www.valvesoftware.com company valve handbook lowres

7/31/2019 Www.valvesoftware.com Company Valve Handbook LowRes

1/37

HANDBOOK FOR

NEW EMPLOYEES============================================================


2/37

HANDBOOK FORNEW EMPLOYEES

A earless adventurein knowing what to dowhen no ones theretelling you what to do

FIRST EDITION

2012

========================================================


3/37

Dedicated to the amilies

o all Valve employees.

Thank you or helping us make

such an incredible place.

Table o Contents

Preace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiHow to Use This Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiPart 1: Welcome to Valve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Your First DayValve Facts That MatterWelcome to Flatland

Part 2: Settling In . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Your First MonthWhat to Work On

Why do I need to pick my own projects?, But how do I decide which things towork on?, How do I nd out what projects are under way?, Short-term vs. long-

term goals, What about all the things that Im not getting done?, How doesValve decide what to work on? Can I be included the next time Valve isdeciding X?

Teams, Hours, and the OceCabals, Team leads, Structure happens, Hours, The oce

RisksWhat i I screw up?, But what i we ALL screw up?

Part 3: How Am I Doing? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

Your Peers and Your PerormancePeer reviews, Stack ranking (and compensation)

Part 4: Choose Your Own Adventure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Your First Six MonthsRoles, Advancement vs. growth, Putting more tools in your toolbox

Part 5: Valve Is Growing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Your Most Important RoleHiring, Why is hiring well so important at Valve?, How do we choosethe right people to hire?, We value T-shaped people, Were looking

or people stronger than ourselves, Hiring is undamentally the sameacross all disciplinesPart 6: Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

What Is Valve Not Good At?What Happens When All This Stu Doesnt Work?Where Will You Take Us?

Glossary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55


4/37

vii

2012 Valve Corporation. All Rights Reserved. Printed in the United States of America.

This handbook does not constitute an employment contract or binding policy and is subjectto change at any time. Either Valve or an employee can t erminate the employment relationshipat any time, with or without cause, with or without notice. Employment with Valve is at-will,and nothing in this handbook will alter that status.

First edition: March 2012

Valve CorporationBellevue, Washington USA

www.valvesotware.com

Designed by ValveTypeface: ITC New Baskerville

10 9 8 7 6 5 4 3 2 1

Preace

In 1996, we set out to make great games, but we knew back

then that we had to rst create a place that was designed

to oster that greatness. A place where incredibly talented

individuals are empowered to put their best work into the

hands o millions o people, with very little in their way.

This book is an abbreviated encapsulation o our guiding

principles. As Valve continues to grow, we hope that these

principles will serve each new person joining our ranks.

I you are new to Valve, welcome. Although the goals in

this book are important, its really your ideas, talent, and

energy that will keep Valve shining in the years ahead.

Thanks or being here. Lets make great things.


5/37

VALV E: HAN DB OOK FOR NE W EMP LOY EE S

viii

How to Use This Book

This book isnt about ringe benets or how to set up your

workstation or where to nd source code. Valve works in

ways that might seem counterintuitive at rst. This hand-

book is about the choices youre going to be making and

how to think about them. Mainly, its about how not to

reak out now that youre here.

For more nuts-and-bolts inormation, theres an ocial Valve intranet

(http://intranet). Look or stu there like how to build a Steam

depot or whether eyeglasses are covered by your Flex Spending plan.

This book is on the intranet, so you can edit it. Once youve read it,

help us make it better or other new people. Suggest new sections,

or change the existing ones. Add to the Glossary. Or i youre not

all that comortable editing it, annotate it: make comments and

suggestions. Well collectively review the changes and old them

into uture revisions.

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Welcome to Valve

1


6/37


2 3

WEL COM E T O VALV E

Your First Day

So youve gone through the interview process, youve

signed the contracts, and youre nally here at Valve.

Congratulations, and welcome.

Valve has an incredibly unique way o doing things

that will make this the greatest proessional experience

o your lie, but it can take some getting used to. This

book was written by people whove been where you are

now, and who want to make your rst ew months here

as easy as possible.

Valve Facts That Matter

Valve is sel-unded. We havent ever brought in outside

nancing. Since our earliest days this has been incredibly

important in providing reedom to shape the company

and its business practices.

