seth bonder interview (mors)

Calhoun: The NPS Institutional Archive

Faculty and Researcher Publications Military Operations Research Society (MORS) Oral History Interviews

2012-17-03

Seth Bonder Interview (MORS)

Bonder, Seth

http://hdl.handle.net/10945/49275

INTRODUCTION

Oral Histories represent the recol-lections and opinions of the personinterviewed, and not the official posi-

tion of MORS. Omissions and errors in factare corrected when possible, but every ef-fort is made to present the interviewee’s ownwords.

Dr. Seth Bonder, FS, was MORS Presi-dent from 1969 to 1970. He received theVance R. Wanner Memorial Award in 1986and was inducted as a MORS Fellow in1994. He was a full-time faculty memberin the Department of Industrial and Op-erations Engineering, the University ofMichigan, from 1965 to 1972, and was anadjunct professor in that department from1972 to 2011. He was the founder, formerchair and CEO of Vector Research, Incorpo-rated (VRI) during the period 1972–2001.Dr. Bonder was president of the OperationsResearch Society of America from 1978 to1979 and inducted as an Institute for Oper-ations Research and the Management Sci-ences (INFORMS) Fellow in 2001 (CharterMember). He was vice president of the In-ternational Federation of Operational Re-search Societies (IFORS) from 1985 to 1988.He received the Jacinto Steinhardt Memo-rial Award from the INFORMS MilitaryApplications Society (MAS) in 1999. Hewas a member of the National Academyof Engineering. This interview was con-ducted on two separate occasions in theMORS office in Alexandria, Virginia: June 23,2004 and November 28, 2005. Dr. Bonderdied in Ann Arbor, Michigan on October29, 2011.

MORS ORAL HISTORYInterview with Dr. Seth Bonder, FSMr. Mike Garrambone and Dr. Bob Sheldon,FS, Interviewers

Bob Sheldon: First of all, tell us whereyou were born.

Seth Bonder: I was born on 14 July 1932in South Bronx of New York City. I grewup there and lived in the same apartmentthrough the depression years and until I leftin 1951.

Bob Sheldon: Give us your parents’names and tell us how they might haveinfluenced your decisions academically orprofessionally.

Seth Bonder: My parents were Al andMinnie Bonder. No real influence. Theywere Russian immigrants. They came overseparately and met here in the UnitedStates. They never had a day of schoolingso it hardly had any influence or effect onmy education. In fact, my education didn’tstart until much later in life, as you will see.

Mike Garrambone: Did you speak Rus-sian at home? What were your parents’professions or businesses?

Seth Bonder: Although some of my rela-tives spoke Russian in their household, wespoke English. I knew a few Russian phrasesfrom spending time with my relatives.

My parents weren’t professionals—remember I said they did not have anyschooling. Both worked in the garmentdistrict of Manhattan. My mother was an‘‘operator’’–she sewed dresses together.My father was a dress presser who workedwith 25-30 other pressers in a large room.He would use a heavy cast-iron steam ironto steam dresses before they were deliv-ered to department stores in Manhattan.It was a real sweatshop and a tough job!He earned every dime he made.

Returning to Bob’s question, I had littleinterest in education as I grew up in theBronx. I spent very little time with books.At 15, I was a pretty good three-cushion bil-liards and pool hustler, played a lot of stick-ball, and also played a little basketball inhigh school. I was a poor student and notinterested in academic activities.

Mike Garrambone: How did you getgood at three-cushion billiards?

Seth Bonder: When I was a youngster,I spent most of my time on the street likemany of my friends, playing stickball, bas-ketball, and getting into trouble. We livedin a one bedroom apartment which waswhere I went to eat and sleep. When I wasabout 12, I started hanging around theneighborhood pool hall (which was illegalin those days). This was not some upper so-ciety pool hall or even what you think ofas a pool hall today. It was a rough crowd,many interested in gambling, corruption,and even some drug peddling. Some ofthe older poolroom crowd took me undertheir wing and taught me how to playpocket pool, but I quickly gravitated tothree-cushion billiards. I thought it wasa fascinating game. Much more skill wasinvolved. After a couple of years with a lotof time on the table and watching good

MilitaryOperationsResearchSociety(MORS)Oral HistoryProjectInterviewof Dr. SethBonder, FS

Mr. Mike Garrambone

InfoSciTex CorporationMichael.Garrambone.ctr@

wpafb.af.mil

Dr. Bob Sheldon, FS

Group W, [email protected]

MILITARY OPSRESEARCH HERITAGEARTICLE

Military Operations Research, V17 N3 2012, doi 10.5711/1082598317305 Page 5Military Operations Research, V17 N3 2012, doi 10.5711/1082598317305 Page 5

players, I became quite good at understandingthe geometries, controlling ball position, andthe strategy of the game. At 15 I started playing insome three-cushion and pocket pool big moneygames, money provided by big betters and Iwould get a small piece of the winnings. WhenI was 16 I had a part-time job running and closingdown the pool hall a few nights of the week.

Bob Sheldon: You said you did poorly in highschool. Even in math?

Seth Bonder: Everything. I was not interestedin school. When I was 15, I became a memberof one of the local gangs, which served as amutual protection society. Gang members didn’tthink much about schooling; we just survivedin very strange ways. So there was really noth-ing driving me to do anything academic.

I actually started at City College of NewYork (CCNY) because you didn’t need gradesto get in, it was free, I could avoid the draft,and I thought I might get to play some basket-ball there. But then they had a big scandal whenthe varsity team threw basketball games. Inthose days, teams could play in both the Na-tional Collegiate Athletic Association (NCAA)and the National Invitation Tournament (NIT)basketball tournaments, and CCNY won bothof them. The next year they started throwingbasketball games. They got caught by the DistrictAttorney, and the whole program fell apart. Notlong after that, I broke a leg, drove a truck fora while, and then the Army came after me.

Bob Sheldon: You took one year of college?Seth Bonder: Well, it wasn’t even that. I

attended for a little more than a semester andhardly ever went to class. I was in the pool roomor on the basketball court. That’s all I did. I wasthere to play, but then I had to leave becauseI never attended class. I went to the Air Forcebefore the Army got me. Not unlike you did,Bob. I went in as an enlisted man but got com-missioned later.

Mike Garrambone: When did you start basictraining and where did you go for it?

Seth Bonder: It was late 1951 at Sampson AirForce Base (AFB) in upstate New York. It wasquite cold. I don’t believe it exists anymore.The Korean War was on. During that time pe-riod they allowed enlisted men who did nothave a college degree to go to flight schoolthrough the Aviation Cadet program.

Bob Sheldon: So you didn’t have a degree,but you were commissioned?

Seth Bonder: Right. I was commissionedthrough the Aviation Cadet program.

Bob Sheldon: How long were you in the AirForce?

Seth Bonder: About five years. It was duringthis period that the seeds of my future life wereplanted. Everyone I worked with and social-ized with had college educations—many withadvanced degrees. It became clear to me thata college education was necessary to have ameaningful career in the Air Force. A relatedanecdote sent a message loud and clear. One ofmy assignments was at Edwards AFB, California,the Flight Test Center. I was at the bar onenight with the personnel officer. He informedme that a quarter of one percent of the officersat Edwards AFB did not have a college degree.And that was me!

He said, ‘‘If you stay in the Air Force youhave to get a degree.’’ I said, ‘‘How do I do that?’’He said, ‘‘You need to get two years of collegecompleted through extension courses beforethe Air Force will send you full time to get a de-gree.’’ I thought this would take forever andwasn’t worthwhile. I decided I’d get out andtry college on my own.

Well, I got out of the Air Force at the endof 1956 and couldn’t get into college. Nobodywould take me. I called Senator Dollinger fromNew York. I said, ‘‘Look, I’m a vet. I don’t havevery good grades but I’m pretty sure I could dowell if they’d let me into college.’’ He con-vinced the University of Maryland to let mein. I picked Maryland because I did not wantto return to New York.

That’s how my education started. My par-ents had little influence on my decision to geta college education. Rather it was my Air Forceexperience and friends, most of whom had oneor more college degrees. I went into engineer-ing because of my Air Force experience and be-cause I thought I could probably dig a betterditch if we had a depression again. I startedin February 1957 and finished in June 1960 witha degree—major in mechanical engineering anda minor in physics.

Actually my first semester in Maryland wasa disaster. I don’t think I got higher than a 40in my first round of midterms. I wanted to be

MORS ORAL HISTORY PROJECT . . . DR. SETH BONDER, FS

Page 6 Military Operations Research, V17 N3 2012

back in the Air Force. But I went to counsel-ing sessions and learned how to study. Thatyear the academic light was turned on. It wasan amazing transition. I realized I actuallyhad a brain and could use it. I finished in aboutthree and a half years even while driving ataxicab 4-5 nights a week. (After one of myMORS keynote addresses in the 1970s, an at-tendee came over and introduced himself asmy Maryland study counselor. He said hewas pleased that his guidance had helped. Sowas I!)

After graduation I went to work at West-inghouse in Baltimore for a couple of months.I hated it. Sitting at a drafting board was notstimulating, nor what I wanted to do.

Bob Sheldon: Did you work as a mechanicalengineer?

Seth Bonder: Yes. And fortunately a friend ofmine who was stationed with me at EdwardsAFB was getting a PhD at Johns Hopkins in anew field called operations research (OR). Hewas doing this in 1958–1960. He actually ap-plied to graduate school for me at two places:Johns Hopkins, where he was going, and OhioState University (OSU). When I was gettingready to quit Westinghouse, he informed mehe did this and he felt it would be an interestingfield for me because I liked people-oriented oper-ational problems.

Bob Sheldon: Did you test for scholarshipsor were you just offered them?

Seth Bonder: No, they were offered becauseI did very well at Maryland. I was in all thehonor and leadership societies. I was older thaneverybody so that was to my advantage. I didsome interesting things at Maryland. A Navypilot and I helped rejuvenate a flying club inCollege Park, and because of my disastrous firstsemester, I started a freshman tutoring servicefor the University.

Anyway, I went over to Johns Hopkinsand talked to Professor Charlie Flagle. I don’tknow if that name’s familiar to you. In lateryears he and I did a lot of work together in theOperations Research Society of America (ORSA).As an aside, very few analysts in MORS everget involved in ORSA. It’s like two differentworlds. It’s a big mistake for the military ORcommunity—I believe it would be useful to theprofession if more military OR analysts were

involved with ORSA (now INFORMS) throughMAS (the Military Applications Society).

Charlie Flagle was one of the pioneers ofacademic OR. He authored the Flagle, Hugginsand Roy book Operations Research and SystemsEngineering. That was one of the primary ORbooks that came out in the late 1950s. CharlieFlagle was a big name in the field. I went tosee him and he told me the scholarship wasfor $1200 a year. And I said, ‘‘That’s pretty tightto live on.’’

When I went to school at Maryland, I hadthe GI bill and I drove a cab at night in PrinceGeorge’s County and the Washington, D.C. areato make a living. I’d get the cab from a privateowner. He’d switch cars with me about 7 pm,I’d drive until about one in the morning, andthen he would switch cars back again about6 am. I would drive about three nights duringthe week and 1–2 days during the weekend.I actually did my first simple operations analy-sis as a cab driver: I collected and analyzed thecall demand data in Prince George’s Countyto learn how to be ‘‘nearby’’ when calls werebroadcast for drivers to bid on.

Bob Sheldon: Do you know the roads aroundhere pretty well?

Seth Bonder: I do. Well, not here in Virginia,but I did know Prince George’s County andD.C. pretty well. Anyway, Charlie said, ‘‘Youhave to pay tuition out of that too.’’ And I said,‘‘How much is tuition?’’ He said, ‘‘$1200.’’I said, ‘‘I must tell you, Professor Flagle, I can’tdo that and will have to go to another school.’’

I called Dan Howland, the professor atOSU who extended the fellowship, and said,‘‘You offered me a fellowship. What is it?’’ Hesaid, ‘‘You get $300 a month. You don’t haveto work for the first year. Tuition is covered.We guarantee you a job the second year as aresearch assistant.’’ And I said, ‘‘I’m coming.’’So it turned out I was the first Systems Fellowat OSU and that’s how I got into the OR field.

When I arrived at OSU, I learned that DanHowland was also the head of the Systems Re-search Group (SRG). Dan is now long gone.He had a bunch of research projects in OR,in the military area and in the health area.He said, ‘‘You can pick an area to work in afterthe first year. Since you have a military back-ground, why don’t you work on a military


Military Operations Research, V17 N3 2012 Page 7Military Operations Research, V17 N3 2012 Page 7

project?’’ That’s how I got into the military ORbusiness.

At the end of my first year, which was in1960, I worked on a little project for the militaryat Fort Knox, Kentucky. Dan had a number ofcontracts with Fort Knox. Nobody in the re-search group knew much about the Army, butwe had contracts. This one was to determinethe optimal path for a tank to drive through acontaminated nuclear battlefield. It was beingdone for the group at Fort Knox that was calledthe Combat Developments Group. I spent timeat Knox learning about the Army. I got intriguedby what one of the groups was doing. They wereresponsible for writing requirements for new ar-mored systems.

Bob Sheldon: Did you drive to Fort Knoxfrom OSU?

Seth Bonder: We’d fly down or drive down toobserve their work activities. That was in 1961.I observed that it didn’t make any sense theway they were drafting requirements for newsystems. They would sit on large butcher paperwith their slide rules drafting requirements forthe next generation tank. They would calculatevalues for many performance characteristicssuch as how fast it should go, how far it shouldbe able to go, how reliable it should be, howwell it should detect targets, the accuracy offiring the tank’s main gun, and the vulnerabil-ity of the tank.

I was amazed how primitive the processwas to develop requirements for the next gen-eration armored systems. It dawned on methat there were two major problems that wouldbe interesting to work on. One was the feasibilityof the requirements. If you wrote numbers downfor those variables or parameters, could you,in fact, build a system with those capabilities?I doubted it because the process they were us-ing didn’t take into account the physical inter-actions among the various parameters.

For example, if they wanted to increase howfast and how far they wanted the tank to go,they would specify the speed of the tank andits cruising range. And if they wanted to, forexample, increase the survivability of the tankthey’d add more armor to it. They neglected torecognize that if you added more armor and, ifyou wanted to go as fast as you said you wantedto go, you needed a bigger engine. If you added

a bigger engine, you needed more fuel to main-tain the tank’s cruising range. This would de-crease the survivability because the additionalfuel made the tank more vulnerable to enemyfire. They had no idea how to handle thoseengineering-level interactions. Based on somerecent experience, I believe this problem maystill exist in a number of military domain areastoday.

The second problem was the issue of whatwould be useful to write as performance specs?Hopefully the specs they wrote would producea tank that was operationally effective. That is,the next generation system would prove tohave a high operational utility on the battle-field. So you had the feasibility issue and theeffectiveness issue. I discussed this with DanHowland who became my PhD advisor. Hesaid, ‘‘Why don’t you write a proposal to ad-dress these?’’ I’d never written one but thoughtit would be a learning experience.

I wrote it for Fort Knox which was then partof the Continental Army Command (CONARC).I’m not sure they understood it. In fact, thefirst time I presented it to the Colonel incharge of the Combat Developments Groupat Fort Knox, he essentially threw me out ofhis office. But his staff convinced him it wasworthwhile doing and they moved it forwardto General Daly who was then the Command-ing General of CONARC. He was a four-starwho looked like General Curtis LeMay, Com-mander of the Air Force’s Strategic Air Com-mand, smoked cigars like him, and acted likehim.

I was asked to give a briefing at FortMonroe, Virginia, the home of CONARC Head-quarters. Fortunately we had a number of Armyofficers who were in the master’s program withme at OSU. One of them was Rick Anson, a cap-tain who went on to become a two-star general.Others were Major Bill Bercaw, LTC Larry Caid,and Captain Dan Schroeder.

Bob Sheldon: These were active duty offi-cers?

Seth Bonder: Yes. They were active duty offi-cers who went to OSU because there weren’tmany OR educational programs in the country.You had Case Western, OSU and Johns Hopkins.These were the three main OR programs thatexisted in the early 1960s.



They helped me write the proposal whichenhanced it by including appropriate Armyjargon. Rick Ansen, who used to be a trainer ofArmy briefers said, ‘‘Let me show you how todo a briefing.’’ This was my first exposure onhow to market in the Department of Defense(DoD). He said we needed to have slides. Inthose days you didn’t have computers to creategraphics. It was very labor intensive at theSystems Group. They’d make big cardboardpictures, photograph them, and then put thenegatives in 3½ x 4 inch glass slides (calledlantern slides).

We went down to Fort Monroe. I gave thebriefing and the argument was over which orga-nization would house it, not whether I’d get it.General Davis said, ‘‘Look, let’s not air our dirtylaundry in front of these guys from the Univer-sity.’’ To make a long story short, I received a $2million contract as a graduate student to runthe proposed four-year research program.

I was the project director and a full-timegraduate student. We started building modelsto address the feasibility issue. That is, predicthow fast the vehicle could go, how well the crewcould detect, how accurately it could fire themain gun, how vulnerable the tank was, etc.My research staff consisted mainly of graduatestudents, supplemented by a few faculty mem-bers from various disciplines and departments.That research slowly brought me into the main-stream of what was happening at the tacticallevel of Army OR at that time.

Although OR was conducted for the Armyin WWII, I believe it first got codified in thelate 1940s with the establishment of the Opera-tions Research Office (ORO). ORO was formedand run by Johns Hopkins University as theArmy’s first FFRDC (Federally Funded Researchand Development Center) to do strategic-leveland policy-level analysis. This later became Re-search Analysis Corporation (RAC) in the early1960s. RAC was sold to General Research Cor-poration (GRC) in Sep 1972 after the Army in-formed RAC it would no longer be supportedas an FFRDC. The Army Strategy and TacticsAnalysis Group (STAG) was established as afield activity under the staff supervision of theDCSOPS in August 1960 and was subsumedby the Army Concepts Analysis Agency (CAA)when it stood up in 1973.

The tactical level of Army OR and experi-mentation in the 1950s and 1960s was conductedand supported by the Ballistics Research Labo-ratory (BRL), run by guys like Joe Sperrazza,Keith Myers, and Dave Hardison, and the Com-bat Operations Research Group (CORG) whichwas a field office of ORO. Some of my futureprofessional friends such as E.B. Vandiver, HunterWoodall and John Riente worked there.

I got into the business at the tactical levelof Army OR at OSU under the contract thatI had with Fort Knox. CORG was doing a bunchof experiments. That was a period in Army ORthat was a scientific era: the 1950s, 1960s, and1970s when the Army did a lot of experimen-tation. They experimented with tank accuracy,firing times, vulnerability, pinpoint detectionand other performance characteristics.

I did some limited experimentation at OSUto assist in developing engineering-level modelsof performance capabilities for armored weaponsystems. We worked on vehicle speed and theimpact of vibrations on cross-country speed. Wedid theoretical work in the engineering me-chanics lab, and ran mobility experiments witha local reserve unit. We did experiments on vi-sual detection to see how well you could detectstationary and moving tanks. I almost lost acouple of my researchers (who were students)because when a tank was coming at you, youdidn’t see any contrast changes. Some of themalmost got run over. It was an exciting time. Wewere supplementing some of the Army’s ex-perimentation activities and building modelsto estimate performance capabilities of futurearmored systems.

Mike Garrambone: Were these mathematicalmodels?

Seth Bonder: Mostly. They were engineeringmodels that described the physics of variousphysical and operational processes.

To assist with our performance-level mod-eling, Bill Bercaw, an Army major who was a stu-dent at OSU, arranged for us to go to Fort Knoxand shoot on the firing range. Most of my re-searchers were students, and some were women.We’d go out to the range and fire the tank maingun and the 7.62 mm machine gun. I had oneresearcher who was an unbelievable gunner.They would pull pallets as targets at differentranges and the idea was to fire a round through



the first pallet and then, after the gun wasreloaded, fire a round through the second pal-let if you could. My researcher did that whenthe pallets were being pulled at 1800 meters!Our project officer was quite proud of the re-search team.

As I said, Army OR was heavily intoexperimentation which was necessary anduseful. We learned about the feasibility ofachieving performance levels, and to modelperformance capabilities such as hit probabil-ities, firing rates, firing times, and loadingtimes. At Fort Knox, I used to have someoneclocking the loading times and the firing timesand used this data to develop models as afunction of the characteristics of the system.BRL was heavily involved in estimating thelethality and vulnerability performance char-acteristics for armored systems.

The Army Materiel Systems Analysis Ac-tivity (AMSAA) was a spin-off of BRL. BRLwas much more a hard engineering group andAMSAA looked at issues of system perfor-mance and effectiveness, just like we were do-ing in a smaller way at OSU and CORG wasdoing. They were also heavily involved in Armyexperimentation.

Experiments were done on many of thearmored system performance capabilities. Someof the early ones were the STALK experiments(Project STALK tank-versus-tank experimentsin 1953) which Dave Hardison analyzed. Itlooked at the impact of different threats, dif-ferent fire control systems, different size gunsand a few other dimensions on hit probabili-ties, kill probabilities, detection probabilities,firing rates, and others.

During the 1950s, 1960s, and 1970s theArmy OR community was involved in exten-sive experimentation to gather data about theseperformance parameters and to understand fun-damental land combat processes. We learnedwhat was important. We learned how to startbuilding models to predict the system’s perfor-mance parameters (capabilities) as a functionof its engineering characteristics. (For exam-ple, the model for hit probability was a bivari-ate normal distribution with a number of fixedand variable biases that you had to estimate topredict the probability of hitting a specific tar-get with a particular type tank gun. Models of

this performance capability, as well as esti-mates of the various biases, were being devel-oped using field and laboratory experimentaldata.) We learned how to improve these perfor-mance capabilities, how to represent them incombat models, and how to model land combatprocesses.

Another experiment was Project Pinpoint,which John Young ran, to determine how wellyou could detect an enemy when they fired around and the associated location errors. Ourresearch at OSU involved theoretical and ex-perimental activities to develop models ofdetection capability based on differences incontrast stimuli.

It was a wonderful community of aca-demics like me, industry researchers, and theArmy that participated in these experiments.That’s when I first met Walt Hollis. In my earlydays at OSU we were trying to build modelsof systems accuracy and we needed estimatesof the biases and how to predict the biases. WaltHollis worked at Frankford Arsenal where hewas a recognized expert in the area. I met Waltthere and learned all about firing main tankguns.

As I said, it was an exciting period of time.I was not only developing models to estimatesome of the performance capabilities but de-veloping software to integrate my models andthe Army’s performance-level models so wecould estimate the feasibility of building a tankweapon system with these capabilities, consid-ering all the interactions among them. Whilethis feasibility-related research was going on,I also started thinking about the potential effec-tiveness of the systems in an operational setting.

Bob Sheldon: When you were buildingthe models, did you use your academic knowl-edge or were you using knowledge based onexperience?

Seth Bonder: I think both—academic capabil-ity and experience from some of the modelingand experimentation work being conducted bythe Army. Obviously, the mathematics was im-portant. My initial foray into building effective-ness models back in 1962 or 1963 (which led tothe Bonder Individual Unit Action (IUA) andmany subsequent developments downstream)was stimulated by a two-star general from theCombat Developments Command. It was Major



General Pickett who was responsible for build-ing the Sheridan system.

He was actually a descendant of GeorgePickett from Pickett’s Charge in the Civil War.He knew I had the contract with Fort Knox.He came to OSU and said, ‘‘I need to knowabout the effectiveness of alternative potentialSheridan light tanks described by alternativesets of performance characteristics. Can youhelp me?’’

There weren’t any combat models aroundthat I could responsively use to address hisproblem. This was a quick reaction six-weekstudy – and I was surprised the University letme do it. We kept it unclassified.

In 1962 there were a small number of com-bat models being used in Army studies butwhich I deemed inappropriate for the Sheri-dan study. There were large gaming modelswhich were inappropriate both because theywere more operational-level models, not tactical-level ones, and due to the requirement for manyplayers to run them. Then there was ATLAS(ATactical, Logistical and Air Simulation), an ag-gregated ‘‘firepower score (FPS)’’ based modelof theater-level combat which was inappropri-ate for many reasons. It was a ‘‘holistic’’ modelthat did not represent details of military sys-tems or processes that I needed to consider. Itrolled all of these up in a many-to-one transfor-mation to produce the FPS for each side. In turnthe downstream loss of each side’s FPS andeach side’s movement was determined via somevery questionable look-up tables. I thought theFPS construct was somewhat ‘‘garbage’’—ithad no scientific basis and in later years I referredto it as the ‘‘Phlogiston’’ theory of combat. (Thephlogiston theory is an obsolete scientific the-ory of combustion from the 1700s—phlogistontheory claimed that combustion gave off oxy-gen, rather than using it.)

Then we had the Monte Carlo simulation:Carmonette. I believe it was developed byDick Zimmerman at CORG back in the 1950s.This was an Army company-level model forblue defense against a red attack or the otherway around. Over the years it was expandedto battalion-level. It was a ‘‘synthetic’’ modelas opposed to holistic. It considered many ofthe relevant tactical-level military processes anddetails of the system performance capabilities

like the vehicle speed, hit probabilities, kill prob-abilities, detection probabilities, etc.

Bob Sheldon: Were you building digital sim-ulation models?

Seth Bonder: I wasn’t, but Carmonette wasa digital simulation model. Carmonette was theonly tactical-level model around; but it took,even at a company level, 30–60 minutes of com-puter time to replicate a single company-levelcombat engagement and 15–30 replications toachieve some statistical stability. Thus we werelooking at 10–15 hours to simulate a singleengagement because computers were slow inthose days. I needed a model that would be ableto run parametric analysis on the Sheridan’sperformance capabilities.

A class of analytic models which were moretheoretical in nature, and primarily of interestto the academic community, was the Lanchesterdifferential equations models with coefficients(‘‘attrition rates’’) no one knew how to estimateexcept after an engagement. There were a lot oftheoretical papers written about various formsof the equations. A square law, a linear law,a logarithmic law—but they weren’t useful inreal studies. Although the differential equa-tions proposed by Lanchester had many prob-lems, their simplicity was appealing.

To do the Sheridan study, I decided I wouldtry to compute attrition in a small unit combatengagement using the differential equationsby developing ways to estimate the attritionrate, a priori. I built simplified models that con-sidered hit probabilities, firing times, and killprobabilities to estimate attrition rates. So forthe Sheridan study, I started using what wethink of as Lanchester models. However, basedon doing this simple study it was clear thata fundamental assumption in the classic Lan-chester models that the attrition rates are con-stant is unrealistic if combatants move duringan engagement. For example, consider some ofthe parameters I used in estimating the ratesfor the study.

The hit probabilities increased the closeryou got to a target. You could detect better asyou got closer to potential targets. Clearly thoseparameters and others were changing over timewhich suggested the attrition rate had to changeover time during an engagement. Recogniz-ing this, I performed the study by building a



simplified scenario involving opposing forces,and solved the variable coefficient differentialequations numerically for the alternative Sheri-dan candidates.

After the study was completed, I comparedthe results with those achieved by using theclassical Lanchester constant-coefficient for-mulation, using various ‘‘average’’ coefficientscalculated with the parameter data. The differ-ences were quite significant. I had stumbled onan interesting research area for my dissertationresearch at OSU (from 1963–1965), and for laterresearch areas at the University of Michigan(1965–1972) and VRI beyond 1972.

As I delved into my dissertation, it was clearthat, not only was the attrition rate a variable asforces moved about the battlefield, but becausemany of the processes I considered in comput-ing the attrition rates in the Sheridan study werestochastic in nature (firing accuracy, firing times,target detection, etc.), then the attrition rate wasalso stochastic. If you think about the conceptof a stochastic attrition rate that varied overtime, the differential equations became variablecoefficient (sometime nonlinear) stochastic dif-ferential equations. Recognizing the extreme dif-ficulty in trying to solve these formulations infinite time, I made some tactical decisions toproceed with the dissertation: I would developmodels to estimate the probability distributionof attrition rates as a function of the fundamen-tal performance characteristics of a weaponsystem. I would then suppress the stochasticaspects by using some statistic from the distribu-tion as the variable attrition rates in the variablecoefficient differential equation formulations.

In the 1964 time period, I assumed that theexpected value of the attrition rate random vari-able was an appropriate statistic to use in thedifferential equations. In 1967, using some ar-guments from renewal theory and Blackwell’stheorem, my colleagues at Michigan demon-strated that the time-to-kill a target, not therate, was the relevant random variable, andthat a theoretically sound definition of the attri-tion rate is the reciprocal of the expected timeto kill a target. Bernie Barfoot (Charles BernardBarfoot) from the Center for Naval Analyses(CNA) made a similar observation in 1969using some harmonic mean and Markov chainarguments.

During the 1963–1965 period, I did someinteresting theoretical work on the variable co-efficient differential equations, particularly look-ing at the impact of movement of forces duringan engagement. The military has known foryears that mass is a nonlinear force multiplierin combat. Some of my research suggested thatattacking with sufficient speed also is a nonlin-ear force multiplier and, along with mass, canbe used to rapidly saturate a defender’s retalia-tory capability.

As an interesting but scary aside, sometimein 1964 I read an abstract in the Defense TechnicalInformation Center (DTIC) by George Gamow,the famous physicist, summarizing work hewas doing on variable coefficient differentialequations for the DoD. I saw my dissertationfly out the window! Fortunately, when I receivedand read the report, it described some militaryactivities that could be modeled by the equationsbut ended up concluding that they were verydifficult to solve.

As part of the dissertation, I developedmodels of the time-to-kill probability distribu-tions for armored weapon systems and thustheir associated attrition rates. This involvedsome fundamental process modeling of the sys-tem’s doctrine for attacking targets and includedits detection, firing, readjustment of fires, lethal-ity and other performance capabilities againstvarious type targets.

During the same period (1963–1965) I con-tinued to run the Fort Knox project, referred toas The Tank Weapon System project in whichwe were developing and integrating models topredict the performance characteristics of pro-posed armored systems, including the inter-actions among them. These were intended toassess the feasibility of achieving the capabil-ities and to feed these parameters into thedifferential models of combat to assess theeffectiveness of proposed alternative armoredsystems.

Bob Sheldon: Who helped you come up withthese variables?

Seth Bonder: A combination of sources. Byreviewing available experimental data; by work-ing with armored officers at Fort Knox whowere writing requirements for future armoredsystems; by talking to other Army analysts likeHunter Woodall and Dave Hardison; and by



reading historical accounts of military engage-ments. Many of the parameters highlightedearlier were included in the attrition rates, in-cluding the lethality of munitions which wehad to estimate through some modeling.

Obviously, the lethality of fire was an im-portant dimension in a tank’s capability to de-feat a target. This was one of BRL’s areas ofexpertise. We needed to be able to predict theprobability of a kill given a hit, and other relatedlethality parameters. BRL was working on modelsof these parameters. They had developed sophis-ticated computer programs that fired rounds intoa tank at four inch squares, assessed damage tocomponents behind it, and computed the rele-vant kill probabilities.

I asked them if we could transfer the pro-gram to OSU. They said to me, ‘‘You couldn’tprogram this in OSU’s computers in less thana year.’’ I said, ‘‘Why don’t you give it to meanyway. We’ll try it.’’ Remember, some of myresearchers were students and some were fac-ulty. I went back home and said to one of mystudent researchers named Dick Freedman,who was an absolutely fantastic computer pro-grammer, ‘‘Dick, I just came from BRL. Theygave me their vulnerability program and saidit would take at least four months for us to con-vert it.’’ He took this as a challenge and had itrunning in eight weeks. And he enhanced itby improving the computational algorithms forthe normal distribution. The students I hiredwere very smart.

I completed my dissertation and graduatedfrom OSU in 1965. The five years I spent atOSU were the beginning of my professionallife in military OR. I had become acquaintedwith many Army OR analysts and familiarwith many of the Army’s tactical-level plan-ning problems. It was the start of the researchthat eventually led to the Bonder-Farrell dif-ferential equations of combat and eventuallythe VECTOR series of campaign models.

When I left in 1965, the Fort Knox projectwasn’t completed. I turned that over to GordonClark who had just returned from the MarineCorps to get his PhD. He took a totally differ-ent approach for more analytic effectivenessmodels. He built the Dynamic Tactics Simula-tion (DYNTACS), a Monte Carlo simulationwhose outputs fed the Combat Analysis model

(COMAN) which was one of the first fitted-parameter analytic combat models. It was avery interesting concept. It was a competitiveapproach to mine. Where mine was purelyanalytic, with independently calculated inputparameters, he fit parameters from his MonteCarlo simulation of combat for use in theCOMAN model. Two different approaches toanalytical combat modeling.

When I graduated, I was going to go to in-dustry to make some money. I was so brokeas a PhD student making $3,000 a year that inmy last year as a graduate student I decided toask for a substantial raise. My advisor said,‘‘What do you want?’’ I said, ‘‘I want $10,000a year.’’ I was married in 1962 and I had my firstchild in 1964. I wanted a raise. The Dean said,‘‘That’s more than some faculty make. I can’tgive you that much.’’ I said, ‘‘Then I’m goingto quit. I’ll just write my dissertation. I’m notgoing to run the project anymore.’’ He said,‘‘You’re holding a gun to my head.’’ They gaveme the $10,000.

In late 1964, Jack Borsting approached meabout joining the faculty at the Naval Post-graduate School (NPS) in Monterey. Jack wasthen growing its OR department and was hir-ing a number of new faculty, including StevePollock who later joined me at Michigan. I toldJack that I was looking for a job in industry.

Bob Sheldon: How did you decide to go tothe University of Michigan?

Seth Bonder: In early 1965 I went to Chryslerin Detroit to interview for a job. They were in thedefense business as builders of tanks. Theyknew what I was doing at OSU, and wantedme to come up there and start a military ORgroup. Prior to going for the interview, one ofthe members of my PhD committee, a mathe-matician named Henry Colson, had a goodfriend at the University of Michigan who headedthe Industrial Engineering Department (since1971 called the Department of Industrial andOperations Engineering). Henry said, ‘‘You’vegot to hire this guy. They throw money at him.’’Universities love professors who bring in money.He called me and he said, ‘‘Would you inter-view for a faculty position?’’ I explained I wasnot looking for a university job. He volunteeredto pick me up at Chrysler so I could give a semi-nar in the afternoon.



So I went to Michigan in the afternoon aftermy Chrysler interview. It was intellectuallystimulating. I presented my dissertation at theseminar and got arguments from some verystrong mathematicians. It was just a fabulousactivity and I thought, ‘‘I can get smarter ifI come here for a few years.’’ On the way backto the airport, the Department Chair offeredme a job as an assistant professor and I wouldbecome an associate in a year or two. That wasa pretty good job offer and it was for all of$16,000 for a nine-month year. I accepted thejob offer on the way back to the airport. WhenI got home, I told my wife and she said, ‘‘Ithought we were going to industry.’’ I said,‘‘I thought we were too, but now we’re goingto the University.’’ My wife had been a cheer-leader at OSU so going to Michigan was like go-ing into enemy territory.

I joined the faculty at the University ofMichigan in 1965. I’d never taught before. Thefirst year all I did was write course materials.The first semester I was asked to start a newcourse to teach linear programming for juniorsand to teach a senior level inventory controlcourse. Through the period 1965–1972, I taughta spectrum of OR courses, developed a three-course sequence in decision theory, and starteda course to teach students how to model oper-ational phenomena—later referred to as the‘‘Modeling Studio.’’

Mike Garrambone: Did you use Saul Gass’book for the linear programming course?

Seth Bonder: No. With few exceptions Iusually taught courses using my own notesand, if they existed, made textbooks availableas references for the students. In those daysthere clearly weren’t any texts to teach model-ing. I can’t be sure, but I believe my course inmodeling may have been the first of its kindsince I was attempting to teach students theart of modeling operational processes (leadingto the name ‘‘modeling studio’’). There werecourses with the word �modeling� in the title,but they were teaching linear programmingmodels, inventory models, etc., not the pro-cess of modeling operational phenomena. Do-ing so is not easy. In fact, my first couple ofattempts were failures, but I think I got it rightthe third time. The students learned how tostart with very simplifying assumptions for

their initial version of a model and how toremove them through a series of enhancementsuntil they had a version that could be usedto address the decision issue I posed to startthe process. They worked on a number ofthese in a semester, with individual critiquesfrom me.

It was a demanding, time-consuming activ-ity for one course. It also was a little frustratingin that I couldn’t describe what I ‘‘covered’’ inthe course. In a linear programming courseyou cover the simplex algorithm, duality, etc;in a queueing course you cover M/M/1 queues,etc; in a decision theory course you cover deci-sions under risk, utility theory, Bayesian updat-ing, strategy development, etc. I finally figuredout that I wasn’t covering material but ratherwas providing the students mentored experi-ence in the art of modeling operational phenom-ena. I was pleased to learn a few years ago thatthe number of such courses has grown in engi-neering and business OR programs around thecountry.

At the end of the first year, the DepartmentChairman said, ‘‘I’ve got money. If you want todo some research over the summer, I’ll pay forit.’’ I formed a little research group which Ieventually turned into the Systems ResearchLaboratory (SRL) at Michigan. I hired somegrad students, one of whom happened to beBob Farrell. Bob was then a 21-year-old typistin the Department Chair’s office. One of my col-leagues said to me, ‘‘You ought to hire this guy.He’s really bright.’’ I said, ‘‘So why’s he typ-ing?’’ Anyway, he joined me. Over the summerI and 10 graduate students developed a long-term research strategy for Michigan’s newHighway Safety Research Institute.

Just a little aside on Bob. This was 1966.He was getting his PhD in theoretical math. Inthe previous year, he claimed to have solvedthe well-known Four Color Problem analyti-cally. And as best I understood, they acceptedit. Then three weeks later, he proved he waswrong. If I were in that department, I wouldhave given him the degree for that proof. Well,they didn’t give him a degree. Then when hejoined me as part of the research faculty in1966, he started taking the comprehensiveexams in the Department of Statistics withoutever taking any of the coursework. On his own



he had learned graduate-level probability andstatistics. The next year he wrote a dissertation.

His two advisors were Bill Ericson and, Ithink, Floyd Hill. They both had been studentsof Savage, a pioneer in Bayesian statistics re-search. They said, ‘‘It’s a great dissertationexcept it’s not for Bob Farrell. He’s too smart.He’s got to do more.’’ So he quit and never re-ceived a PhD. But he was, without a doubt, thebrightest man I’ve ever known—immeasurableIQ, unbelievable individual and wonderful towork with. He was a close collaborator, partner,antagonist, and friend for over 30 years.

Anyway, Bob joined my SRL Lab in 1966.The first project was a contract out of Redstoneto do some missile air defense simulationmodeling and another out of Rock Island Ar-senal to do some air defense gun modeling.Then I obtained a contract with Office of Na-val Research (ONR) to continue the researchI started at OSU on analytic models of combat.

I also received a contract from the ArmyAssistant Vice Chief of Staff’s (AVCS) office inthe Pentagon. My contacts in that office wereDick Trainor and John Honig. John was thenand has been very active in MORS for manyyears. The AVCS was Lieutenant General BillDePuy who, as I will comment on later, playedan important role in my professional life. Manyofficers in the AVCS office became generals laterin their careers, including Max Thurman, DaveMaddox, and others.

The ONR and AVCS contracts with the SRLwere to continue my theoretical work on theanalytic models of combat. It involved a broadspectrum of research on tactical-level combatoperations. It also included building predictivemodels of attrition rates for different weaponsystems. Not just tanks but infantry systems,for helicopters, for artillery, for various otherweapon systems. It included expanding thedifferential models of combat to represent mul-tiple types of systems in combat operations—multiple systems on one side and multiplesystems on the other side. We expanded theconcept of the attrition rate to the attrition co-efficient to include the allocation of fires amongdifferent targets. The latter included one of mystudent’s theoretical research with differentialgame theory for his dissertation to determinethe optimal way to allocate fires across enemy

forces. We expanded the target acquisitionprocess to include line of sight (LOS) effects,visual pinpoint detection and false targets.There were models of other combat processes,all of which fed into the attrition coefficientwhich, in turn, fed into expanded differentialmodels of combat. That work took place from1966 through 1972 in the SRL. Much of this re-search was documented in an extensive reportfor ONR and the AVCS office—referred to asthe ‘‘Bible.’’

Bob Sheldon: This laboratory was at the Uni-versity?

Seth Bonder: Yes. It was part of the Industrialand Operations Engineering Department ofthe University of Michigan. The SRL includeda number of bright PhD students (includingPeter Cherry, George Miller, and some militaryofficers), a cadre of masters students (includingDavid Thompson), and some faculty (includingRalph Disney, a superb stochastic process re-searcher). Bob Farrell was a full-time researchscientist in the lab.

During my time at Michigan, I also didsome consulting for industry. One of my clientswas the Chrysler Defense Group in Detroit.They asked me to work with them four days amonth for a couple of years as they preparedto bid on the next generation tank system. Al-though I had not consulted before, one of mysenior colleagues said my rate should be $300per day. Senior management at Chrysler saidthat was way too high. So I offered them an al-ternative: I would work four days a month untilthe program was awarded (which could havebeen 3–4 years) for free, but if they won thecontract (worth billions) I would get a milliondollar bonus. Needless to say, they paid the rateI originally requested. With the Chrysler work,I was consulting about 5–6 days per month,while teaching 1–2 courses, doing research,managing the research and marketing for theSRL, guiding about 3–4 PhD dissertations andsome masters students, serving on departmen-tal committees, teaching summer short courses,and trying to participate in professional socie-ties. It was a very busy time in my professionallife.

I would be remiss if I didn’t mention someof the wonderful faculty colleagues I had at theUniversity of Michigan. This included Ralph



Disney, Bob Thrall, Herb Galliher, Dick Wilson,Walt Hancock, and others who taught me muchabout teaching and academic politics. In 1969I recruited Steve Pollock from NPS who pickedup my modeling and decision theory courseswhen I left the University in 1972. Of local rel-evance here in Virginia, in the late 1960s I ad-vised a bright assistant professor to move tothe business school since our department wasnot going to grow an IT program—his nameis Alan Merten who is now the President ofGeorge Mason University. A wonderful guy whoI understand has done a great job at Mason.

In addition to the extensive research onland combat, I did some research at Michiganin the area of search theory with some of mymilitary students. Bernie Koopman in WorldWar II developed search theories for the Navywhich I’m sure you’re well aware of. Thatwas powerful work for sea warfare, but itwasn’t appropriate for land combat becauseof LOS issues.