Valve owns its intellectual property. This is ar rom the

norm, in our industry or at most entertainment content-

producing companies. We didnt always own it all. But

thanks to some legal wrangling with our rst publisher

ater Half-Lifeshipped, we now do. This has reed us to

make our own decisions about our products. Valve is more than a game company.We started our

existence as a pretty traditional game company. And

were still one, but with a hugely expanded ocus. Which

is great, because we get to make better games as a result,

Fig. 1-1

Fig. 1-2


7/37


4

i

1

3

and weve also been able to diversiy. Were an entertain-

ment company. A sotware company. A platorm company.

But mostly, a company ull o passionate people who love

the products we create.

Welcome to Flatland

Hierarchy is great or maintaining predictability and

repeatability. It simplies planning and makes it easier to

control a large group o people rom the top down, whichis why military organizations rely on it so heavily.

But when youre an entertainment company thats spent

the last decade going out o its way to recruit the most

intelligent, innovative, talented people on Earth, telling

them to sit at a desk and do what theyre told obliterates

99 percent o their value. We want innovators, and that

means maintaining an environment where theyll fourish.

Thats why Valve is fat. Its our shorthand way o saying

that we dont have any management, and nobody reports

to anybody else. We do have a ounder/president, but

even he isnt your manager. This company is yours to

steertoward opportunities and away rom risks. You have

the power to green-light projects. You have the power to

ship products.

A fat structure removes every organizational barrier


8/37


6

between your work and the customer enjoying that work.

Every company will tell you that the customer is boss, but

here that statement has weight. Theres no red tape stop-

ping you rom guring out or yoursel what our customers

want, and then giving it to them.

I youre thinking to yoursel, Wow, that sounds like a

lot o responsibility, youre right. And thats why hiring is

the single most important thing you will ever do at Valve

(see Hiring , on page 43). Any time you interview a potential

hire, you need to ask yoursel not only i theyre talented or

collaborative but also i theyre capable o literally running

this company, because they will be.

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Why does your desk have wheels? Think o those wheels as a symbolic

reminder that you should always be considering where you could move

yoursel to be more valuable. But also think o those wheels as literal

wheels, because thats what they are, and youll be able to actually move

your desk with them.

Youll notice people moving requently; oten whole teams will move

their desks to be closer to each other. There is no organizational

structure keeping you rom being in close proximity to the people

who youd help or be helped by most.

The act that everyone is always moving around within the companymakes people hard to nd. Thats why we have http://usercheck it

out. We know where you are based on where your machine is plugged

in, so use this site to see a map o where everyone is right now.

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Settling In

2


9/37


8 9

SETTLING IN

Your First Month

So youve decided where you put your desk. You know

where the coee machine is. Youre even pretty sure you

know what that one guys name is. Youre not reaking

out anymore. In act, youre ready to show up to work this

morning, sharpen those pencils, turn on your computer,

and then what?

This next section walks you through guring out what to

work on. Youll learn about how projects work, how cabalswork, and how products get out the door at Valve.

What to Work On

Why do I need to pick my own projects?

Weve heard that other companies have people allocate a

percentage o their time to sel-directed projects. At Valve,

that percentage is 100.

Since Valve is fat, people dont join projects because

theyre told to. Instead, youll decide what to work on

ater asking yoursel the right questions (more on that

later). Employees vote on projects with their eet (or desk

wheels). Strong projects are ones in which people can

see demonstrated value; they sta up easily. This means

there are any number o internal recruiting eorts

constantly under way.

I youre working here, that means youre good at your

job. People are going to want you to work with them on

their projects, and theyll try hard to get you to do so. But

the decision is going to be up to you. (In act, at times

youre going to wish or the luxury o having just one

person telling you what they think you should do, rather

than hundreds.)

But how do I decide which things to work on?

Deciding what to work on can be the hardest part o your

job at Valve. This is because, as youve ound out by now,

you were not hired to ll a specic job description. You

were hired to constantly be looking around or the most

valuable work you could be doing. At the end o a project,

you may end up well outside what you thought was your

core area o expertise.

Theres no rule book or choosing a project or task at

Valve. But its useul to answer questions like these:

O all the projects currently under way, whats the

most valuable thing I can be working on?

Which project will have the highest direct impacton our customers? How much will the work I ship

benet them?

Is Valve not doing something that it should be doing?

Whats interesting? Whats rewarding? What leverages

my individual strengths the most?


10/37


10 11

SETTLING IN

How do I fnd out what projects are under way?