In the sea, except in rare cases, you alwayshave LOS to a target although you may not de-tect it. In land warfare, which is much morecomplicated, you may lose your LOS intermit-tently so you have to start the detection processover. I had an Air Force major work on this forhis dissertation—the impact of LOS on searchtheory and ability to acquire targets. It turnsout the optimal search concepts are quite differ-ent than the original Koopman research.

Following that we added another dimen-sion to the search process: What are the conse-quences if you find a target? There are timeswhen finding it may be detrimental to yourhealth. How does that affect your search pro-cess? That led to another dissertation. Withina span of four years, I probably had six or sevendissertations and many master’s theses writtenbased on the research we did on military opera-tions in the lab.

Bob Sheldon: Were most of these active dutyofficers doing research?

Seth Bonder: No. There were about 25 stu-dents and faculty doing research in the lab. Thisincluded 5–6 Army, Air Force, and Navy offi-cers. When I was at Michigan, the Army offeredto send me 30–40 Army students a year. I said,‘‘There is no way I would take on that manyArmy students a year.’’ This was true for a lot

of reasons. One, it was too big an overload forour department. But more importantly, I thoughtsince OR was such a burgeoning field, it wouldbe appropriate for Army soldiers who get edu-cated in this field to have different perspectiveson it. They should have faculty members otherthan me because I had one view on how to doresearch and analysis that likely would be dif-ferent from other faculty members.

They eventually gave contracts to, I think itwas Tulane, and then to NPS in Monterey wherethey sent most of the officers. I got two or threea year from the services. One Navy lieutenantcommander was writing a classified disserta-tion on fleet air defense. I asked Ervin Kaposwho was then at CNA to serve on the commit-tee. (I’ll tell you shortly how I met Ervin in 1966.)

This Navy lieutenant commander was al-most done with his dissertation, but ran outof time and had to go back to the Navy. SinceI had invested so much time in this research,I arranged with Jack Borsting for him to go toNPS to finish his dissertation. But then he gotan offer from Navy Admiral Zumwalt to cometo a new Navy analysis office called OP-96. Hesaid to me, ‘‘What should I do?’’ I said, ‘‘Well,we spent a lot of time on this dissertation but,if I were you, I’d go to Zumwalt. You’ll becomean admiral.’’ Well, he eventually became a two-star admiral, but never finished the dissertation.

I should backtrack to Ervin Kapos becauseI’m bouncing around a bit.

Bob Sheldon: That’s fine. I interviewed Ervinthis summer.

Seth Bonder: My first interaction with MORSwas through Ervin Kapos. I arrived at the Uni-versity of Michigan in 1965. In 1966, I receiveda call from Ervin who said, ‘‘My name is ErvinKapos. I’m running a symposium for an orga-nization called the Military Operation ResearchSymposium (later changed to Society) andwould like you to referee the submitted pa-pers.’’ Stroking my ego, he said he heard Ihad high standards and he wanted a qualitysymposium. I believe he heard this from a cou-ple of Army officers who worked with me atOSU where I did develop some high standardsfor analysis and research. (Looking back, I be-lieve I embedded these standards into the VRIculture and to other activities in my profes-sional life.) Being a good academic and not



understanding the magnitude of the task, Iagreed to review some papers. Big mistake onboth our parts! He sent all the submitted papersfor me to review (I expected 10–20), and I rejectedall but a handful—clearly not enough for aMORS Symposium. (Knowing Ervin, if hewere the reviewer I’m sure he also would haverejected most of the papers.) But he had a sym-posium to run. At his request, I gave him thetop 20 which he included in the symposium.

That’s how I first met Ervin and learnedof MORS. I attended the 1966 symposium inMonterey and began a long association withthe Society. I joined the Board of Directors in1967 and became President in 1969. I continuedto work at many of the subsequent meet-ings and presented some plenary and keynotespeeches over the years. In one of my early key-notes, I tried to highlight the ‘‘tyranny of num-bers’’ and ‘‘rubber threat’’ which the Office ofSecretary of Defense (OSD) Program Analysisand Evaluation (PA&E) used to control the Ser-vices’ budgets and expenditures. At a later oneI introduced the methodology of ‘‘versatilityplanning’’ as a rational alternative to the ‘‘cost-effectiveness’’ paradigm for defense planning.(I believe many of the versatility concepts arenow embedded in some of the services and JointChiefs of Staff [JCS] planning approaches.) Ina MORS plenary talk a few years ago I presentedsome ‘‘lessons learned’’ for military model build-ing and analysis based on my 401 years of expe-rience in the field. I’ve always cherished my timewith MORS and the MORS Board. Early on Imet some very bright dedicated analysts (suchas John Honig, Murray Greyson, Art Stein,Gene Visco, Joann Langston, John Kettelle,Clayton Thomas, Jack Borsting, Dave Schradyand many others) and established some life-time friendships through MORS.

Bob Sheldon: Are we still in the 1965–1972period?

Seth Bonder: Yes. A number of key events oc-curred in the 1965–1972 period which had sig-nificant impact on my future professional life.I’ve already mentioned the phone call fromErvin Kapos which led to a lifelong associationwith MORS.

During the 1960s, many individuals in in-dustry, the Army, and the DoD who were prac-ticing or wanted to do military operations

analysis were engineers and physicists (evena few economists!). They had good technicalbackgrounds but little formal training or expo-sure to the concepts, mathematics, and phenom-enology of military operations analysis (e.g.,model building, decision analysis, probabilitytheory and stochastic processes, simulation ex-periments, combat processes, effectiveness mea-sures, life cycle cost analyses, etc.). Believingthere was a demand for this type of course, inthe summer of 1967 I started a one-week, 8 hoursper day, short course titled ‘‘Topics in Mili-tary Operations Research’’ at the Universityof Michigan’s continuing education program.Given the large enrollment the first year, I ex-tended the course to two weeks in 1968. Alongwith some of the other lecturers, I developedcourse notes that were 1,000 pages of lectureson background material and the work I hadbeen doing in the 1960s on predicting systemperformance capabilities, on estimating forceeffectiveness, and on how to perform logicallysound analyses. I had approximately 100 stu-dents attend each year. Dave Maddox attendedin the late 1960s when he was a captain or major.In addition to offering the course at Michigan,I taught it at a couple of DoD agencies. If I re-member correctly, one of them was the Army’sVirginia-based Systems Analysis Group in 1969–1970 run by Marion Bryson, and another wasBRL in 1971 or 1972. Although I stopped offeringthe course after 1972 when I left the University,I continue to run into former students who tellme how useful the course was and how valuablethe course notes have been. I met many interest-ing folks who were involved in DoD-related ac-tivities through the course.

Bob Sheldon: Did you ever publish a bookout of it?

Seth Bonder: No, I never made a book out ofit. But a lot of people still have the notes anduse them as reference materials. I noted earlierthat much of the theoretical underpinnings forthe ‘‘Bonder-IUA’’ and subsequent models (dis-cussed later) were codified in a large SRL reportreferred to as the ‘‘Bible.’’ The course notes con-tain some of this material.

In 1965 I met Wilbur Payne for the firsttime. He was the Deputy Under Secretary of theArmy for Operations Research [DUSA (OR)]and was running the Hawthorne Committee.



He had obtained my name from an Army officerwho worked for me at OSU. He called me atMichigan and asked if I would join the com-mittee which had many senior Army and OSDmembers on it. It was a large committee. Proba-bly because of my lack of knowledge about thesubject, I asked some embarrassing questionsat the first meeting. Wilbur loved it and we be-came fast friends and continued for the rest ofour professional lives. For many years after thisinitial meeting, we spent a lot of time togetherworking on committees, discussing Army issues,brainstorming new approaches and methods foranalysis. I tutored him on some new stochasticprocess book, and drinking. Over the years wedid a lot of the latter at his houses in Washington,D.C. and El Paso, Texas, in Juarez, Mexico, andon temporary duty (TDY) together. Wilbur workedfor me for 2–4 days per month as a VRI Associ-ate after he retired and until he died in 1990.Wilbur was bright and dedicated to makingthe Army better, and had a unique talent forgetting change to happen in the DoD bureau-cracy. I learned much from him, and he sur-faced at many places in my career.

Bob Sheldon: In what way?Seth Bonder: He was instrumental in my get-

ting involved with the Army Science Board, wasa strong advocate for implementing the newmodeling approaches I developed in my re-search, and, as I said, was a VRI associate fora number of years after he retired from govern-ment service.

While still in the 1965–1972 period, in addi-tion to my teaching, I directed the SRL at theUniversity of Michigan which conducted exten-sive research on military operations, some ofwhich I discussed earlier. Most of that workwas theoretical in nature with lots of mathemat-ical formulations. Although our research wason a spectrum of combat processes, we did nothave a model, per se, to use for analysis of com-bat issues. In August of 1969, I received a callfrom Dick Trainor, who was my sponsor fromthe Army’s AVCS office, asking me to meet withthe AVCS, General Bill DePuy (then a three-stargeneral). I had not met him before, but appar-ently he had an innate feeling for the power ofanalysis.

At the meeting, General DePuy said, ‘‘I’mhaving some problems with Mr. Packard (who

was the Deputy Secretary of Defense) on thenew Main Battle Tank (MBT-70). I need someanalysis of the MBT-70 alternatives, but theLockheed Corporation wants to charge me anoutrageous amount of money to do the studyusing their new IUA simulation model. I under-stand we’ve been giving you money to buildmodels of warfare. I want you to use them todo an analysis to determine the pros and consof the MBT-70 for my meeting with Mr. Packardin mid-November. I want the results by 1November.’’ I explained I had two problemswith his request: (1) my theoretical research wasto develop mathematics of combat processes—I did not have a credible model of small unitland combat to do the study; and (2) the studyis obviously going to be classified and I can’t doclassified research at the University—it hada moratorium on classified research. DePuytold me he needed my help so I should usewhatever mathematics I had to do the study,and that I should start a company to handlethe classified material. I said, somewhat face-tiously, ‘‘Right, I’ll go start a company.’’

At the end of that week, I had security folksfrom the Pentagon come to Ann Arbor to clearan office in a company that I didn’t have. I didn’thave an office, and I didn’t have a company. Butthat’s how VRI got started. Bill DePuy was thestimulus behind the founding of VRI, with me,Bob Farrell, and an economist friend, DaveBrophy, as the founders. The security peoplesaid, ‘‘We’ll be back next week. Get an office.’’I went to a friend of mine who had a littlecompany and he gave me a desk, and I boughta safe. They came back and said, ‘‘This isn’tgood enough. You have to have an office.’’I rented a room in a motel-like building andmoved the safe there. (The safe moving processbecame a classic traditional story in Vector’shistory.) They cleared the office. Bob and Imoonlighted from our university responsibili-ties and did the study from September to mid-November. In addition, since I was President ofMORS at the time, I had to attend the MORSSymposium in early November.

Bob Sheldon: Interesting way to start a com-pany. Did you complete the study?

Seth Bonder: Yes, and it became the seed ofmy career after teaching and research at theUniversity. It is interesting to remember this



stressful three-month study because it gaverise to the first hybrid analytic-simulationmodel of land combat engagements, later re-ferred to as the Bonder-IUA model. Soon afterthe office was cleared, we received descrip-tions of the different MBT candidates; descrip-tions of scenarios for six different land combatengagements involving battalion size forces;100,000 IBM cards containing performancedata for the different MBT candidates, terraindata for the scenarios, etc. We began a hecticprocess of building a new model of land com-bat from our theoretical formulations. Bobwrote code to implement the formulations andthe scenarios (because I can’t type!), and I modi-fied the mathematics to represent the differenttypes of weapon systems in the scenarios andtheir different firing doctrines.

Then we had to drag boxes of IBM cardsover to the computer center to make some trialruns with the candidate MBT systems. JohnHonig was the government project officer forthe study. He wanted us to verify that our ana-lytic model would produce results similar tothe IUA simulation before making runs foranalysis of the candidate systems. (I was notsure why that was a good credibility criterionfor our model.) Our first set of test runs didn’tmatch at all—they were way off. I reviewedour results and believed them to be consistentwith the mathematical structure of our model—that is, the code that Bob had written was cor-rectly transforming the model inputs and ourmathematical formulations to output combatresults for the test cases. We reviewed the IUAsimulation code and found numerous program-ming errors as well as some different decisionlogic representations between our model andthe Lockheed simulation. Once these were cor-rected, we and Lockheed made some new testruns. John made this a blind comparison—weprovided him with our results which he com-pared to Lockheed’s. John thought the compar-ison was good enough to go forward with thestudy. That is, the deterministic results of ourhybrid analytic-simulation compared quite wellwith the mean of the results from the IUAMonte Carlo simulation over the six differentengagement scenarios.

Bob Sheldon: Was that like a verification, val-idation, and accreditation (VV&A) process?

Seth Bonder: Not quite. We did not compareour results to real combat operations (which istoday’s validation process), but did more thancheck to see that the code implemented ourmathematical formulations (which is today’sverification process). We compared our hybridanalytic–simulation model’s results to resultsfrom the IUA simulation which led to the Armyapproving (this is today’s accreditation process)our model for use in the study.

In the remaining time we made hundredsof runs over the six different engagement sce-narios and analyzed them in our make-shiftmotel office. Amidst boxes of pizza, we calledthe results into John on a Sunday evening whothen prepared a briefing for General DePuy topresent to Mr. Packard that week. General DePuywas very pleased with the results and our re-sponsiveness. John referred to our battalion-level combat model as the ‘‘Bonder-IUA.’’ Laterhe and I wrote a paper describing the study,presented it at a MORS Symposium, and wonthe Rist prize.

The first VRI study in November 1969 wasa success and had many positive downstreamimpacts on VRI’s future reputation and growth.The structures and equations developed in theSRL and used in the Bonder-IUA (and subse-quent) models appropriately became knownas the Bonder-Farrell methodology. Because themethodology and models involve differentialequations, and because Lanchester pioneereddifferential equation models of attrition, somepeople call these Lanchester models. To theextent that this is intended as a well-deservedtribute to a pioneering analyst, this is fine ter-minology. However the methods of Bonder-Farrell are synthetic in nature and are quitedifferent from Lanchester’s, which are holisticmodels.

So VRI was formed in 1969 at the impetusof General DePuy, with me and Bob Farrell asprincipals. I moonlighted with VRI for a coupleof years during which we hired some early em-ployees (most notably David Thompson, GeorgeMiller, Peter Cherry, Dick Freedman and AlanWeintraub, all of whom worked for the com-pany for 301 years), and moved into a sizableoffice in 1971. Although I was due for a sabbat-ical, I took a six month leave of absence fromthe University the same year. In 1972 I left my



full-time faculty position at the University ofMichigan, became an Adjunct Professor, andjoined VRI as its full-time Chairman, CEO, andPresident. Bob joined as the first Vice Presidentand Chief Scientist.

Bob Sheldon: After starting Vector, did youcontinue activities with the University?

Seth Bonder: I did. I periodically taughtcourses in decision theory and the modelingcourse after 1972. The period 1960–1972 was fo-cused in academia and was a busy and pro-ductive period of my professional OR life. Theperiod 1972–1989 started the OR business partof my professional history which, like the previ-ous period, was affected by many related eventsand activities of others.

In 1972, Bill DePuy formed and became theCommanding General of the Army’s new Train-ing and Doctrine Command (TRADOC). In thisrole he became the architect of the post-VietnamArmy, responsible for the planning of all theArmy’s new tactical and operational-level sys-tems, new force designs (organization andcomposition of fighting units), new operationalconcepts and doctrine (how units should fightand be supported), and for schoolhouse train-ing. And he became a champion of operationsanalysis to help him with these responsibilities.He formed a number of OR analysis groups inthe command: sizable ones at Fort Leavenworth,White Sands, and Fort Lee; and smaller ones atmany of the branch (e.g., armor, infantry, artillery,etc.) schools.

Bob Sheldon: Did you provide support tohim?

Seth Bonder: Yes, to him and the commandfor many years. At the tactical level of opera-tions, we (that is, me and my colleagues atVRI) and the Army analysis community hadextensive experimental, research, and modelbuilding activities to use as a solid base forconducting analyses at that level. In the early1970s, VRI began providing analysis support toTRADOC for the development of new tactical-level systems such as the M1 tank, the Bradleyfighting vehicle, the Apache helicopter, andothers. We provided analytical support for newtactics, and the incorporation of new tech-nologies such as fiber optic, laser-designated,self-designated precision munitions into newsystems.

Our initial studies for the Army in the early1970s used the Bonder-IUA model to examinethe impacts of different armored systems onthe outcomes of battalion-level engagements.Even though we had used it to support Gen-eral DePuy in his successful negotiations withMr. Packard, it was not an easy sell to convinceother senior military and the Army’s analyticcommunity that it was an appropriate vehiclefor analysis of small unit combat engagements.The model, which I referred to as a ‘‘hybridanalytic-simulation model,’’ considered groupsof homogeneous systems (i.e., multiple systemsof a single type all at one location) on the battle-field and analytically integrated their detection,firing, accuracy, lethality, and vulnerability sto-chastic processes to compute attrition deter-ministically during the course of a battle. Itsimulated movement of the groups over theterrain, command and control of the groupsthrough decision rules, and LOS computationsbetween combatants by embedding a digitizedmap into the model. This was in contrast to theArmy’s primary model of small unit engage-ments, Carmonette, which kept track of indi-vidual systems and explicitly sampled eachof the stochastic processes using Monte Carlotechniques to develop a distribution of attri-tion results during the course of the battle. Al-though both were useful tools, I personallythought the hybrid model was easier to use ina study. It ran faster to simulate a battle, didnot require replications, and most importantly,I thought it was easier to interpret the exten-sive outputs by referring to the underlyingmathematical formulations. It allowed us toperform the necessary extensive parametricand sensitivity analyses during a study.

Bob Sheldon: Were you successful in gettingthe Army to use the model?

Seth Bonder: Only with extensive influenceand help from Wilbur Payne, initially in his po-sition as the DUSA(OR), (the highest level ORposition in the Army), and later when Bill DePuymade him the Director of the TRADOC SystemsAnalysis Agency (TRASANA) at White Sands.We started doing studies for DePuy and Wilburand after a while various Army agencies be-gan using the Bonder-IUA along with theirother models. Using a ‘‘prototyping process’’ ofmodel enhancements to conduct studies of new



systems, technologies, and tactics, VRI and thegovernment developed a large family of hybridanalytic-simulation models of battalion-level en-gagements during the period 1972–1989. A VRIbranch of the developments led to the BattalionLevel Differential Model (BLDM) which wasused at least through 1990 and the Bonder-Farrellmethodology was adapted for use in the JANUSgaming version which I believe is still being usedtoday.

I think it is important to remember that thisfamily of battalion-level models was built ona strong base of field experimentation (directedor performed by Army analysts such as WilburPayne, Dave Hardison, E.B. Vandiver, Walt Hollis,Marion Bryson, Hunter Woodall, and others),and theoretical research using this experimentaldata. They were not built on a ‘‘base of sand’’ assome military analysts who were not involvedwith or aware of the experimentation and re-search like to publicly broadcast. As we willdiscuss shortly, these models later became thebasis for building the VECTOR series and othercampaign models used in the DoD.

While he was still the AVCS, Bill DePuy andsenior leaders in the Air Force, JCS, and OSDwere expressing new interest in operational-level issues (corps and echelons above corps)—new operational concepts and doctrine, newforce designs, and new operational-level sys-tems. This was perhaps driven by the Sovietsfocus on the operational level of war as de-scribed in a number of classified documents.Efforts were underway to develop or enhanceexisting operational-level models that couldconsider the effects of combined arms (multiplearmy branches) and joint (multiservice) forcesin a large-scale campaign.

Bob Sheldon: What were some of the cam-paign models?

Seth Bonder: If I remember correctly, therewas IDAGAM (Institute for Defense Analy-ses Gaming Model) developed by the Insti-tute for Defense Analyses (IDA). This laterbecame TACWAR (Tactical Warfare simulation)that used an eigenvalue methodology to assessattrition in large unit battles. And there wasATLAS run by RAC which in the early 1970sused a FPS approach to assess attrition andmovement of major units. Because of DePuy’sinterest in operational-level issues in 1970, we

did some research in the SRL at Michigan onthe possibility of cascading our battalion-levelcombat methodology into a campaign model.

In early 1971, I received a telephone callfrom Air Force Lt Gen Glenn Kent, who wasthen the Director of the Weapons Systems Eval-uation Group (WSEG), an OSD agency that Ibelieve served as the military side of IDA. Hesaid he had heard of my modeling researchefforts and wanted my opinion on some of theongoing campaign model development efforts.At a subsequent meeting I told him I thoughtthe IDAGAM eigenvalue approach led to somevery unstable attrition predictions and that theFPS approach used in ATLAS was an unscien-tific hypothetical construct that was not credibleto assess the utility of future systems or forces.I think he agreed with me, but then out of theblue said he wanted me to go to Europe forabout six weeks on an intelligence related taskfor the Secretary of Defense. I told him I thoughtit would be interesting, but couldn’t do it sinceI had just taken a leave from the University toget a new company off the ground. He saidhe wanted me to go and would have somecampaign modeling work for me to do whenI returned.

Bob Sheldon: What did he think your qualifi-cations were for the European task?

Seth Bonder: I had no idea then but after thefact believe he thought I had sufficient techni-cal training in OR and reasonable knowledgein the state of the art in military OR modelingand analysis. (Glenn has confirmed that theproject and associated reports discussed belowwere declassified many years ago.)

The project involved interrogation of a de-fector from one of the Warsaw Pact nationswho claimed to be a military OR analyst knowl-edgeable in Pact and Soviet modeling and anal-ysis activities. I went to Europe with two otherindividuals—one from the Central IntelligenceAgency (CIA) and one from the Defense Intelli-gence Agency (DIA)—to assess his credibilityand learn what we could about the Pact’s anal-ysis capabilities and analyses. We operated outof the Fifth Corps HQ (the I. G. Farben build-ing) in Frankfurt and met with the defector atsome safe house daily. The other two analysts(clearly intelligence; not OR) worked with meonly part of the time I was there. During our



discussions, I tried to evaluate his technical ORcapabilities, his knowledge of Pact and SovietOR analysis methodologies, his knowledgeof ongoing studies, and other topics I thoughtwould be of interest to the DoD. Although hehad some superficial knowledge about US tools,(e.g., Carmonette and my attrition rate work),his technical capabilities were light, and hisknowledge about Soviet studies was limitedto artillery firing table developments. I believedhe had very limited access to what the Sovietswere actually doing. I drafted my report toGlenn on the weekends at an office in the Farbenbuilding.

Although it sounds like a benign assign-ment, I should have asked Glenn for hazardousduty pay due to the Baader-Meinhof gang. Onenight I was evacuated from the VIP quartersdue to a bomb threat. A bomb exploded at thefront of the Farben building hours before mymorning jog past the building, and a bomb ex-ploded near the Officers’ Club one day whichblew out all the windows on the backside ofthe Farben building.

Bob Sheldon: How long were you in Europeon this task?

Seth Bonder: It was a little less than the sixweeks Glenn thought it would. When I returnedand debriefed him, General Kent later askedme to start a program to build a campaignmodel that didn’t rely on the techniques usedin IDAGAM or ATLAS, but was more related tothe work I had done in developing the battalion-level models. I thought this was appropriatesince, as I noted earlier, our models had stron-ger linkages to extensive field experimentationand related research.

While still in the SRL, I, Bob Farrell, anda couple of others had given some thoughtto nesting or embedding the hybrid analytic-simulation battalion-level models inside a cam-paign model to compute the land warfare resultsduring a campaign. (This is in contrast to the hi-erarchy approach used later in CEM [ConceptsEvaluation Model] whereby land combat attri-tion data are input to the campaign model formany different situations based on runs of aseparate ground combat model). Bob thoughtwe would have some major computational prob-lems in doing so. A campaign could involvea number of corps-size units on the friendly

side, with each corps containing about 25 battal-ions. Recall the Bonder-IUA (and all succeedingones in the differential family) had an embed-ded digitized map of the terrain which was usedto locate and move system groups and to com-pute the duration of LOS dynamically betweenall potential shooters and targets. The latter wasa major share of the run time for a single battal-ion engagement. Bob estimated that it wouldrequire eons to compute LOS in this fashionfor a large scale campaign. We needed a wayto analytically represent LOS in a campaignmodel rather than try to simulate it as we hadin the battalion models. We considered the fol-lowing logic:

As a shooter moves around a battlefield,he will have a specific realization of LOS win-dows as he approaches a target. If he takes aslightly different path, he will have a differentrealization of LOS windows. Slight variationsin attack paths produce different realizationsof LOS windows. Although in planning stud-ies we usually script specific attack paths, wedon’t know what attack routes will actuallybe used in a future combat operation, even ifit is on the exact piece of terrain used in a plan-ning study. Accordingly, we can think of theseas realizations from some underlying LOS sto-chastic process. Conceptually, as any shooterapproaches a target there is some probabilitythat he will obtain LOS with a potential targetand the duration of that LOS is a random variabledescribed by some probability distribution. Onemight expect that the distribution would havedifferent statistics as the range between shooterand target change. Given this simplified logicand a couple of assumptions, one might mathe-matically characterize the LOS process as a non-stationary Renewal Process.

About the same time, Bob (and I believeWilbur Payne) conducted an extensive analy-sis of the LOS data generated in the TETAM(Tactical Effectiveness Testing of Antitank Mis-siles) experiments. (The TETAM experimentswere conducted by Marion Bryson at the Com-bat Developments Experimentation Command[CDEC] and I believe Dave Hardison, based onobserving in an earlier experiment that therewere a reasonable number of 30001 meter LOSin Europe, developed the tube-launched, opti-cally tracked, wire-guided [TOW] long-range



missile.) Bob’s analysis suggested that the LOSprocess could reasonably be approximated bya Markov Renewal Process with parameters thatvaried with range to the target. We developed themathematics to integrate LOS into an extendedversion of the Bonder-Farrell methodology andprocedures for estimating the LOS parameters(mean LOS, mean non-LOS windows) from dig-itized maps of a battlefield area. Subsequent testsusing digitized LOS in the Bonder-IUA modelwith a version that used the analytic LOS for-mulation produced comparable combat results.

Bob Sheldon: When did you develop the firstcampaign model under the program with Gen-eral Kent?

Seth Bonder: Based on my experience withthe battalion-level models, my approach to build-ing models was (and is) quite different fromwhat I see in the community today. Using manysimplifying assumptions, I believe an initialprototype should be built as quickly as possi-ble, used in studies, and enhanced (continualremoval of simplifying assumptions) basedon those studies and the needs for future stud-ies. I guess in today’s vernacular it would becalled a ‘‘prototyping’’ development process.I can’t prove it, but I believe the recent JSIMS(Joint Simulation System) model developmentprogram failed (at a cost of 1–2 billion dollars)because it didn’t follow this type of approach.

Starting in 1972, the first prototype ver-sion, originally called the BATTLE (BattalionThrough Theater Level Engagement) modelbut soon changed to VECTOR-0, was devel-oped in 10 months. It included the effects ofsome air operations but was primarily a landwarfare model. It was used to address someissues for General Kent. I like to tell a littlestory about one of my interactions with Glennin this time period.

Bob Farrell and I met with him to discussresults of one of the issues we were looking atwith VECTOR-0. Glenn has a wonderful se-mantic technique for boring in on study results(and the briefer!). We were arguing back andforth about a particular result which was notintuitive to him when he said something like‘‘no, no, no, you don’t understand.’’ Sitting onthe side, Bob shrugged his shoulders. Glennwhirled on him, and said ‘‘Don’t look at me likeI’m a stupid general,’’ and went to the white

board to make his point. At the board hestopped, thought for a moment, understoodthe point I was trying to make and said‘‘I guess I am a stupid general.’’ Obviouslynothing could be further from the truth—he’sone of the smartest military analysts I’ve known,but confident enough as a senior military leaderand analyst to admit to missing a subtle result.In our subsequent discussion we agreed thatthe validity of a planning study’s results shouldnot be based solely on agreement with one’sintuition because an individual’s intuition isbased on his or her past experience. Valid model-based analysis about future operations mayappear counterintuitive initially, but eventuallycan be used to enhance one’s intuition aboutcomplex operations.

The issues we addressed for General Kentserved as a basis for developing VECTOR-1in 1974 which added some rear area opera-tions. Using this prototyping process, a lineageof VECTOR campaign models was developedfrom 1974 to 1990, including VECTOR-1 Nuclear,VECTOR-2 (which enhanced the rear area rep-resentations, including complete joint air oper-ations), VECTOR-2 SWASIA (Southwest Asia),VECTOR-3 (which added complete joint logis-tical operations, including intratheater airlift),VECTOR-3 intelligence and electronic warfare(IEW), and other versions as needed to addressrelevant military decisions. New code was writ-ten and new documentation prepared with eachnew major version. This codified all the changesmade to the earlier version as it was used in stud-ies and all the new content (entities, processes,etc.) added to create the new version.

Bob Sheldon: When you added logistics, didit include munitions?

Seth Bonder: Yes, for all service systems. Itincluded all classes of supplies, transshipmentof supplies, maintenance activities, and all otheractivities of a logistical system integrated withcombat operations.

Bob Sheldon: Did the VECTOR models gothrough the VV&A process?

Seth Bonder: I don’t think so, but let me tellyou what I do know regarding the VECTORmodels and that process. In the early 1990s,VECTOR-3 was accredited by TRADOC Analy-sis Center (TRAC) or Walt Hollis for VRI to dosome deep strike studies that were mandated



by Congress. Perhaps more significantly, in1982 I directed a study for the CIA in whichVECTOR-2 was successfully blind tested againstthe 1973 Golan Heights campaign (before per-forming a Mideast balance analysis for thesame sponsor). This was what I believe wouldbe called a validation study in that we comparedthe results of VECTOR-2 when it simulated theGolan Heights campaign to historical results ofthat campaign. I believe that study producedtwo useful results for the military OR commu-nity and profession.

I believe it is the first campaign model thathas been ‘‘validated’’ against real-world mili-tary operations and the study provided a logicalmethodology for doing so. VECTOR-2 is a de-terministic model which produces a single,central-value realization of a simulated war drawnfrom some underlying simulation stochastic pro-cess. I expected that we would be comparingthis to a single realization of the actual wardrawn from some underlying real-war stochas-tic process. It turned out that there were eightdifferent reports of how the war unfolded (i.e.,unit locations, attrition, etc. over the course ofthe war). There were two classified reports andsix unclassified ones. (I’m not sure if anyonecan ever really determine how a war unfolded.)From a methodological perspective we treatedthe eight reports as eight realizations of the un-derlying stochastic process of the real war andtested the hypothesis that the VECTOR-2 reali-zation came from the same underlying stochas-tic process. It was a blind test in that we didnot see any of the eight realizations until thecomparisons were made. As input, the CIA pro-vided us with system capabilities, force listsand how each side (Israelis and Syrians) wouldemploy them.

The comparisons suggested that there wasno reason to reject the hypothesis that the dy-namics of the VECTOR-2 simulated war wassimilar to the dynamics of the actual GolanHeights campaign. From a broader perspectivefor the military OR profession (not just theVECTOR models), the results verify (or do notreject) our implicit hypothesis that we canmodel the operations of large-scale militaryconflict. About the same time AMSAA was test-ing one of their division-level models (which Ibelieve used the Bonder-Farrell methodology

to assess ground combat operations) againstsimilar data.

Bob Sheldon: You modeled the weapon sys-tem capability. Did you model the differencein the capabilities of the various countries’ground fighters like Trevor Dupuy did in hisHistorical Evaluation and Research Organiza-tion (HERO) model?

Seth Bonder: VRI and Mathematica hada contract to review the sources of the HEROdata and its statistical methods years ago. Atthat time I thought the HERO model contentwas spurious so don’t want to discuss how herepresented different countries. In this vali-dation study, the CIA provided us with thesystems’ capabilities as used by the differentcombatants. In the first iteration of compari-sons, they neglected to tell us that a Syriantank crew would abandon their tank if a shotfired at them was a near miss or hit their tankbut didn’t cause any damage. The CIA countedthat as a killed tank and thus estimated signifi-cantly more Syrian losses that we did. The com-parisons were much better when we simulatedthe same Syrian behavior.

Although the VECTOR models have notgone through the formal VV&A process, the sto-chastic LOS version of the Bonder-Farrell attri-tion methodology used in the VECTOR modelshas been used by other modelers in campaignmodels such as Vector-in-Commander (VIC),STAR, Eagle, and most recently the Joint War-fare Simulation (JWARS) model. I believe someof these have been through the VV&A process.I was surprised to learn recently that somemodels are still using a form of FPS methodsto assess land combat results.

Returning to my association with GeneralKent, he was a wonderful sponsor for the de-velopment of the VECTOR campaign models.He had great insight and vision as to the kindsof models needed to support the DoD. We main-tained our professional association after he leftWSEG. It was always a pleasure to occasionallysee him at various meetings. The next directorof WSEG was Admiral Ed Waller who wascharged with closing the WSEG. A smart mil-itary man but not analytically oriented likeGlenn. During Ed’s tenure, I used to meet withhim for breakfast once a month and spend anhour teaching him the basics of decision theory.



They were enjoyable sessions since I really do liketeaching.

Bob Sheldon: Let me backtrack a bit. Wheredid the name Vector come from?

Seth Bonder: I’m not sure whether you meanthe company name or the name of the cam-paign models, but let me comment on both.In selecting a name for the company, we wantedsomething with a mathematical connotation.After exploring a number of the obvious Greekletters (the Alpha Corp, etc.), Vector was sug-gested. We all liked it since it not only had themagnitude and direction associated with a math-ematical vector, but also the n-dimensionality ofa vector. Our VRI logo (created by one of my gradstudents, my secretary, and me) contains associ-ated vector line segments (the arrows) and amathematical vector dot product. My friendlycompetitor Dan McDonald from BDM used tosay, ‘‘They don’t know which way they aregoing—up, down, or sideways.’’

As noted earlier, the original name createdby Bob Farrell was the BATTLE (BattalionThrough Theater Level Engagement) modelwhich was a perfect acronym. But I changedit to VECTOR to get visibility for our new andvery small company. In 1972 I think we had eightemployees. At that time I did not have anygrowth plans or desires, but wanted to buildour reputation as a high quality, innovative re-search and analysis company that could helpaddress potential clients’ most difficult deci-sion problems. That philosophy continued un-til 1988–1989 when I had to establish specificgrowth plans.

Before continuing with my professional his-tory in the 1972–1989 time period, I think it isimportant to note a somewhat random eventin 1972 that led to some of VRI’s downstreamgrowth. That year, strictly on a whim, I respondedto a Commerce Business Daily Request for Pro-posal (RFP) for an analysis of national nursesupply and demand issues. I knew very littleabout the area. Probably the worst proposalI have ever written, but (of course) we wereawarded the contract. That got VRI into thehealth-related business. I hired an Army MedicalCorps captain, Tim Doyle, to run the contract andbuild that side of the business. Over time he de-veloped significant OR, econometric, and in-formation technology (IT) and database work

with the Military Health System. I did not getdeeply involved with VRI’s health business until1994–1995, surprisingly enough at the request ofGeneral Max Thurman. More on that later.

Although we were getting some supportfrom Bill DePuy to do small tactical-level stud-ies, and from Glenn Kent to work on the devel-opment of campaign models, I did not know,and was not known by, the senior-level Armyplanning and operations community. In 1975,General DePuy set in motion activities for anArmor/Anti-Armor Systems Program Review(AASPR) with all the Army’s senior leadershipto review plans for new major armor programs.All the analyses for that program review werebeing conducted at the Combined Arms Cen-ter in Fort Leavenworth, Kansas. About threemonths before the review, General DePuy askedme to meet him at Leavenworth to see what MGCushman (the Commanding General [CG] ofFort Leavenworth) was doing in preparationfor the AASPR. He and a couple of majors pre-sented some slides with the worst analyticgibberish I had ever seen. General DePuy askedme ‘‘What do you think?’’ I said, ‘‘I think hehas no idea what he is doing and may be havinga strong negative effect on preparation for theAASPR.’’ Bill asked me to spend a little timewith Colonel Reed Davis who was directinganalyses in preparation for the AASPR withanother analysis group at Leavenworth, andto attend the major review.

The AASPR was attended by a cast ofthousands—a few Army four-stars, some three-stars, a host of two- and one-stars, and a largenumber of other Army military officers and ci-vilians. It was so large that only the senior lead-ership were seated in the main presentationroom, while the others watched on TVs in ad-joining rooms where I had planned to be. WhenGeneral DePuy arrived with his entourage, hebeckoned me to join him saying, ‘‘I have a seatfor you.’’ I thought that was nice of him. Littledid I realize how nice—I sat with him and Gen-eral Kerwin, then the Army’s Vice Chief ofStaff. Bill introduced me to a large number ofthe current and future Army senior leadershipand set the image that I was someone who,through quality analyses and consulting, couldbe helpful to that leadership. I believe Bill pur-posely did this to help me and VRI.



At that meeting and subsequent ones, I metArmy officers who became clients and eventu-ally lifetime friends such as Generals Max andRoy Thurman, Glenn Otis, Dave Maddox, JohnFoss, Jack Merritt, Ted Stroup, Paul Gorman,Gordon Sullivan, and others. He providedthe entree for VRI to perform work for manyTRADOC agencies, other organizations in theArmy, and the leadership at Department ofArmy (DA) Headquarters in the 1972–1989time period. Using the VECTOR series models(including one that had player interaction capa-bilities), I studied a broad range of operational-level issues in the 1972–1989 period, includingoperational concepts and doctrine (e.g., AirLandBattle doctrine), new operational organizations(e.g., Division-86, the Light Division, Army-86,the Army of Excellence), and operational-levelsystems (e.g., the Army Tactical Missile System[ATACMS], the Joint Surveillance Target AttackRadar System [JSTARS]).

Bob Sheldon: How many studies were youinvolved with during this time period? Do youhave any favorite ones?

Seth Bonder: Not counting my Army ScienceBoard work, prior to 1989 (when we switched toquick response analyses) I would estimate thatI technically directed or closely oversaw five tosix projects per year at VRI for the Army, otherDoD agencies, and industry, so probably 75–80from 1975 to 1988. Many were studies and somemodel development projects. VRI had great tech-nical professionals to work with which madethis workload feasible and stimulating. Evenwith my business development and managerialresponsibilities, I tried to spend at least 30 % ofmy time working on technical projects (basedon my usual 70–75 hour weeks). I was nevera believer of having separate marketing andproduction staffs. In an analysis firm, I alwaysthought that technical folks should be inter-facing with potential clients to determine whatwas needed to help with their decision issuesand to assess what was feasible to achieve inthe allotted time. I believe the pure businessdeveloper did not have these capabilities.

I really liked doing analyses for clients, somany were favorites. I tried to make sure thatthe VRI studies and development projects wereuseful, useable, and used by the clients. I particu-larly liked those in which I not only helped the

client with his decision issues, but learnedsomething about operational phenomena. Someexamples during this period might illustratethe point. Through Glenn Kent I met JasperWelch at the Air Force Studies and AnalysisAgency (AFSAA) when he was a Colonel. Laterwhen he was a one-star and I believe the Assis-tant Secretary of Defense (ASD)-Atomic Energy,he asked me to do ‘‘something bright’’ for theDefense Nuclear Agency, an agency which hada small set of favored contractors that did notinclude me. At that time the North AtlanticTreaty Organization (NATO) strategy against aSoviet offensive in Europe was to try to stoppenetration of the Inter-German Border (IGB)conventionally and if that failed, use tacticalnuclear weapons to stop the penetration. Wethought it would be useful to understand andperhaps develop some good nuclear targetingpriorities. In doing the study we developed thetargeting priorities for our nuclear systems. Inaddition, this study in conjunction with a cou-ple of others, gave rise to a powerful insightabout tactical nuclear operations which I thoughtbrought the whole NATO strategy for Europeandefense into question. So did Jasper. (I can’t de-scribe the specific insight because I don’t knowif the study has been declassified.) Jasper askedme to brief the SACEUR (Supreme Allied Com-mander, Europe) within a couple of weeks. Thestudy was useful to the DoD and I developedsome insights regarding the interactions be-tween tactical nuclear and conventional war-fare operations.

As another example, in doing a numberof studies for General DePuy during the early1970s, we analyzed a large number of simulatedUS battalion-level defensive engagements againstan attacking Soviet force. We observed that theinstantaneous loss exchange ratio (LER) – the ratioof the rates of attacker and defender losses –was very high and relatively independent ofthe threat size early in the battle because of con-cealment and first shot advantages accrued tothe defender. The LER advantage moves to theattacker as the forces become more decisivelyengaged and the concentration and saturationphenomena come into play for the attacker. Thissuggested that an indepth use of a large numberof small-unit engagements in which defendersget off a small number of shots (operate at the



high end of the LER) and then fall back to pre-pared positions to repeat the process would bean effective tactic for Europe. In essence, to pre-vent a Soviet penetration, the concept was totrade ground for high attrition of the attackerwhile continually increasing the density of thedefensive positions. I subsequently wrote a pa-per describing this as a ‘‘Variable-Density’’ de-fense. One of my colleagues convinced me thiswould not be a good acronym, so I changed itto the ‘‘Dynamic Density’’ defense. GeneralShy Meyer who was the CG of the 3rd InfantryDivision (ID) in Europe at the time (later be-came the Army’s Chief of Staff) thought thathe could find a large number of such fall-backpositions in his sector to make the conceptwork. I believe Bill DePuy used these ideas asthe basis for his Active Defense concept. Thestudy helped the client, and I learned some-thing about combat operations.

Bob Sheldon: Sounds like you and GeneralDePuy had a good working relationship.

Seth Bonder: We did and it blossomed intoa lifetime friendship after he retired in 1977.When he retired he wanted to do some con-sulting for VRI. I told him we needed to waita year to avoid any misperceptions related toour contractual work with TRADOC. After 9months of consulting with some of my compet-itors, he complained that they never gave himsubstantive work to do—he would spend mini-mal time on a large number of proposals andprojects each consulting day so they could usehis name for marketing purposes. He becamea VRI Associate shortly thereafter, working twoto four days per month until a year or two be-fore he died in 1992. Most often he would cometo our Ann Arbor office (he loved walkingaround the University’s campus). He’d walkin, salute, and say: ‘‘Reporting for duty.’’ Hewould help fix a project that was having somedifficulties, help structure a new project, or thetwo of us would work on an important issuefor the Army. He often prepared papers de-scribing the work he did on fixing and structur-ing projects or think pieces on Army and Jointservice issues – products not often providedby senior military consultants. He was the firstmember and Chairman of my small AdvisoryBoard (AB) that I formed, not only to reviewongoing programs, but to serve as a sounding

(more often pinging) board on how I was man-aging and leading the company.