There are lists o stu, like current projects, but by ar

the best way to nd out is to ask people. Anyone, really.

When you do, youll nd out whats going on around the

company and your peers will also nd out about you. Lots

o people at Valve want and need to know what you care

about, what youre good at, what youre worried about,

what youve got experience with, and so on. And the way

to get the word out is to start telling people all o those

things. So, while youre getting the lay o the land by

learning about projects, youre also broadcasting your

own status to a relevant group o people.

Got an idea or how Valve could change how we internally

broadcast project/company status? Great. Do it. In the

meantime, the chair next to anyones desk is always open,

so plant yoursel in it oten.

Short-term vs. long-term goals

Because we all are responsible or prioritizing our own

work, and because we are conscientious and anxious to be

valuable, as individuals we tend to gravitate toward projectsthat have a high, measurable, and predictable return or

the company. So when theres a clear opportunity on the

table to succeed at a near-term business goal with a clear

return, we all want to take it. And, when were aced with a

problem or a threat, and its one with a clear cost, its hard

not to address it immediately.

This sounds like a good thing, and it oten is, but it has

some downsides that are worth keeping in mind. Speci-

cally, i were not careul, these traits can cause us to race

back and orth between short-term opportunities and

threats, being responsive rather than proactive.

So our lack o a traditional structure comes with an

important responsibility. Its up to all o us to spend eort

ocusing on what we think the long-term goals o the com-

pany should be.

Someone told me to (or not to) work on X. And

theyve been here a long time!

Well, the correct response to this is to keep thinking about

whether or not your colleagues are right. Broaden the

conversation. Hold on to your goals i youre convinced

theyre correct. Check your assumptions. Pull more people

in. Listen. Dont believe that anyone holds authority over

the decision youre trying to make. They dont; but they

probably have valuable experience to draw rom, or inor-mation/data that you dont have, or insight thats new.

When considering the outcome, dont believe that anyone

but you is the stakeholder. Youre it. And Valves custom-

ers are who youre serving. Do whats right or them.


11/37


12 13

SETTLING IN

There are lots o stories about how Gabe has made important decisions

by himsel, e.g., hiring the whole Portal 1 team on the spot ater only

hal o a meeting. Although there are examples, like that one, where

this kind o decision making has been successul, its not the norm or

Valve. I it were, wed be only as smart as Gabe or management types,

and theyd make our important decisions or us. Gabe is the rst to say

that he cant be right nearly oten enough or us to operate that way.

His decisions and requests are subject to just as much scrutiny and

skepticism as anyone elses. (So i he tells you to put a avorite custom

knie design into Counter-Strike, you can just say no.)

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Whatever group youre in, whether youre building Steam

servers, translating support articles, or making the ten-

thousandth hat or Team Fortress 2, this applies to you. Its

crucial that you believe it, so well repeat it a ew more

times in this book.

What about all the things that Im not getting done?

Its natural in this kind o environment to constantly eel

like youre ailing because or every one task you decide

to work on, there will be dozens that arent getting your

attention. Trust us, this is normal. Nobody expects you

to devote time to every opportunity that comes your way.Instead, we want you to learn how to choose the most

important work to do.

How does Valve decide what to work on?

The same way we make other decisions: by waiting or

someone to decide that its the right thing to do, and then

letting them recruit other people to work on it with them.

We believe in each other to make these decisions, and this

aith has proven to be well-ounded over and over again.

But rather than simply trusting each other to just be

smart, we also constantly test our own decisions. Whenever

we move into unknown territory, our ndings dey our own

predictions ar more oten than we would like to admit.

Weve ound it vitally important to, whenever possible,

not operate by using assumptions, unproven theories, or

olk wisdom.

This kind o testing takes place across our business, rom

game development to hiring, to selling games on Steam.

Luckily, Steam is a antastic platorm or business learn-

ing. It exists to be an entertainment/service platorm or

our customers, and as such it also is a conduit or constant

communication between us and them.

Accepted truisms about sales, marketing, regionality, sea-

sonality, the Internet, purchasing behavior, game design,economics, and recruiting, etc., have proven wrong surpris-

ingly oten. So we have learned that when we take nearly

any action, its best to do so in a way that we can measure,

predict outcomes, and analyze results.