Bob Sheldon: Was all of your work at VRIfor the federal government?

Seth Bonder: Initially yes, but through con-tacts I made at the ASB, MORS, and other meet-ing venues, we started doing a small amountof work with defense industry. I believe ourfirst industrial client was Raytheon in the earlyto mid-1970s. Initially it was analysis in sup-port of their air defense programs and thengrew to support their land warfare programs.I believe we had more than 25 quick responseprojects with them over the years. Generally,our analyses examined the operational utilityof the systems that industry was developingand assessed design capability tradeoffs to makethe systems more cost-effective. Through the1970s and 1980s I expanded our industrial clientbase to include Lockheed, IBM, Boeing, UnitedDefense, LTV, and a number of others who weredeveloping systems for the military services.

One of our largest studies for an industrialclient, and perhaps the most difficult one forme, was done for the Georgia Lockheed AircraftCorporation (GELAC) in the 1984–1985 time pe-riod. The purpose was to assist them in the de-velopment of a new intratheater airlifter thatcould be operationally effective in the NorthernArmy Group region of NATO and in SouthwestAsia. It involved the addition of complete jointservice logistical operations to the VECTORseries models to create VECTOR-3, the devel-opment of scenarios and databases for bothregions, and extensive parametric analyses toexamine the operational effectiveness of alter-native airlifters. We had the continuous partici-pation and review by a board of senior Armyand Air Force generals chaired by Bill DePuy.This is where I first met Bill Moore, former Com-mander in Chief (CINC) of the Air Force AirMobility Command and Fritz Kroesen, the for-mer CINC, US Army Europe (CINCUSAREUR)whose sedan was fired on by the Baader-Meinhof group in Germany. He looked likea Hollywood type-cast Army general and wasdeceptively very, very bright. The program wassubstantially underbid and grossly mismanagedby one of my VPs. The project was headed fora gigantic loss (about a year’s earnings forVRI), and possibly default. At Bill’s urgings,



with six months remaining I relieved the VP,and spent five to six months of 15–17 hour days,six days a week managing and working on theproject team to complete it with about 50% ofthe anticipated loss. I was so fatigued that a car-diologist friend made me take a stress test andgo through a catheterization procedure. He as-sured me nothing was wrong that a long beachvacation wouldn’t cure. He was right.

Not long after that I directed what wasprobably the first operational effectivenessanalysis of a potential stealth (low-observable)fighter for the Lockheed’s Skunk Works. Itwas interesting to learn that it could have a dra-matic impact on an air campaign if employedappropriately. It not only was itself more sur-vivable, but its low-observability made it easierfor it to take out the threat’s air command andcontrol system, thus making all the nonstealthyfighters in the fleet more survivable.

Bob Sheldon: Did you do any work with theEuropean OR community?

Seth Bonder: Yes. That’s a dimension of mycareer that started with a 1974 NATO conferencein Munich organized by Reiner (Sam) Huber,who then was the director of military analysis atthe IABG (Industrieanlagen-Betriebsgesellschaft)of Messerschmitt Aircraft. Although I had nevermet him, he invited me to present a paper andchair a session on new combat models. (Thatwas the beginning of a lifelong friendship withSam and his family, with visits in Germany andthe US every couple of years.) I was also askedby Ted Roderberg, the Director of OR at theSupreme Headquarters Allied Powers Europe(SHAPE) Technical Centre (STC), to presentsome talks to his staff before attending theMunich meeting. At the Munich conference Imet a number of other European analysts includ-ing David Dare, Jeff Hawkins, Hans Hoffman,John Gibson and others from many differentcountries whose names I have unfortunatelyforgotten. In subsequent years I and my VRIcolleagues attended a number of other con-ferences with this community, which led tostudies for some of them in the 1970s and1980s. A number of the analytic groups through-out Europe implemented various versions ofthe Bonder-Farrell hybrid analytic-simulationdifferential models of combat for analyses ofbattalion-sized engagements.

The most significant project was performedfor STC during the period 1979–1982. The pur-pose was to perform an Armor/Anti-armor ca-pabilities analysis for each of the eight Corpssectors in NATO’s Central Region using theVECTOR-2 model, and in the process, transferthe model to STC and train their analysts inits use. The study required an enormous datacollection effort to obtain data for each of theCorps and their opposing Warsaw Pact threats.The training was accomplished by having theSTC analysts work hand-in-hand with the VRIanalysts, some of whom were stationed at STCfor extended periods. The study was completedand Corps-by-Corps analysis results providedto NATO’s senior leadership in early 1983. Basedon some of the initial work by Paul Haraschu atSTC, Bob Farrell built a methodology for con-structing an aggregate ‘‘meta-model’’ to simulateVECTOR model campaigns based on previouscampaign results. He applied and tested themethodology using all the STC results to developthe first version of the ‘‘MACRO’’ model whichcould simulate central region campaigns ordersof magnitude faster than VECTOR-2. The meth-odology and subsequent versions of MACROwere extremely valuable in conducting analysesafter the Cold War ended.

Mike Garrambone: You’ve talked about yourwork in the 1972–1989 time period, and I knowyou were involved with other professional ac-tivities besides MORS. Could you comment onthese?

Seth Bonder: Sure. In addition to periodi-cally giving some talks at MORS meetings,I was actively involved with the ASB, the Op-erations Research Society of America (ORSA),the International Federation of Operational Re-search Societies (IFORS), and with advisoryactivities to a number of universities. Let mecomment on each of these.

My association with the ASB started inthe early 1970s and was instigated by WilburPayne. Wilbur and Dave Hardison wanted toput an operations researcher on the Army Sci-ence Board. (In those days it was called theArmy Science Advisory Panel – the ASAP).The Army’s Chief Scientist, Marvin Lasser,refused stating they just wanted technologistson the Board. Wilbur and Dave threatened toform an ORSAP, an OR Science Advisory Panel,



with a member of one on it, me. Marv was wor-ried about the competition so he agreed tohave an OR type on the Board. I believe Iwas appointed in 1972–1973 and have beenon it for many years.

In the early 1970s there were only 25 fullmembers on the Board, mostly from hardwareindustry firms. Given their particular corpora-tion’s business, it was easy to identify potentialconflict of interest issues on ASB studies sothey could recuse themselves when appropri-ate. Not long after I joined, Norm Augustine(then the Army’s Assistant Secretary for Re-search and Development [R&D]) told me hehad received a letter objecting to me (an ORtype) being on the Board because (as part of myVRI business) I did studies across a spectrumof Army problem areas and couldn’t be objec-tive in performing a study for the ASB. Norm’sresponse to the letter was typical Augustine:‘‘If he were objective he wouldn’t know any-thing.’’ I stayed on the Board.

Bob Sheldon: How long were you on theASB?

Seth Bonder: I was on the ASB until 1978 andleft after an argument over the Copperheadsystem with Percy Pierre who was the new As-sistant Secretary of the Army for Research, De-velopment, and Acquisition (RDA). Percy wasa big advocate for Copperhead (which was alaser-guided artillery round), but apparentlywas having some problems selling it to Con-gress. At one of the plenary meetings, he andthe four-star CG of the Army Materiel Command(AMC) touted Copperhead’s virtues. Having re-cently completed some studies on the system,I pointed out that it would be effective if andonly if the Army made some significant doc-trinal changes for its employment. Percy was fu-rious. At break time he cornered me and said,‘‘I’m reconstituting the ASB and if you want tobe on board the ship when it leaves the dockI want your support for Copperhead.’’ I real-ized he was trying to buy my advocacy so Igot upset and said something like, ‘‘What makesyou think I would want to be on your ASB?’’Needless to say I wasn’t.

The story raises an important principle, atleast for me. I believe analysts should be objec-tive in developing and providing informationto assist clients in making rational decisions.

The client may use that information to advocatetheir position on an issue. I don’t believe theanalyst should become an advocate – you needto maintain your reputation as an objective de-veloper of the information. I have no problembriefing study results if they support the client’sposition. And I have no problem with clientsnot using my study results if they don’t supporttheir position. I do have problems when clientswant to ‘‘modify’’ some of the results as a basisfor advocacy–then my name comes off the study.Ethics and principles are critical in maintainingone’s professional reputation.

Bob Sheldon: That’s what General Kent saystoo.

Seth Bonder: That doesn’t surprise me. Heunderstands the importance of honest and ob-jective analysis.

I was off the ASB for four years, and in 1982the new Assistant Secretary Jay Scully put meback on the Board. I was on through 1992 whenI quit due to business commitments and sug-gested that Peter Cherry become a member. Iwas reappointed in 1998 when Peter’s tour wasup. My latest tour on the ASB expired in 2005.

As with MORS, I met some wonderfulbright individuals on the ASB (such as RussO’Neill, Norm Augustine, Larry Delaney, JackVessey, Larry O’Neill, Gil Decker, Walt Laberge,Dick Montgomery, Joe Braddock, GeorgeSingley, and many others) who were dedicatedto helping the Army. Many of these individualshave become lifelong friends. Although I workedon a number of studies for the National Acad-emy of Sciences and the National ResearchCouncil, the ASB was the primary vehicle forperforming many of my pro bono study activitiesduring the 1972–1989 period. One of the earlystudies was analysis of the Dragon antitank sys-tem operational test data to help General DePuy(then CG TRADOC) make a procurement buydecision. Another helped General Miley (thenAMC Commander) determine how to fix theSAM-D (now Patriot) air defense system andits testing process when one of the first onesfailed and essentially fell off the test pad. In1983, I chaired a panel for Secretary Scully toreview the development of the Army’s AllSource Analysis System which was experienc-ing significant delays. With the help of somebright ASB intelligence and IT members, we



determined that applications software devel-opment was the critical problem causing de-lays. I recommended that an initial system withlots of memory be procured from the black com-munity, fielded to an operating division with acadre of programmers, and that the softwarebe developed jointly with the operating per-sonnel. We estimated a full Division set couldbe operational by 1987. The Under Secretarycomplained that I was going around the estab-lished acquisition process (which was correct),and gave the program to one of the nationallaboratories to complete. I don’t believe any-thing was fielded until the late 1990s.

One of the nice benefits of being on the ASB,at least years ago, was the occasional opportu-nities to be with the operational troops in thefield. As an example, in the mid-1980s five ofus on an Intelligence and Electronic Warfare(IEW) panel were invited by General Pat Criser,CG of the 3rd ID, to observe EW operations aspart of the REFORGER (Return of Forces toGermany) exercises in Germany. During theexercise, we were helpful in getting EW activi-ties more integrated with the operations sideof the division and I was able to observe the‘‘fog of war’’ first hand. For a couple of daysPat put us in a helicopter to observe operationsof various units during the exercise. At the com-mand briefing one evening, one of the unitsreported that it had successfully fought its wayacross the Donau River, which I knew was nottrue since I had just returned from that unita few minutes before. I expect today’s IT capa-bilities might prevent errors of this type.

Perhaps the most memorable enjoyableASB study, and possibly the most useful one,was the ad hoc ASB study I chaired in 1986 todevelop competitive strategies for defeating theUnion of Soviet Socialist Republics (USSR) andthe Warsaw Pact. The ASB Chairperson wantedme to have 10–15 ASB members on the study.Fortunately, I was able to argue for a muchsmaller team which included Larry Delaneyand Larry O’Neill –both broad thinkers withstrong technical capabilities. This study gaverise to some new doctrinal thinking and starteda large number of black programs, many ofwhich have resulted in technologies used byour operational forces in the field today. I thinkthis study is an example of the wise axiom that,

given a critical mass, productivity of a studyteam is inversely proportional to the numberof team members.

Mike Garrambone: Any relation betweenyour ASB and ORSA activities?

Seth Bonder: No, not really. The ASB is a sci-entific advisory body while ORSA is an ORprofessional society, although it has been in-volved with a few national issues such as scien-tific assessment of the START-II Treaty. I wasORSA’s representative and Norm Augustinewas the Aeronautical Society’s representativeto discussions with the State Department re-garding the impacts of START-II. Since ORSAwas heavily governed by academics, therewas a little MORS overlap with folks like JackBorsting and Dave Schrady. ORSA is the venuewhere Saul Gass and I became professional col-leagues and good friends.

I joined ORSA as a student member in 1962.I started attending the national meetings aftergraduating from OSU in 1965 and became activein the Military Applications Section (MAS) formany years. I chaired the MAS in the late 1960sor early 1970s, but don’t remember the exactyear. I ran numerous technical sessions, orga-nized and chaired a national meeting, becameSecretary in 1974 and President in 1979. Thatwas a difficult time financially for the Society.We only had about 20% of our annual expenses(for the meetings, journals and the administra-tive office) as a cash reserve. Any slip in revenuesfrom dues, journal subscriptions, or meetingattendance would have been a disaster. DaveSchrady was Treasurer at the time. With hishelp and Council approval, I drastically cutexpenses, raised dues by almost 300 percent(that made Council members very nervous),and set a policy that no new initiatives couldbe started until the cash reserve equaled theannual expenses. I cut all journal pages in half,which had all the editors screaming at me andDavid. But the Society survived and grew overthe years with a sound financial basis.

For many years, The Institute of Manage-ment Sciences (TIMS) and ORSA ran competi-tive meetings and a lot of redundant activitiesfor essentially the same membership. Duringmy active time on the Council, we combinedthe meetings and (with the help of Sid Hess)I created the concept of a Joint Council of the



two organizations to minimize the redundantactivities between them. This trend culminatedwith the merger of ORSA and TIMS in themid-1990s into the Institute for Operations Re-search and Management Sciences (INFORMS).

As I did through my activities with the Uni-versity of Michigan, VRI, the ASB, and MORS,I developed professional associations, and insome cases lifetime friendships, with yet an-other group of professionals. Some of theseinclude Al Blumstein (Carnegie Mellon), SaulGass (University of Maryland), Jack Borsting,Dave Schrady, Bill Pierskalla (UCLA), MikeThomas (Georgia Tech), Tom Magnanti (MIT),George Nemhauser (Georgia Tech), Art Geoffrion(UCLA), Dick Larson (MIT) and many otherbright academics and researchers. (Some evenknew how to have fun—you should ask SaulGass about the late evening parties after longCouncil meetings and dinner.) I have very fondmemories of actively working with all of themfor many years to establish and grow the ORprofession. I was able to expand these pro-fessional activities on an international scale bypresenting plenary papers at the tri-annualmeetings of IFORS – an organization with over130 nation members –and even more so whenI was elected as a Vice President of the Federa-tion in 1985.

During this period I was also involved withORSA’s founding of the ‘‘Visiting LectureshipProgram’’ and was one of its early lecturers.The objective of the program was to expandthe academic side of OR by having seniorprofessionals visit with departments who hadexpressed an interest in starting an academicOR program in their university. The lecturerswould present applied or theoretical researchseminars to interested faculty and students,participate in a question and answer (Q&A)session about the OR profession, meet with po-tential faculty regarding possible curriculumdesigns, and other activities to promote thedevelopment of OR programs and attract stu-dents to the OR profession. Because of my aca-demic activities via ORSA and the Universityof Michigan, and my applied OR, engineeringand business experience through VRI, I haveserved on a number of advisory committeesfor engineering and business schools through-out the country, including the University of

Southern California (USC), Virginia Tech,OSU, University of Miami, and others.

A couple of points before turning to the1989–2000 period of this history which wasexciting for me professionally and challengingfor me personally. I noted earlier that the Armyhad performed extensive experimentation inthe 1950s, 1960s, and 1970s to develop funda-mental knowledge about combat processes atthe tactical level which we used as a basis fordeveloping many of the combat models. Muchof this experimentation was conducted at theArmy’s Combat Developments Experimenta-tion Command (CDEC). In 1981 there appearedto be a shift to more operational testing of sys-tems at CDEC and less real experimentation.This trend continued until the early 1990s whenI believe CDEC was closed.

Bob Sheldon: What was the rationale for thatdecision?

Seth Bonder: I really don’t know. TRADOChad some good senior leadership at that timewho understood the value of analysis andexperimentation, including Generals DePuy,Starry, Otis, Thurman, and Foss. The best per-son to ask would be Marion Bryson who ranCDEC for many years. Perhaps there was nolonger a need for it then. I believe there clearlyis a need today given the major changes in op-erational processes and technologies associatedwith net-centric warfare, systems-of-systems,counterinsurgency operations, stability opera-tions and other new activities for our operatingforces. I don’t think we have an empiricallybased understanding of these processes fortheir effective implementation or experimen-tal data as a basis for building models to helpus develop this understanding.

As noted earlier, I did not have any growthplans when I started VRI, effectively in 1971.I was primarily interested in establishing a qual-ity reputation for the company. I enjoyed doingtechnical work and didn’t want to be just a cor-porate manager. Although we had grown ev-ery year until the mid-1980s, after a couple ofminimal growth years, Bob convinced me weneeded to develop plans for more aggressivegrowth. This would provide needed career-growth paths for long-term employees and al-low us to bid on larger programs in the DoD.I decided we would try to achieve this growth



internally (i.e., not by acquisition) by expandingour military and health analysis lines of busi-ness and by leveraging these to make an entreeinto the IT business. Much of this happened inthe 1989–2000 period of my career. This is prob-ably a good time to segue into this period ofdramatic changes in my professional and per-sonal life.

Bob Sheldon: What made it so different fromthe 1972–1989 period?

Seth Bonder: So many things changed. Therewas a major change in the global security envi-ronment which led to changes in how I did stud-ies, the types of decisions addressed, and thedecision makers I worked for. In addition to thetactical and operational-level issues addressedin the previous two decades, in the 1990s myanalysis activities expanded to include strategicand policy level issues for senior leadership ofthe Army, JCS, and CINCs of Unified Com-mands (now referred to as Combatant Com-mands). It also was the first time I becamedeeply involved in VRI’s health analysis busi-ness. I had a major life style change towardthe end of this period, but let me take a momentto remind you what the security environmentlooked like before this period of time as back-ground for these changes.

Bob Sheldon: Are you referring to the ColdWar?

Seth Bonder: Exactly. Throughout most ofthe Cold War, the Soviet-led Warsaw Pact hada significant conventional force advantage overthe NATO alliance—approximately two-to-onestrategic advantage in armored systems and amuch greater operational and tactical advan-tage if they chose to mass in a specific area foran attack against NATO. NATO relied on nu-clear means to deter a massive Soviet offensivein Europe. Strange as it may sound, this wasa relatively stable global security environment.The US and the Soviet Union were military su-perpowers who maintained strong influenceover policies and activities within their alli-ances. The Soviets simultaneously maintainedsignificant control over many Third World coun-try military activities through economic means,technology controls, and military pressure. Al-though regional conflicts occurred during thisperiod, with hindsight from 2006, it really wasa relatively stable security environment.

Although many analysts may not remem-ber this, there was also stability for defenseplanning and associated OR analyses duringthis time frame. And there was very little uncer-tainty in our defense planning activities. Thefocus was Europe—the US and NATO werecommitted to stop a Soviet-led Warsaw Pactattack. The threat was clear. We knew the sizeand location of Soviet and Pact forces. We knewtheir attack options, how they would fight, theirequipment, their command and control pro-cesses, and many other characteristics. Basedon years of analyses, we understood the war-fighting dynamics of a potential conflict, andwe recognized that feasible conventional changesin various components of warfighting capabil-ity (e.g., new systems, force designs, and forcesize) could not alter the conventional forceimbalance. Changes that were made I believewere intended to raise the nuclear threshold.I apologize for the political-military historylesson but it provides the backdrop for why Imade so many changes in VRI’s analysis busi-ness starting in 1989–1990.

Bob Sheldon: What motivated the changesat that time?

Seth Bonder: Eventually it was the end ofthe Cold War, but I started making changesbefore then due to a couple of conversations Ihad with General Max Thurman in 1987. I metMax many years before when he worked forGeneral DePuy, and did some personnel-relatedstudies for him in the early 1980s when he wasthe Army’s Deputy Chief of Staff for Personnel(DCSPER). He was still the Vice Chief of Staffin 1987. We would occasionally get together todiscuss some of the studies I was doing forTRADOC and other Army agencies.

During one of these meetings he casuallysaid, ‘‘Soviet communism will collapse soonand the Soviet Union will be dissolved,’’ whichalso implied that the Warsaw Pact would breakup. I thought he was smoking something or justlosing it. He said, ‘‘The Wall will come down ina couple of years.’’ Max obviously had knowl-edge that I didn’t have. My guess is that hehad been privy to conversations taking place be-tween the Reagan administration and the SovietUnion led by Mr. Gorbachev. Not long after that,Max became the TRADOC Commander. At asubsequent meeting in his office at TRADOC



Headquarters, we discussed the ramificationsof his conjecture. It seemed clear that withoutthe influence of both superpowers, the worldwould become much more unstable. He sug-gested that there would be major changes inthe military forces of the US and NATO nations,and that I should ‘‘set up a hot plant to conductextensive analyses over the next five to tenyears.’’ Sure enough, the Conventional Forcesin Europe (CFE) talks and negotiations startedin 1989, which eventually led to the treaty. Theevents in 1989–1995 proved how omniscientMax was.

Bob Sheldon: Were you involved in the CFEactivities?

Seth Bonder: I was not in the line activityof structuring the negotiation packages, but Istarted doing CFE related studies in late 1989in support of it. In 1989, General Thurman leftTRADOC to become Commander of US South-ern Command (SOUTHCOM) where he ledthe Panama invasion (‘‘Operation Just Cause’’)and captured Mr. Noriega in 1990. General JohnFoss replaced Max as the TRADOC Com-mander and asked me to do a risk analysis ofthe potential CFE treaty. I was pleased thatGeneral Glenn Otis, who had just retired asthe CINCUSAREUR, joined VRI as an Associateand agreed to work on the CFE study with me.

The intent of the CFE talks was to asymmet-rically reduce forces in both the Warsaw Pactand NATO (accordingly, oftentimes referred toas the Asymmetric Force Reductions talks) toa position of parity of both forces. This was agigantic step for the Soviet Union. I may beslightly off from the exact numbers, but theWarsaw Pact had something like 56,000 tanksand NATO had 20,000 tanks –reductions wouldbring both alliances to the same level some-where below 20,000! There were to be similarreductions in artillery, aircraft, and other weaponsystems. The eventual treaty involved a largenumber of details regarding which nationswould take which reductions, verification ofreductions, etc.

Bob Sheldon: Why would the Soviets agreeto such a large disparity in reductions?

Seth Bonder: I don’t know, and I would besurprised if many people knew at the time. Thelogic provided to the public was that the Sovietscould not afford to keep up with President

Reagan’s large expenditures to modernizethe US military forces (his ‘‘offset strategy’’).I’m sure there must be a number of books writtenon that topic since then. Anyway, this quick-response study was completed in late 1989 orearly 1990. It concluded that there was no ‘‘oper-ational minimum,’’ and that there could be sig-nificant risks with the treaty. Let me brieflycomment on each of these.

Although there was agreement that thereshould be parity of forces, it wasn’t clear whatthat level should be. Studies by RAND sug-gested that there was a floor that the reductionsshould not go below. The VRI study indicatedthat there was no such floor—it would not bedetrimental, from a warfighting perspective, togo further down the parity line than the level be-ing discussed by the negotiators. After lots ofdiscussions and comparing analyses, the RANDanalysts sort of agreed that there might not bean operational minimum.

Based on the VRI analysis, I concluded thatthere could be significant warfighting riskswhen the forces reduced to the agreed-uponparity levels because the basic warfighting phys-ics appeared to change when the major forceimbalance was removed. In contrast to the ColdWar period when feasible changes in warfight-ing components (such as new systems, addi-tional forces, and reinforcement capability) couldnot alter the warfighting imbalance, this studysuggested that relatively minor advantages orsmall changes in many of these dimensions byeither side could significantly affect warfight-ing capability when the alliances were at thosesuggested parity force levels. Modernization hadan especially large impact on warfighting capa-bility. Accordingly, I thought that slight modern-ization improvements by the Soviets, additionalunilateral reductions by some NATO nations,slight delays in expected warning times, andother small changes in any of these dimensionsafter the treaty was implemented posed significantwarfighting risks for the NATO alliance.

When the study was completed, John Fossasked me to brief the results to General JackGalvin who was the SACEUR (the military com-mander for all of NATO, and also the Commanderof the US European Command –EUCOM). Al-though I briefed him at SHAPE (Supreme Head-quarters Allied Powers Europe) Headquarters



in Mons, Belgium, I only briefed his US com-manders because the study was classified Se-cret, No Foreign.

General Galvin’s SHAPE HQ was in Monsand his EUCOM HQ was in Stuttgart, Germany.NATO HQ is in Brussels. At the end of my brief-ing in the morning, Jack thought the resultswere so important that I should brief it to allthe NATO commanders in the afternoon. (Imay have this number wrong, but believe hehad 60 four-star generals reporting to him.) Ireminded him that the study was classified NoForeign. He got on the phone, quickly removedthat impediment, and I briefed the NATO com-manders that afternoon. Although he liked thestudy, General Galvin was upset and wantedto know why a study about his theater of oper-ations was being done by someone else. Before Icould attempt an answer, he said: ‘‘If you dostudies about my theater, I want you to do themfor me.’’ And we did a few studies for him overthe next three years. One of them examined theimpacts of nonlinear operational concepts anddeep-strike precision munitions on warfightingcapability after CFE. Another was to help himdesign NATO’s Rapid Reaction Force in 1991which I will comment on a little later in thisinterview.

As you know, the CFE treaty led to thebreakup of the Warsaw Pact alliance, reunifica-tion of Germany, collapse of Soviet commu-nism, and dissolution of the Soviet Union inthe early 1990s. This further reduced the mas-sive threat to NATO and led to unilateral reduc-tions by our NATO allies and major cuts in USdefense budgets to achieve the ‘‘peace divi-dend,’’ which of course led to major reduc-tions in our military forces.

Before coming to TRADOC, General Fosswas the CG of the 18th Airborne Corps whoseforces would fight in Southwest Asia (SWA—Iraq, Iran, etc.) if a conflict required US forcesin that region. Apparently, John was alreadyconcerned about the region’s instability becausein late July 1990 he asked me to assess US warfighting capability in the region using one ofTRADOC’s Southwest Asia scenarios (calledSWASIA). In early August, Saddam Husseininvaded Kuwait—John must have had a crystalball! Operation Desert Shield started thereaf-ter and the buildup to Operation Desert Storm

began. The air war started in January 1991,and the ground war about a month later. Wecompleted our study by February 1991, in-cluding assessment of the scenario’s versionof the left-hook strategy.

Bob Sheldon: Sounds like General Thurmanhad it right.

Seth Bonder: Absolutely. He was right aboutthe Soviet threat to NATO going away; aboutthe world becoming more unstable with DesertStorm, Bosnia, Kosovo, some others, and thecurrent Iraq conflict as evidence; and the needfor major changes in the military forces of theUS and NATO. He was also right about VRI per-forming extensive analyses to assist militaryleadership with some of these changes. I’ve al-ready mentioned the CFE and Desert Stormones and will get to others shortly after I com-ment on what the global security environmentnow looked like after collapse of the SovietUnion and Desert Storm.

Desert Storm gave us a view into future po-tential conflicts: They would be ad hoc coali-tion conflicts, contingency operations requiringpower projection, and opponents could havemodern technologies for use on the battlefieldbecause they were becoming available to allnations. In a few short years, the Cold War’stwo-superpower era transformed into a newmultipolar world that was more disorderly, moreunstable, and more uncertain, with increasedlikelihood of multiple Third World conflicts.And the uncertainty was large. We no longerwere going to Europe to fight the Warsaw Pact.We didn’t know what type of contingencyoperations we were going to be committed to(later referred to as major theater of war [MTW],small scale contingency [SSC], stability andsupport operations [SASO]); we didn’t knowwhere geographically we might have to oper-ate; didn’t know who our coalition partnersmight be; and we didn’t know who the threatmight be, how he would be equipped, and howhe would fight.

The US was the sole remaining superpowerin this environment. We created a new NationalMilitary Strategy which emphasized regionalconflicts and crisis response. The National Mil-itary Strategy success criterion was to apply‘‘decisive’’ force to win swiftly and minimizecasualties. The strategy was to draw down



overseas forces and rely on power projectionfor contingency operations. As Max had forecast,US military capability had to be restructuredto achieve the peace dividend by consideringchanges in force structure and design, modern-ization, forward stationing, strategic lift, mobi-lization, active-reserve force mix, and otherdimensions of our military capability.

I believe VRI and other OR analysis orga-nizations such as Center for Army Analysis(CAA) formerly Concepts Analysis Agency, andTRAC, played a significant role in this restruc-turing of the Army and the other military services.

I think it would be appropriate at this pointto discuss the 1989–2000 period of my career.But first, I have to go back a bit to 1987 and setthe stage by explaining my role with VectorAnalysis.

I ran a lot of the projects. I would directthree or four at a time, and people would beimplementing and creating results, but I alwayswanted to be involved no matter how big thecompany got because I liked doing analysis.So I would direct two or three, and I may evenhave done one with two or three people; I hadsome superb people like Mike Farrell (as distin-guished from Bob Farrell, whom I mentionedearlier) and George Miller who could imple-ment. We would brainstorm together; then theywould go run some models and prepare someresults. When I say ‘‘I,’’ I really mean I and myteams at Vector. But, for lots of different reasons,during the period of 1989–2000, we grew be-yond the operational level of analysis to whatI think is strategic policy-level analysis. We cre-ated a new OR analysis paradigm that had notexisted prior to 1989.

VRI expanded significantly in size. AndI had a lifestyle change at the same time. Therewas a new military strategy that came out at thattime. I think the buzzwords were �Win Deci-sively�; that is, you have a decisive force towin swiftly and minimize casualties. That wasthe new criterion. During the Cold War, we didstudies and, if we lost 100,000 people, this wasexpected. Then, in Desert Storm, we lost around100. This led to the opinion that from now onor in any war, we cannot lose more than 100.But now and at the end of the Cold War, theentire US military force had to be restructuredbecause we needed a 30–35% cut in forces in

order to achieve that peace dividend. We knewthere were going to be changes in force struc-ture and force design, there were going to bebig changes in modernization, there were goingto be big changes in strategic lift and strategicmobility because we went from prepositionedforces in Europe to a force in CONUS (Conti-nental United States) that we had to take wher-ever we wished to go fight. The whole conceptof how we were going to fight wars changed ina period of three or four years. That led to lotsof changes in how we did analysis. VRI andmany other organizations played a big role inthis restructuring of the Army, and I assumethe Air Force and the Navy, although most ofmy work was for the Army. It was done in jointanalysis, but I was not restructuring the AirForce. I was restructuring the Army.

Analysis became very difficult to performfor a number of reasons. First of all, budgetsgot tight. Not only the budget for analysis, butthe budget for the things you wished to buy.We had to think about power projection. Wenever thought about that before; the analysiswas all done, we’re in Europe, and we’re start-ing a war. Now the issue was that we had tomobilize back in the US, then we had to embarkto go some place, then we had to debark, andthen go fight a war. We were then faced withthe mobilization problem, the deployment prob-lem, and then the theater employment problem.We had to worry about them simultaneously,because they traded off between each other.Let me give you an example. You can have aforce you deploy very fast because it is light,but when it gets there, it cannot do anything.On the other hand, you can take a long timegetting a force there, and when it gets there, itwill be useful. How to make those trades andalso trades back into the mobilization processof railroads and materiel getting to our ports,what ports do they go to, what airfields do theygo to, what do you fly over first, what do yousend by ship—all of those topics became anal-ysis issues from 1989 to 1996, and I think re-main even now.

So, we had the power projection problem,and we had a new National Military Strategy.You had to have decisive force, so you couldwin swiftly and lose almost nobody. The deci-sion making process in the Pentagon became



very volatile. I do not mean they were scream-ing at each other, but things changed rapidly.The Services filling the various roles werechanging. Who was doing deep strike? Wasit Army or the Air Force? We weren’t quite sure.Funding changed. You may have had fundingfor a study, which had funding for the systemsyou’re looking at, and the next morning thatwould change. The assumptions on fundingwould change; the administrative policies ofhow you did studies would change.

The things I refer to as uncontrollable, I sumall up in what I call operations situations, in-cluding who the threat is, where the threat is,what does the environment look like, what doesthe threat have, how are they going to fight, allthe things you cannot control in an analysis.During the Cold War, we knew all of those.Post-Cold War, you knew none of it. It camedown to: you can’t just pick a spot to fight andtherefore, have these systems. I had to worryabout where in the world I was going, who Iwas going to fight, and who my coalition part-ners were in order to arrive at some reasonablealternative systems or force structures or forcedesigns or strategic lift or all the other issues.Analysis thus became very different. I knowa number of studies during the Cold War thattook two years to do, would involve 20 to 30analysts, and, after two years, nothing muchcame of it because you just kept doing the samestudies over and over because you knew whatwas in Europe. Now the paradigm for analysischanged, and I made a big change at Vector. Ipresented some papers on it, and I think otherpeople did similarly. We moved to what wasthen called ‘‘quick reaction analysis.’’ I thinkthe longest one we ever did was five months.On average, they were two or three months.Some of them were two weeks. Some of themwere days.

Mike Garrambone: What was the pressure forthose older analysis types? They used to takemuch longer.

Seth Bonder: There was no pressure forthose.

Mike Garrambone: Now all of a sudden youhave one-third the time or less for the ‘‘quickreaction’’ types?

Seth Bonder: Much less, because theyneeded answers. They were making decisions

on restructuring on all of the types of problemsI mentioned earlier: the force structure, forcedesign, new modernization, and strategic lift.They wanted answers for all of these strategicpolicy-level issues tomorrow. Let me give youan example. In fact, it led to the concept that Icall continuous analysis.

Analysis used to be an event. Somebodyhad a problem, they came to you, you did youranalysis and you gave them the answer and youwent away. I convinced General Carl Vuonowhen he was Chief of Staff of the Army, by say-ing, ‘‘Look, we have a million issues to look at.Why don’t we just start doing analysis continu-ously, and when you have questions, we’ll try toaddress them.’’ It had a high value because ifyou did continuous analysis, when our intui-tion got better, we would work on problemswe thought were important to them. We dis-cussed with them which were important, andwe could respond quickly with analysis wehad already done to issues that got raised later.Let me give you the prime example. We weredoing studies on roles and missions for theArmy.

Mike Garrambone: That’s a volatile issue.Seth Bonder: Oh, it was very volatile.Mike Garrambone: I think that has been

kicking around since 1947.Seth Bonder: Yes, but after the demise of the

Soviet Union and Warsaw Pact it raised itshead again because we started to get into deepstrike systems, who’s really going to do deepstrike, and who’s in charge of what.

The point I’m making is the concept of con-tinuous analysis gets you smarter and makesyou more responsive to the decision makers,and I thought that was very useful. We did alot of that.

Quick response analysis was done by fouror five analysts—not 20, and may have in-cluded the computer programmer who madechanges for us in the models. We created somenew metrics and used some old ones. We usedto call it the warfighting-casualty curve. On thex-axis was the FER, the Force Exchange Ratio,and on the y-axis were casualties. It was bytheater. We would run hundreds and hundredsof simulation runs and plot these curves to tryto get a handle on what the word decisivemeant. We used the FER against casualties.



Dr. Bob Helmbold at CAA used another vari-able, and he said ours was wrong. Our variableswere inverted, so his curves looked differentfrom our curves. But once we got on the samewavelength, the curves looked very similar.

Winning was defined in a certain way inthe historical data. We used what Bob did, anddid our own analysis with it, and created whatpeople used to call the horseshoe curve or theheel. It allowed us to take a look at how to define‘‘decisive’’ to meet the National Military Strat-egy. I’ll get to that more in a little bit. We createdsomething called the window of risk. Becausewe were a deployment force now, we were in-terested in how fast you had to get there, andthe window of risk. It was defined as the periodof time from C-Day (the day a deployment op-eration begins) until you arrive. You had to getthere within that window or you could not win;the threat would win. Take over the ports orwhatever you needed. The US couldn’t get therefast enough to do the job.

We also changed the concept of analysis. Weused to make maybe 15 simulation runs dur-ing the Cold War, that is, 15 runs for everycase; so we’d make thousands of runs becausewe had extensive sensitivity analysis. Well,suppose you didn’t have this system but hada different system. Suppose this force was a littlebigger than we assumed it was, there were somany uncertainties, we had to do sensitivityanalysis around all the uncertainties.

Bob Sheldon: How did you account for falsealarm rate, i.e., mobilize when you really shouldn’thave?

Seth Bonder: We couldn’t care less (Laughs).That is, if we mobilize and are ready to gosomewhere – to start early. We did some studiesfor General Galvin on a Rapid Reaction Force,where in fact, we said, you have to start beforeC-Day (Laughs). You have to have started backthen if you’re going to be able to do anythinguseful tomorrow.

Mike Garrambone: Who made that call?Seth Bonder: The military did. A lot of our

briefings said the following: ‘‘If you believe,A, B, C and D, then, do this.’’ If a US decisionmaker doesn’t think that’s a good assumption,we have to go back to the drawing board. Be-cause this depended on so many underlyinguncertainties, we tried to look at and see the

impact of assumptions and point out the bigones. In our Rapid Reaction Force Study in1992, we recommended that for him to go toa certain place, he had to start way earlier thanhe thought and get his boats moving (Laughs)to where they needed to be. If they got thereand nothing happened, then fine, turn aroundand come home. That’s okay. Because the down-side risk of delaying was the real disaster.

In Operation Desert Storm, if Saddam hadhad half-a-brain (Laughs) he’d have kept righton coming down over the ports. We couldn’tstop him; we couldn’t get there soon enough.He waited until we built up this gigantic forceand then decided to come. I have no idea why,but our analysis suggested he could have donebetter; too bad for him he didn’t have a goodanalysis team.

Let’s continue with this new paradigm. Wehad all these extensive sensitivity analyses todo, and in a previous phase of this discussionwe’re having, I talked about all the VECTORSeries, campaign models that we used. Butyou couldn’t make 3,000 sensitivity runs ofVECTOR 3 or 2, so we learned to build ana-lytical models that simulated the VECTORmodels. We called these macro models. We’drun the VECTOR Series, do some fitting, runthe macro some more, come back and test it outin VECTOR. So, as part of the new paradigm,we had models that simulated the models.

Bob Sheldon: Were those macros statisticalregressions?

Seth Bonder: Yes, they were regressions. Wegot a lot of data, not only out of the VECTORSeries here, but out of SHAPE Technical Centre;we did studies for them. They did a bunch ofArmor/Anti-armor studies when they broughtour VECTOR 2 model on board in Europe. Weused a lot of that data to create a macro, to doa lot of their studies, especially the sensitivityanalyses. We also changed some of the admin-istrative aspects. We had all of these results,and the question was, ‘‘How do you presentthem?’’ During the Cold War, briefings wentas follows: ‘‘Good afternoon. Here’s the ques-tion we addressed. Let me tell you how weaddressed it.’’ We spent 45 minutes describingall of the models and all of the assumptions,everything and then, in the last 15 minutes, ifthere was any time left, we’d explain the results.



Now, we turned that whole paradigm back-wards. When we talked to the senior guys, thedecision makers, which were the four-stars orthe Secretaries, we would go in and say, ‘‘Here’sthe issue, here’s what we did—not how, andhere’s the results.’’

Now the question is, ‘‘How do you presentthose results when you have thousands of datapoints?’’ We figured out how to use color torepresent content, not aesthetics, not to lookpretty, but the color represented content. Orig-inally, we used stop light charts, which relatedto some criteria, but we got even better thanthat. Learning how to use color, so when thedecision maker looked at it, he could get a goodfeel for the results, even though he was lookingat maybe a thousand campaign results. In a colorgraph—one graph, two graphs, and he could geta feel for where the power was and where hisweaknesses were, for the decision he was tomake. Now, how we did this. We would nor-mally meet with his staff earlier, discuss thehow, and go through that wicket with them be-fore we got to the boss. Although for manyof our senior decision maker studies, and I’mtalking three-star, four-star, Secretariat, it wasclear to me that we could not do a study forthem unless they were involved; I mean reallyinvolved.

When, for example, General Galvin said,‘‘I want you to do a study on the Rapid ReactionForce,’’ I said, ‘‘You have to be part of it.’’ Hesaid, ‘‘What do I have to do?’’ I said, ‘‘Just meetat the beginning with me, so we can make surewe have the right question. Meet about half-way through, for an hour, and I’ll show youwhat we’re doing and if it makes sense to you;and then I’ll give you a sense of what kind of re-sults you might expect to see. If you think it’snonsense, we learn right there, and then on theback end, I want to spend time really debriefingyou on what we did.’’ Some of the clients actuallygot involved in the studies. It was so importantthat you didn’t address the wrong question. Wealways made sure we had the right question be-cause at the very first meeting, I’d say, ‘‘Let mepose to you the type of output results we mayget for you. Will that satisfy you?’’ And he’dsay, ‘‘I don’t want that.’’ I said, ‘‘Well, that’s thequestion we thought we heard you ask’’ andhe’d say, ‘‘Well, let me rephrase the question.’’

That was very useful and to the extent youcan do that on studies – even if you’re doing itfor lower level staff, you then have an advocatefor the study. Of course, they’re part of thestudy. We, Vector, got involved and I ran quitea few of the studies, not all of them, but quitea few of them. If I didn’t direct them, I knewwhat was going on in them because nothing leftVector for high level review that I was not in-volved in.

For force levels to meet the National Mili-tary Strategy, we started to really look at howmuch is enough. That was never touched yearsago, because we never could get enough infor-mation (Laughs). When we did the Army Rolesand Missions studies, we did stationing strate-gies—where should we place forces, if weplaced any; we did the tradeoff study betweenmobilization, deployment and employment. Inthose days, we worried about Southwest Asiaand Korea. We created the concept for zero-based force design. Everybody kept saying,you can’t have the same forces you had duringthe Cold War. So we said, ‘‘Okay, we’ll designa force from the bottom up.’’ We created a meth-odology for doing that, starting with a totallyclean slate. We knew we would have to havea command and control function. We knew wewould have to have an intelligence function,and some sort of fundamentals. After that, webuilt it all up; all the warfighting parts, the logis-tic parts, intelligence parts – built up, from thezero-base. I think that study may have beendone for Mike Stone, the Secretary of the Army.

We did a really interesting study on forcestructure versus modernization for Carl Vuono.The issue he had for the overall Army was tocome down from 680,000; the question waswhat level to come down to? And how muchmoney do I put into modernization? On a pre-vious study, I’d convinced him that moderni-zation is an important player. The catchphraseI created was, ‘‘The smaller the force, the moremodern it has to be.’’ And that’s a true state-ment. It has more of an impact. So we did thestudy in which we started at 680,000 and re-duced the end strength at each introductionchanging the whole force design, reducing theend strength. As we did that, we saved moneyand put more money into modernization, andwe created one curve that had for the x-axis



end strength versus the modernization, howmuch modernization and how much endstrength. We had different budget levels. Weshowed effectiveness-investment curves, as afunction of all those data points. Each datapoint created a force, which we then fought awar with, and did this again and again for all dif-ferent combinations and created effectiveness-investment curves. But, we overlaid budgets ontop of that. If you look at the curves, you’d seewhat’s called an efficient point. It’s where thehighest effectiveness just touches the budgetlevel that you can get. Clearly, the more budget,the better you can do; you can get more outthere. We learned a lot of things such as if youmove off that efficient point, you need a lot ofmodernization to account for the small falloffin end strength. And, we learned a lot aboutthe dynamics of trading modernization for endstrength.

That one curve was the big study result; ittook us four months to get it. I called GeneralVuono and I said, ‘‘I’ll be coming into town.Can you spare time to look at this?’’ He said,‘‘How much do you want?’’ I said, ‘‘At leastan hour.’’ He said, ‘‘Come on into town and thencome fly with me.’’ So I flew with him, out tothe west coast. We had lots of time, and we satdown and went through the curves. It was verycomplicated to look at, until you got an explana-tion. Anyway, we recommended 508,000 at thetime and that’s what they eventually decided,but then they dropped to 480,000. Wonderfulstudies are the things that I love to do becausewe were really helping people, we were applied,and learning a lot about tradeoffs in planningfor forces, and then tradeoffs in operations. Wedid our analysis in both Southwest Asia andKorea; we didn’t do much in Europe anymore,and all of our analysis tended to be joint and co-alition campaigns. So we had the Air Force, theNavy offshore with deep strike, and coalitionpartners.

We did studies for a lot of senior leadership.I already mentioned some: the study of con-ventional forces in Europe was for General JohnFoss, who was head of TRADOC; we did stud-ies for JCS, the CINCs and the Chiefs of Staffand the Secretaries. We did force structure ver-sus modernization for General Vuono. We didthe Air Force CAS (Close Air Support) to the

Army for General Sullivan; that was a four-day study—it was short and easy. We did astudy for General RisCassi, who was in chargein Korea; I’m guessing the time period wasSeptember 1992 or 1993. General RisCassi toldme, ‘‘I want you to do a study so that I can pres-ent a paper, along with General Colin Powell,Chairman, JCS, telling the ROK (Republic ofKorea) Army how they should modernize forthe next ten years.’’

I said, ‘‘Okay, this is September, when doyou want it?’’ He said, ‘‘Well, the ministerials arein November.’’ (Laughs) I said, ‘‘It’s September.’’He said, ‘‘I know; we need a six-week study.’’I said, ‘‘You know, you’re over there and we’reover here.’’ Well we had all these classifiedlines for phone talk; I made two trips over therein between, meeting with him. He came backon one trip and we went over and presentedthe study to him and gave him all the chartshe then took to the ministerials.

Bob Sheldon: What assumptions did youmake on US force presence?

Seth Bonder: We then had 37,000 – whateverforces we had at the time, but now we broughtmore in. Not on D-Day, not in the beginning.We showed how ROK modernization could re-duce their dependence on the US; that was theintent. The US would basically pay for a lot ofthe modernization and give them our systems.It was interesting because there is also an anal-ysis shop in Korea. They redid the study andcame up with the same answers.

Bob Sheldon: Korean Institute for DefenseAnalyses (KIDA)?

Seth Bonder: Yes, something like that. I guessRisCassi said they should do a parallel study,and they came up with very similar results –not exactly the same, obviously. We did a studyfor General Galvin, which I will expound ona little bit because it was an important study;it led to another sort of criterion. I think it was1992, maybe 1993, when he said, ‘‘Look, the So-viet Union’s gone; we have a new NATO now.I want to have a force that I can use for crisesresponse.’’ He called it the Rapid ReactionForce; it eventually became the Rapid ReactionCorps. He said, ‘‘I want a force that is politicallyfeasible.’’

Each of the nations had to play. The US hadto commit some combat forces, because if we



didn’t have combat forces, we wouldn’t suffercasualties, and the other nations wanted us tosuffer with them. There were a lot of politicalconstraints like this on this force. So GeneralGalvin said, ‘‘Politically feasible and capableof crisis response and capable of holding insome areas until the bigger force can get there.’’The first part was a political analysis, in whichwe were trying to find out what the constraintswere that would limit what the force could looklike. He put one other constraint on us, whichwas critical. He said (I may be exaggerating)‘‘I don’t want this �fully adapt� stuff anymore;we’re done with that.’’ He said, ‘‘I want you tothink about wars anywhere that NATO willhave to go.’’ I said, ‘‘It’s a lot of places, right?’’We created the concept of a parametric threat,parametric scenarios. In fact, what we createdwas 27 different scenarios that were used. Wedid the analysis and built the force from theground up; we did a zero-base build. We startedwith zero and built up the force, to get a RapidReaction Force that was limited by the con-straints. Eventually we would add a package,run it through the political constraints—and itwas okayed. We would run it through a warfightto see if it enhanced warfighting. There werelots of alternative packages, so we ran differ-ent kinds of alternative packages through andadded the one that was a heuristic that gavethe highest return on the margins. This was notconventional; there are better algorithms, butwe were using a heuristic.

Once we got a force that was good for thisscenario, we switched in a new scenario, andwe’d build again. We built 27 different forces,and then looked for the commonality acrossthem, to see if we could arrive at a force thatwas politically feasible, and did okay in the27 scenarios. The force we eventually arrivedat was green (OK) in about 19 of them, yellow(iffy) in about 2, and it was red (meaning itwas bad) in a few others. We figured out waysto supplement that quickly with tactical airwings. We had some air force assets standingby for that part of the force, but we added sep-arate tactical air wings that could come fromItaly or Germany or elsewhere.

Jim McCarthy, an Air Force four-star, wasGalvin’s deputy at the time. Every time I’d doa study for General Galvin, McCarthy would

say to me, ‘‘You don’t have it right. You don’thave enough air power.’’ And I’d say, ‘‘Well,that’s the air power we have; that’s the data,’’and he said, ‘‘Well, go to Air Force Studies andAnalysis.’’ I’d go to Studies and Analysis andthey said, ‘‘We don’t have data on that.’’ (Laughs).‘‘Where does McCarthy think we have thatdata?’’ Not only sorties, but attrition on eachsortie, because we were representing flightsof aircraft in the models.

By just trying to come up with a way tomake the Rapid Reaction Force cover all of the27 scenarios, this highlighted the whole con-cept for a need of battlefield versatility. I sug-gested adding two or maybe four tactical airwings when I briefed General Galvin. I hadMike Farrell with me (Mike, mentioned earlier,was a VRI employee and a retired Army lieu-tenant colonel). He gave me two hours whenwe were at Ramstein Air Base in Germany, andI hadn’t finished. General Galvin said, ‘‘I’ve gotan activity. Let me make a call.’’ He called hisaide and said, ‘‘Bring my Class A uniform overand I’ll change here.’’ And we stayed for threehours and that’s when McCarthy said, ‘‘Yougot it right now.’’ Meaning, I had enough airpower in it.

That was the first real application of theversatility planning concept that I actually cre-ated in 1975. Nobody paid any attention to methen in 1975. But I brought it up again in JohnGalvin’s study, where it worked, and we wereable to ‘‘not try to get’’ the most cost effectivealternative, but one that was the most versatile.We wanted one that could cover all of thesepotential situations, and, as expected, you don’thave any probability distribution on that. I de-veloped some theories back in 1975 about thewhole notion of a state space without the prob-ability distribution; and how do you plan, giventhat circumstance. Here you can use variouscriteria like min-max and stuff of that kind.Anyway, that was the first application forGeneral Galvin in 1992.

The next really big study we did was for Gen-eral Maddox, who was then CINCUSAREUR.His job was to reduce the total Army personnelin Europe, including civilians, from 300,000 downto 65,000. I was going to a conference in Portu-gal with my wife. General Maddox called meand asked, ‘‘Can you stop on the way over?’’



We stopped and had dinner, and we did somebusiness and chatted about this study we de-cided to do for him. The study was to helphim determine how to draw the force down,so he could perform the missions that he wassure he would have in this new environmentwe lived in. He had all the special missions andhe had warfighting missions. He had noncom-batant evacuation missions; he had securitymissions; he had all sorts of different missionsto perform. He said, ‘‘I want you to bring theforce down to 65,000 and tell me how manyof these missions that I can perform.’’ I said,‘‘Okay. We’ll go home.’’

I went home and what we built—myselfand two other analysts—was eventually calledthe bean counter. We looked at all those differ-ent missions and built a system that we coulduse interactively. That is, we could say to thesystem, I want to do one of these types of mis-sions out of the potential four, or I want to dotwo of these.

Bob Sheldon: Simultaneously?Seth Bonder: Simultaneously. And the sys-

tem would print graphics that showed the effec-tiveness and the risk for each of the missions.We completed the study in Orlando, Florida.We went down for an AUSA (Association ofthe United States Army) meeting and I asked,‘‘David, are you coming over?’’ He said, ‘‘Yes’’because I was going to go there; he would come,too. I said, ‘‘Let’s put aside, at least the morn-ing, probably most of the day. We’re going tobuild you a force interactively.’’ It covered allof NATO where he had to play.

I had a guy on the computer, myself andMaddox, and I said, ‘‘Okay, start off with whatyou think you’d like to do,’’ and he said, ‘‘Well,I want to do two of these. I want to be able to doone of these,’’ and we had a list. I said, ‘‘That’sabout 182,000 troops you need, and here arethe curves.’’ He said, ‘‘Well, I can give up riskover here; I could accept more risk over here.’’We ran it, accepting more risk, and you get thetroops down lower. We interactively did whatI call a multi-objective decision problem, inter-actively with the decision maker. We looked atall these different objectives: the cost, risk andeffectiveness, and troop levels, which were thecost function. We got it down to 67,000 troops.We not only gave him that answer, but we

delivered the bean counter system to them,over there. So they used it, and continued touse it, to learn more about it. It was a wonder-ful study and one I enjoyed doing. Workingwith clients who have analytic skills is won-derful. They really become part of the team.

In my theoretical research, I learned aboutthe importance of attack speed and how speedinteracts with mass; that was all theory. But,in doing studies, I have over the years learnedabout the dynamics of warfare, the dynamicsof forces, how they interact with each other.Let me give you a couple of examples.

When General DePuy was looking for newoperational concepts to take on tactical fightsin Europe, we looked at battalion firefights andlearned something about their dynamics. Inoted that when an attacking brigade of sev-eral battalions takes on a single battalion, wehave up to a six-to-one force ratio. If we lookat the ratio of attacker losses to defender losses,the LER, we could tell if it was going to be a goodcampaign or a not-so-good campaign. By plot-ting the instantaneous LER over the course ofthe battle, the change in the LER shows that atthe very beginning of a battle, the defender hasa big advantage. That is because the defendersees the attacker, because the attacker is movingand the defender is hiding in defilade—you canhardly find him. The defender is getting a shot ortwo or three shots off before the attacker getsa shot off. The attacker is looking to find the de-fender but it’s harder for the attacker to bring tobear his mass of fires early on. The longer youstay engaged, you start to come down thatcurve. The longer the defender stays in defi-lade the more rapidly the curve falls off, sothe advantage shifts to the attacker. Becausethe attacker is firing much more at you, he’sgetting the effect of mass and speed. The at-tacker is starting to saturate the defender’s re-taliatory capability. If you look at that curve,you say to yourself, if I were a smart general,I’d want to fight with the ratio high; that is,I’d want to initially fight very short little battles.

In our discussions with Bill DePuy, he calledthis the active defense. General Meyer, the 3rdID commander, then looked at places to con-duct these early defenses by building a set ofcompany-size positions, in depth. When some-body attacks you, just take two or three shots,



then get out of there by falling back. Now youcannot do that ad infinitum but you are reallytrading ground for attrition. After a while youhave to stiffen the fighting up a bit. GeneralMeyer and I walked the terrain and he showedme where we could build those positions.So the active defense became a concept thatTRADOC would embrace, that later on wentto other approaches like AirLand Battle. Thepoint I’m making is analysts can do more thanhelp to solve the decision problem. You can ac-tually help with learning battlefield dynamics.If you ask the right questions, you will under-stand why things are occurring, rather than justnumbers. That was at the tactical level. WhenI got to the operational level and the strategiclevel, we learned more things that were ofinterest.

There are things I learned at the opera-tional and strategic level which I published ina long article in the AUSA Journal. I think itwas Gordon Sullivan or Ted Stroup who askedme to write the article. What I’d been learningfrom 1989 to 1993 were important things aboutthe new security environment. I had mentionedearlier that if we were still going to be in Europe,with the big reduction in forces, at parity, ev-erything became important. What I found wasthe conventional warfighting would become verysensitive to many different things: moderniza-tion, tactics, and lots of different factors thatcould affect warfighting. Modernization pro-vides major leverage; the smaller you are, themore modern you should be. The example Ialways use is one tank on one tank. The guywho has the better tank is going to win, partic-ularly if he gets the first round off. We learnedthat in order to minimize casualties, you reallyneeded to have a decisive overmatch. Now thequestion was, ‘‘What does decisive mean?’’When the National Military Strategy came outat the same time we were doing these studies,everybody was saying, ‘‘It’s decisive becauseit’s in the strategy.’’ By doing a bunch of analyses,we created what we called rainbow charts forparticular regions of the world.

Bob Sheldon: Describe the rainbow chart.Seth Bonder: Part of the dynamics we

learned, when the National Military Strategy de-scribed decisive overmatch was to win swiftlyand minimize casualties. We had already been

working on how to quantify that notion withBob Helmbold, and were able to create curves,that we called the casualty warfighting relation-ships. The force exchange ratio (FER) was on thex-axis and the casualties for particular regionsof the world were on the y-axis. The curve showshow many casualties you get, as a function ofthe FER.

What we did in doing a lot of studies inSouthwest Asia, was to define ‘‘decisive over-match’’ as having an FER of at least five, orgreater. That not only provided good warfight-ing capability, it minimized casualties, it in-creased significantly the win probability, and itprovided you with a degree of robustness.

Given the shape of the curves, if you de-cided to mount a force that could only registeran FER 3, even though the casualties were low,you were a lot less robust. Because if you madea mistake on what the threat was, say by 20 per-cent, it would jump you way up that curve,which is exponential. We like the definition,FER 5 or greater; it minimizes our casualties,it maximizes the probability of a win—withinsome constraints—and gives you robustness,in case you make a mistake.

The same curves can be used to describe theprobability of a win along the FER axis. To lookat alternative solutions to problems we woulddisplay other information such as Research andDevelopment dollars on the curve to show whatyou get if you spend 12 billion dollars versus 6billion dollars. We did a study for Jay Garner,when he was the Director, Force Development,to show the impact of different amounts ofmoney devoted to Research, Development andAcquisition. We were able to plot the differenceson the curves and put the alternatives right onthe curves and show him how robust it was.What you bought for more money was not onlybetter capability, but robustness of the force.

Another piece of dynamics we learned wasin looking at force projection—some of this workwas done with General Max Thurman. We foundthat early arriving forces in theater, nonlinearlyreduces the risk in warfighting, and gives youa better warfighting capability. It reduces over-all casualties, increases the probability you willwin and gives you some robustness. You need toget there with the right things earlier—includingyour intelligence and air defense.



As General Thurman said, ‘‘You have to beable to see, you have to be able to protect your-self, provide air defense, and communicate witheach other.’’ You must get these initial units in,and come in right behind them with the maneu-ver units – if you didn’t get there fast enough,then you suffer higher casualties. That’s whatled us to create the window of risk we spokeabout earlier. The window was the limited pe-riod of time that you were going to lose, withthe force that you were going to bring over,given the mobilization, the deployment capa-bility, and the warfighting capability. If youcould increase the deployment capability andmake it better, so you got there faster or yougot there with better stuff, then you could re-duce the window of risk.

Bob Sheldon: Is this assuming a major re-gional conflict (MRC), where you’re in the haltphase?

Seth Bonder: Yes, it was actually on the de-fensive part, but we would fight the defenseand then fight the counterattack. We’d fightboth of them. That was the window, and again,we learned something about dynamics.

Bob Sheldon: Somewhat in the operationalart?

Seth Bonder: Yes, the operational physics iswhat you’re learning about war. It was nicebecause I could talk to the guys who really un-derstood it, it was really the military. I’d saysomething and they’d know it because they’remilitary people, and very smart. During thatsame period of time, Mike Garrambone sug-gested I give a paper at the MORS Symposiumin Colorado Springs in 2000, and I presenteda paper on Lessons Learned from Modelingand Analysis.

I included in those lessons at the Sympo-sium several I had written in an article for theOperations Research journal, the 50th Anniver-sary Issue where I wrote about the history ofArmy OR. Later, I included some of those les-sons learned in a book on modeling and analy-ses, which I think were very useful to a lot ofpeople.

In 2000, I also got elected to the NationalAcademy of Engineering. A lot of my friendswho were in the Academy said I should havebeen there 15 years earlier, but they’re slow inwhat they do. It was quite an honor. I was very

pleased to be a member of the Academy andthey make use of you.

Bob Sheldon: The MORS Symposium you’rereferring to is when you presented in the samesession with Wayne Hughes?

Seth Bonder: Yes. ‘‘Army and OR analysis—Is the marriage over?’’ Our review showed thehistory of OR and I raised questions aboutwhere it was going from 1995 out.

Let me go back to the 1987–1988 period anddiscuss the VRI component. I’ve talked aboutsome studies, but VRI changed a lot in that timeperiod too. Prior to 1988, company growth wasnot a VRI objective for a lot of reasons. Themain one is I wanted to continue doing myown hands-on analysis. If we were going togrow, the way you really grow—big time—isby acquisitions. That meant I’d be out search-ing for companies and working to acquire com-panies, which I didn’t want to do.

In the early 1980s, Bob Farrell – Bob was myTreasurer and Executive VP and everythingelse—asked me if I wanted to grow by acquisi-tion, and I said, ‘‘No.’’ So we started givingmoney out to people around me and also savingmoney to buy companies in the future. That wasprior to 1987. In 1988–1989, Vector was about80 people. I recognized then that we had togrow, because I was losing people who wantedcareer paths. If you’re not growing very much,you don’t have career paths. So we made a de-cision to grow, but to grow internally. Militaryanalysis grew from 1989 to 1995. We also sawsignificant growth in the Health Analysis Divi-sion, and we added an IT Division eventually.

Our clients wanted to use the study resultswe had, so we created decision support sys-tems for them. Then they needed databases touse the data, the decisions part. So we learnedhow to create databases—we hired IT guys todo that. Finally they said, ‘‘Why don’t you justbuild systems for it?’’ So we got into the sys-tems building business, which we did for DoD,the Military Health System. We got some newclients. We started working for various Head-quarters, Department of Treasury, the IRS, andfor Commerce, building systems.

One of the big studies we did for Treasurywas the Y2K (the Year 2000) Case Study. Webuilt them a Command Center so they could ob-serve the progress of the 600 computer systems,



getting ready for Y2K. We had this big Com-mand Center, with graphics, so we could keeptrack of the developments.

The Secretary of the Treasury, LarrySummers, would occasionally come in and lookat the Center. We would show them which sys-tems were behind, and needed more money. Itturns out, on New Years Eve of 2000, he camethere. I wasn’t there but the guys at the Com-mand Center said he brought them cookies.(Laughter). It was very nice. We did some otherstudies with Treasury, which were IT-relatedfor the IRS and other people. So that helpedus grow some. By roughly 1995, we numbered350. By 1997 or 1998, we were over 400. I thinkour peak was about in 1999, when we numberedabout 430 people. We were grossing roughly $55million a year. It was a nice size company.

A lot of generals retired and came to workfor us. When General Thurman got out, he gotleukemia, and was very sick. He was at JohnsHopkins Hospital, survived that, and came hereafter they let him out of Walter Reed Army Med-ical Center. He came to work for me three orfour days a month. He was a wonderful guy,with great ideas, but would never put anythingon paper, (Laughs) never. He said, ‘‘You gotta getin the health business.’’ I said, ‘‘Max, Vector isin the health business. We’ve been in the healthbusiness since 1972, when I wrote that first pro-posal, when we worked for National Instituteof Health (NIH).’’ He said, ‘‘No, no, no – youhave to be in the health business,’’ meaning me,personally. I said, ‘‘I have enough to do withrunning the company and trying to get 30 per-cent direct charge.’’ That was because I alwayswanted to work on projects. Even when we werethat big, I was still doing 30 percent directcharge out—if I could—on contracts. It’s addi-tional to the marketing I was doing. I said, ‘‘Okay.’’So he introduced me to a gentleman named RussZajtchuk, who was a Major General and head ofthe Army’s Medical Research and DevelopmentCommand at Fort Detrick, Maryland. He hadworked for Max and in fact, may have extendedMax’s life; they were very good friends.

So, as my first foray into the health business,General Zajtchuk invited me to a Military HealthSystem Telemedicine Conference, a workingsession, a small group of maybe 50 peopleworked with us. I went there and I asked Russ

if I could take Peter Cherry and George Millerwith me. We observed what they were doingin the health business. The guys I broughtwith me were military analysts, not healthanalysts. The health people tended to be muchmore economics-oriented, running large infor-mation systems for the military health organiza-tions; something called DMIS, Defense MedicalInformation System.

They were talking about what services youwere getting. So I took my two military guysand the moderator was Charlie Flagle, who hadoffered me the scholarship at Johns Hopkins,way back in 1960. (Laughs). I’d seen him occasion-ally through my ORSA associations, but hadn’tseen him in quite a few years, and he was moder-ating this. It ran for two days, and I just wentaround and listened to what people were talkingabout. At the end they had a debrief, and some ofthe key panel heads got up and talked aboutwhat they were doing or what they learned.

It was then that Charlie made a mistake. Heasked me what I thought about the conference.And I said, ‘‘Well I think that the conferencewas good.’’ I said, ‘‘I’m just shocked at thestate of the art of analysis, if you call it that, inthe health area. It’s archaic. It’s in the Dark Ages.’’I went on to really tear into the Telemedicine partof the Military Health System, and Charlie said,‘‘You know, he’s right, we really are. We havenot pushed and moved to the state of the artof what could be done.’’ Peter Cherry andGeorge Miller both agreed with me. So we didthe Telemedicine thing and then moved over,more into the broader aspect of the healthcare de-livery system within the Military Health System.

As an example of telemedicine, you are a pa-tient and you’re out in Seattle, but you’re notnear where your dermatologist is. You get ona TV screen and the dermatologist looks at thething on your hand and says, ‘‘Oh, I know whatthat is’’ and he prescribes something for you,and he never sees you. It’s medicine via televi-sion. It’s more sophisticated than that now. Alot of ophthalmology can be done this way now.You don’t have to be anywhere near the ophthal-mologist. But he can look right into your eye,from afar. My early work was in the telemedicinearena in trying to understand how it could havesome value. That got me personally into thewhole issue of the healthcare delivery system,



with two aspects. One is the treatment of acutedisease, and how we treat people when you findsomething wrong with them. And, second, theprevention and management of disease. It be-came clear to me that everything they did to an-alyze what to do for the future was retrospective.

They looked at what they had been doingand to make a change, it was trial and error.They’d say, ‘‘We’re going to try this new pieceof equipment.’’ They’d put it in, they’d watchit for three years and then throw it away. Thewhole business was a trial and error. Costs wereescalating, it was inefficient. I said, ‘‘There hasto be a better way to do this.’’ I said, ‘‘We needto bring analysis into this process—an analysisprocess—not retrospective statistical analysis,because that’s nice for evaluating how well you’redoing. And maybe we need to introduce inferencea little beyond what you’re doing.’’ So we use thesemodels much more for prediction and planning,not evaluation. Instead of evaluation and infer-ence, we’re talking about planning and prediction.

All of the models that they used were statis-tical models. All the economists regress on someparameters, and they have Cobb-Douglas func-tions and various things, but it’s all statistical. Itdawned on me, and I think others will agree,that’s nice for evaluation and inference; but themodel depends on the data that you feed themodel. That data describes a current system. Ithas nothing to do with the future, and in fact,it’s irrelevant when you talk about the future.That is, your model has intrinsic value, the cur-rent system. You can’t get to the future system.

And I said, ‘‘But you’re using all the datafrom the current system. You can’t do 10-yearforecasts. We’re going to build the analog ofthe VECTOR Series models and help.’’ My guysthought I was nuts, because I didn’t know any-thing about health. But, I had read and I hadgone to conferences. Some of the words wouldrun way over my head but I said, ‘‘We’re goingto do that.’’

I got Russ Zajtchuk to fund us and we builtwhat was called the Healthcare Complex Model,because it models a large complex that delivershealthcare services. It is a model that simulatesa large enterprise. Let me give you an exampleof size. The model can be used to analyze thedelivery of healthcare in an enterprise thathas one or two major medical centers, like

Bethesda and Walter Reed, operating, with pa-tients coming in and out. Five or 10 hospitalsand maybe 10 or 15 clinics, that’s an enterprise,that’s complex. The model simulates, just likea simulation, but has a lot of analytic stuff init. The model simulates individual patients com-ing in and following them through a completeepisode of care.

Now an episode of care means, you walk in,then you get diagnosed and you have a chestpain and you need to have tests; you get anEKG, you get a thallium stress test, take blood;you assess that, you come up with a diagnosis,you get treated. And when you come out theother end—you’re not just there to visit—youcome out the other end and you are a recoveringpatient. Or, maybe, you go to surgery. Then youare a recovering surgical patient, and the simu-lation keeps track of you. The model simulatesall of that. I’ll give you an example of this. Wetested it at Madigan, which is an Army MedicalCenter in Seattle, with some small number ofhospitals and clinics. We ran that model for ayear and then compared it to the results thatthey had the year that we built the model for.

The model simulates more than a millionoutpatient services. Inside it are the brains ofthe models called protocols. If you say some-body comes in with this pain and here’s whatyou do with that person: get these tests, do this,check his ear, have him see an ophthalmologist;they’re called protocols, they’re called path-ways, sometimes guidelines, that’s what physi-cians do. It has something like 1,500 guidelinesfor different diseases and actions you couldtake. It keeps track of 60 different physiciantypes. It’s fairly sophisticated—it’s like fight-ing a war in Europe, in the VECTOR Series.We built that and we started using it in the mil-itary health system.

I believe they’ve been using it for somecounter-terrorism stuff in looking at the impactof a terrorist attack that would result in masscasualties and how the health system would re-spond. They need to see what kind of capabilitythe hospital system has to take care of patients.They keep adding to the model. So that was oneof the models we built, which was really di-rected by me. Not that I know a lot about thehealth care industry personally, but we got allthe health people to build it, and I was there,



making sure we were physics-based, to the ex-tent we could be.

The other piece we did, that I got involvedwith, was diabetes. One of my guys built a modelof the aging of diabetics. You’re a diabetic andwe’re going to age you 20 years—we’ll see whathappens to your disease, given different treat-ment protocols. So we simulated out 20 years.What was interesting is that after five years ofthe simulation, we could see that the number ofgood protocols, as opposed to the bad ones, re-duced the number of emergency visits by almost30 percent. The number of really bad physicalproblems—like going blind, and having to cutlimbs off—dropped by about 25 percent. TheArmy liked this study so much, they then fundedclinical trials for five years at Brooks Air ForceBase in San Antonio.

I haven’t tracked it since then, but afterthree years of the trials, they looked pretty muchlike the simulation. This doesn’t say that weshowed how to fix diabetes, because the realissue is to get people to follow the protocols.But we can do analysis on the protocols.

I think I stimulated getting more OR intoour Health Division, and they still have lots ofwork to do. A lot of the military guys havemoved over to the health side. George Miller isin the health business; Mike Potter, and a wholebunch of others, have moved over there andthey are very powerful people. That was thehealth stuff in Vector as the company grew. Idid get Vector into the Veterans Administration(VA) for studies and we started to expand outin the military health system but most impor-tantly, I think we got the military health systemto start thinking about their system and thatI think was very useful. Now that ends the pe-riod up to 2000, so let’s talk about 2000 to thepresent. Which is fortunate—my career is stillgoing (Laughs). As I said earlier, I call this myage of freedom. It’s freedom because I don’thave to manage anybody anymore.

Bob Sheldon: Before, you said you had amajor lifestyle change?

Seth Bonder: I didn’t quite finish there. Acouple of things happened. In 1999, I did decidethat we would change how we would grow be-cause we started to level off again after 10 years.I now decided we’re just going to have to doacquisitions and we were a $50 million dollar

company and I thought that the only way totake it to a $500 million dollar company wasto start acquiring companies. We had somemoney saved—we had not issued many divi-dends. So that’s what I decided in 1999; andwouldn’t you know it, in October of 1999, Icame down with kidney cancer.

A friend of mine is an oncologist and oneday he said, ‘‘You know, you never had Com-puterized Axial Tomography (CAT) scans, andyou haven’t had a colonoscopy.’’ I was then68 years old and didn’t feel bad. I said, ‘‘Okay.’’So I submitted to all those.

He had my secretary, Lisa Gaines, get holdof me and asked me to come in to see him.I did and he said, ‘‘You have a lesion in yourkidney.’’ For some reason lesion and tumordidn’t connect. I said, ‘‘A lesion?’’ He said, ‘‘Atumor.’’ I said, ‘‘Is it cancer?’’ He said, ‘‘Yes.’’I said, ‘‘You sure?’’ He said, ‘‘Absolutely.’’ AndI said, ‘‘What are we going to do?’’ (Laughs)‘‘We’re going to take it out.’’ I said, ‘‘The cancer?’’He said, ‘‘No, the kidney.’’ ‘‘When are we going todo that?’’ He said, ‘‘Next Wednesday. The sur-geon’s coming over now, the urologist, a guynamed Hugh Sullivan.’’ He came in to see meand he said, ‘‘We’re going to take it out nextWednesday because it’s encapsulated, meaningit’s inside the kidney; that’s good.’’ I said, ‘‘Is itpossible to postpone it for a week?’’ He said,‘‘Why?’’ I said, ‘‘Well, I’m lecturing at NPS nextMonday and Tuesday and then I want to go tothe wine country and party for a little bit.’’ (Laughs)He said, ‘‘Of course, you’re joking?’’ And I said,‘‘No, I’m serious.’’ The surgeon looked at thex-ray and he said, ‘‘We can see how fast it’s beengrowing, you have a week, but you have to beback the following Monday.’’ I said, ‘‘Okay.’’

So I went to Monterey and I came back, andthey took the kidney out. I was out of work forabout four weeks, recuperating. It was a sig-nificant surgery. Today it’d be a lot better; theywere doing laparoscopically. Vector used tohave three or four physicians in the companybesides nurses, and Jack Taylor said, ‘‘You getthe doughboy operation. They cut you and thenthey twist you open and they stick (Laughter)your head in there.’’ ‘‘That’s great Jack. Thankyou.’’ Anyway, they did that and I went backto work. A little hard walking, I got on the tread-mill early and started doing stuff.



It became clear to me that there was no wayI would have the energy to take the companyto the $500 million level. So I told folks in thecompany, ‘‘I think it’s time that I have to sellit.’’ I’ve got 31 years into it and I offered man-agement first option. I said, ‘‘If you can raiseenough money to buy it at what I think is a rea-sonable price, you guys can buy it.’’ A lot ofthem already had stock. Over the course of theyears, I ended up with 51 percent of this com-pany. I wouldn’t go lower than that. My goodfriend John Kettelle had taught me a lessonyears ago. He once had a company, Ketron. Hebrought in people, gave them stock, gave themtoo much and they fired him. So John taughtme a lesson: it’s a benevolent democracy. Any-way, I was committed to trying to sell it, and in2000 I started looking and offered managementthe option. They could raise some money—notthat I was outlandish—but I knew they wouldbe lower than other corporations would bidfor it. But they were way low (Laughs). It justseemed not appropriate. So I had a change ofpolicy that we’d go outside, rather than me justtry to grow it again. I looked around, and hadone of the local companies here in Washingtonlook for potential buyers. We had some comein on their own, when the word got around.They started with 17 buyers—eventually theynarrowed it to seven. We had a few others, fromthe ecological players here in town and endedup with, surprisingly, one in Ann Arbor. It wasthe company that used to be called the Environ-mental Research Institute of Michigan (ERIM).It was the old Willow Run Laboratories fromthe 1950s that they spun off for a dollar fromthe state and then they had this multispectralsensing company. They created two companies,a for-profit ERIM and a not-for-profit ERIM.They sold the for-profit a number of years be-fore we got together to Veridian. Then theyhad all this cash (Laughs) and there I was andthey needed projects of a research nature. Theymade an offer, we negotiated. I got the offer Iwanted and sold it in May 2001, so it’s nowbeen four and half years. That’s why I said, in2000 and maybe it was more like 2001 (Laughs),it became my age of freedom. The companychanged its name; it’s now called the AltarumInstitute. It’s a nonprofit. Some people I knoware on the Board: Jacques Gansler is on their

Board, a good friend. He’s the Chairman of theBoard now. It was a good sale. There wasn’t anystock involved because they’re a nonprofit. Weknew we were going to consummate this earlierthan May, probably six months early. We wereworking towards getting all the paperwork andall the arrangements and the incentives and bo-nuses, just to keep the good people on board. I be-came unemployed (Laughter). Not true, and theyactually wanted me to work for 18 months; thenew CEO, who came from General Motors,wanted me to stay on and advise him. When any-body asks me to advise them, I always tell them,‘‘Don’t ask for what you may not like.’’ (Laughter)And sure enough, his style of management wasdifferent from mine, and after six months, weagreed to part ways. It was fine.

They owned the company, they should runit the way they want to run it, and I becamefree. I sold VRI in 2001. I’ve continued my ex-tensive work with the Army Science Board.When I got off in 1992, Peter Cherry was onthe Board and when he got off in 1998, eitherPaul Kern or Walt Hollis asked me if I’d comeback, so I went back on the Army Science Boardin 1998. I’ve done a bunch of studies with themover the last six years. I just got off the Boardagain; it’s a six-year tour or two three-yeartours, or something like that.

We did some interesting stuff with theArmy Science Board, and I grew up with them,starting in 1972, so I really have a great alle-giance to the Board. I’ve since then become amember of the Board on Army, Science andTechnology, the BAST, which is a NationalAcademy Board; it’s chaired by George Singleyand John Miller now. Glenn Otis asked meto accept the offer to join the Board. MalcolmO’Neill is on the Board. These are retired gen-erals, and a bunch of fairly senior technologypeople are on the Board. The Air Force has ananalogous board, the Air Force Scientific Advi-sory Board and the Navy has an analogousboard, as part of the Academy. So I becamea member of the BAST.

I provide strategic advice to the Universityof Michigan Comprehensive Cancer Center.I’m on their advisory board to the Chairman;same with the Kellogg Eye Center. What I bringis not a lot of knowledge about cancer, but I’m in-terested in cancer, obviously. I bring OR thinking,



which although those of us in the business don’tthink it’s that special, it is quite different fromthe rest of the people in the world. I bring a differ-ent thought process to that. I sit on the Commu-nity College Board in town. I sit on two startupboards, and I sit on the OSU and University ofMichigan Engineering External Advisory Boards.

The reason I call it my age of freedom, I’mfree to pick and do what I want to do. I workon studies that I want to work on; I turn downthose I don’t want to. I go wherever I want togo. (Laughs) I’m free, which is very nice, and Idon’t have to manage, I don’t have to market.The only one I’m responsible to is me and myfamily. I created a number of endowments whenI made money selling Vector, not as much assome people think but enough to keep mehappy the rest of my life. I decided to givea bunch away, so I established a scholarship atOSU, to mimic the same fellowship I went onin 1960. Of course, the original has disappeared.

As you probably gather, I do a lot of probono stuff and some of it is done for the Na-tional Academy. I do studies for them. I’ve saton peer review committees for the NationalAcademy to get people into the Academy. I’mjust finishing up one study looking at opera-tional testing at the DoD. For evolutionaryacquisition, what did the testing look like?We’re just finishing a study on that. I did an-other study, just finishing, on how to bring moreengineering into healthcare delivery. Not justinto better technology, but how to make the de-livery system better. There’s a report coming outon that soon. The National Academy has threepieces: the National Academy of Engineering,engineers go here; the National Academy of Sci-ences, biologists go here; and then the Instituteof Medicine (IOM), where all the medical peoplego. They’re all commissioned studies. I got a callfrom somebody high up in IOM, who asked me ifI would join a study they’re going to start onlooking at how to restructure the mental health-care delivery system in the country.

I said, ‘‘I know nothing about mentalhealth’’ and just jokingly said, ‘‘I’m not goingto be a case study either.’’ (Laughter) They said,‘‘No, we need a systems engineer.’’ There aregoing to be some 13 psychiatrists, three or fourpsychologists, and one or two public healthpeople, who have all been working in mental

health and substance abuse for years. ‘‘But weneed somebody with a different thought process.’’Well, he was right (Laughs). The study was 15months long and the findings briefly got releasedto the press at the Jimmy Carter Center in Geor-gia. The report will be out in early December. Sothat’s done, but it was a very difficult study forme in a sense that I kept bringing OR thoughtprocesses to this and they think differently.

If I say to you, look Bob, you’re at A; and toget to B, you do this and then you get to B; andthen you do this to get to C; and once you’re atC, then you’ve got an integrated system. Theydon’t think that way at all. Just as sort of anaside, the mental healthcare delivery system isalmost separate from the rest of the healthcaredelivery system in this country, for lots of rea-sons, both in public and private sector. I waspushing for more integration, and since the re-port’s not out, I can’t tell you some of the inter-esting findings and recommendations. But thatwas a study that I just finished. We have onemore meeting to figure out what to do to getit implemented. I had recommended, as partof our report, we have an implementation strat-egy. That is, just telling somebody to do some-thing doesn’t get it done. They didn’t do that.There were 13 reviewers of the study; the Acad-emy has all those studies heavily reviewed beforethey go out. Four of them recommended –what’s your strategy for implementation? Wedon’t have it; so we have to create one, eventu-ally. I’m just starting a new one, this one’s comingout of the Air Force Studies Board, part of theAcademy, and that’s why I’m in town; the firstmeeting’s tomorrow. I’m not sure I know muchabout it, other than it has to do with aircraft,stealth, and some things. It’s going to be codeword stuff, so that’s all I know about the studyand I’ll find out more about it tomorrow. Butthat’s some of the pro bono work that I enjoy.But I picked those that I want to work on. Iprobably got asked to do three times as muchand turned most of them down, because theyjust don’t look interesting.

Interesting things are those that I can con-tribute to and when it’s done, it’s going to be use-ful to somebody. I have been, over this five-yearperiod, presenting some keynote talks in forumsand other places and presented some memoriallectureships at various schools throughout the



country. And I’ve been doing some consulting inthe defense industry and some in the healthindustry.

Let me wrap it up with an overall comment.I could talk about OR and military OR, with les-sons learned and what students ought to do andall of that. I don’t want to do that. What I reallywant to do is to have people recognize you don’thave a career without a lot of other people. Ca-reers aren’t solos—maybe if you’re a singer—but not in OR. Even though I did the initial the-oretical work on the differential models, thegrowth on those depended on a lot of people.Many academicians mentored me. I had mathe-maticians to bang ideas off of when I started get-ting into equations that I couldn’t solve, andneither could they. You get help along the way,all the time, as an academician, and you makefriends and you get a lot of mentors who helpyou in various ways from how to teach ethicsin analysis and a bunch of other stuff. I can’toverstate the importance of all of the supportthat came from guys like Bob Farrell, PeterCherry, Dave Thompson, George Miller, AllenWeintraub, and many others, on the growthof the whole differential models of combat.These are used, not only in the Battalion, butthe VECTOR Series; as they go from VECTOR 0all the way to VECTOR 3; there must be eightor nine versions, for different regions of theworld. That doesn’t happen with one person;it happens with everybody contributing. I di-rected and managed, but a lot of ideas camefrom others and we implemented as a team. Itcomes from having great clients. I have to tellyou, all of my clients understood what I wasdoing and they were willing to contribute.I’m talking about senior people like DePuy,Thurman, Otis, Maddox, Foster, and Kent.Wilbur Payne was a major influence on my ca-reer, in his sardonic ways. I value all of our ses-sions where we taught each other so manythings. All of that work contributes to havingyour career. My clients, who were not only clients,but helped me do some of the work, and evenwhen they were working for me, they’ve been

friends, my whole life. It’s been a wonderful ca-reer, full of great intellectual things, full of greatjoys, full of lots of sadness when you do thewrong things, and full of a lot of learning, whichis what I always have to do. It never stops. I hopethis history is useful to some other people whoget into this business. It’s such a great business.

Bob Sheldon: You mentioned your relation-ship with other people like Saul Gass. Is your re-lationship with him via INFORMS?

Seth Bonder: It was professional throughORSA, before INFORMS. I think it was JackBorsting and myself who started the visiting lec-tureship program, and Saul was part of that.I met Saul through ORSA. He and I presentedpapers at various conferences together. For oneof them, an OR-themed applied mathematicsconference organized by Saul for the AmericanMathematical Society (1979), I presented andwrote a paper, ‘‘Mathematical Modeling ofMilitary Conflict Situations,’’ that appearedin the proceedings Saul edited.

Yes, I’ve known Saul for a long time, but it’sall through ORSA. Comradeship. What is nicein my career is that I’ve had that comradeshipthrough ORSA. A lot of them were academics,so in part, they related to my academic side.But I had a separate group of academicianswho are part of my career and we were a sepa-rate group of comrades.

Obviously, growing Vector, with all of thepeople in Vector, who were like family, and weran it like a family, are among my comrades.The clients of Vector were among my comrades.Then you go to the Army Science Board and it’sa whole new group of people. And MORS is an-other group of people.

So I had ORSA, MORS, the Army ScienceBoard, the National Academy of Engineering,Vector and the teaching part. They’ve all beenpieces of an integrated career. I’m not surehow I did all of it, probably not well (Laughs).When I look back on my life and wonder ifI’ve contributed something, I think: ‘‘Yes, Idid.’’ And, you look back on all the friendshipswhich are everlasting, they just go on.



seth bonder interview (mors)

Documents