12/37


14 15

SETTLING IN

Recruiting can be a dicult process to instrument and

measure. Although we have always tried to be highly ratio-

nal about how we hire people, weve ound much room

or improvement in our approach over the years. We have

made signicant strides toward bringing more predict-

ability, measurement, and analysis to recruiting. A process

that many assume must be treated only as a sot art

because it has to do with humans, personalities, language,

and nuance, actually has ample room or a healthy dose

o science. Were not turning the whole thing over to

robots just yet though(see Hiring , on page 43).

Can I be included the next time Valve is deciding X?

Yes. Theres no secret decision-making cabal. No matter

what project, youre already invited. All you have to do is

either (1) Start working on it, or (2) Start talking to all the

people who you think might be working on it already and

nd out how to best be valuable. You will be welcomed

there is no approval process or red tape involved. Quite the

oppositeits your job to insert yoursel wherever you think

you should be.

Teams, Hours, and the Oce

Cabals

Cabals are really just multidisciplinary project teams. Weve

sel-organized into these largely temporary groups since

the early days o Valve. They exist to get a product or large

eature shipped. Like any other group or eort at the

company, they orm organically. People decide to join the

group based on their own belie that the groups work is

important enough or them to work on.= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Fig. 2-1

For reerence, read the article on cabals by Ken Birdwell. It describes

where cabals came rom and what they meant to us early on:

http://tinyurl.com/ygam86p.


13/37


16 17

SETTLING IN

Team leads

Oten, someone will emerge as the lead or a project.

This persons role is not a traditional managerial one.

Most oten, theyre primarily a clearinghouse o inorma-

tion. Theyre keeping the whole project in their head at

once so that people can use them as a resource to check

decisions against. The leads serve the team, while acting

as centers or the teams.

Structure happens

Project teams oten have an internal structure that orms

temporarily to suit the groups needs. Although people at

Valve dont have xed job descriptions or limitations on

the scope o their responsibility, they can and oten do

have clarity around the denition o their job on any

given day. They, along with their peers, eectively create a

job description that ts the groups goals. That description

changes as requirements change, but the temporary struc-

ture provides a shared understanding o what to expect

rom each other. I someone moves to a dierent group or

a team shits its priorities, each person can take on a com-pletely dierent role according to the new requirements.

Valve is not averse to all organizational structureit

crops up in many orms all the time, temporarily. But

problems show up when hierarchy or codied divisions o

labor either havent been created by the groups members

or when those structures persist or long periods o time.

We believe those structures inevitably begin to serve their

own needs rather than those o Valves customers. The

hierarchy will begin to reinorce its own structure by hiring

people who t its shape, adding people to ll subordinate

support roles. Its members are also incented to engage in

rent-seeking behaviors that take advantage o the power

structure rather than ocusing on simply delivering value

to customers.

Hours

While people occasionally choose to push themselves to

work some extra hours at times when something big is

going out the door, or the most part working overtime or

extended periods indicates a undamental ailure in plan-

ning or communication. I this happens at Valve, its a sign

that something needs to be reevaluated and corrected. I

youre looking around wondering why people arent in

crunch mode, the answers pretty simple. The thing we

work hardest at is hiring good people, so we want them tostick around and have a good balance between work and

amily and the rest o the important stu in lie.

I you nd yoursel working long hours, or just generally

eel like that balance is out o whack, be sure to raise the(contd on page19)


14/37

A Timeline of Valves History

Valve is formed in Kirkland, WA,

by Gabe Newell and Mike Harrington.

1996

Formation papers are signed on the

same day as Gabes wedding.

Quake engine license is acquired

from id Software.

Production commences on the game

soon to be known asHalf-Life (HL).

Production commences on Valves

second game,Prospero.

Valve recruits and hires two game

teams, including the first international

employee from the UK.

1997

Gabe promises that ifHL becomes

the #1- selling game, the company

will take everyone on vacation.

After internal review,HL deemed

not good enough to ship.

HL team returns to the drawing boardand essentially starts over.

Prospero permanently shelved.

HFNE:96:97::01VA L 19

step 1. Unplug cords rom wallstep 2. Move your desk

step 3. Plug cords back into wall

step 4. Get back to work

Fig. 2-2 Method to move your desk

1.

3.

2.

4.

A LV E METHOD DIAG. 1


15/37

1999 2000 2001

Valve establishes a pattern

of supporting the best

mods and occasionally

acquiring them.

Half-Life: Opposing Force

is released.

Expansion pack follows

events in Black Mesa

from the viewpoint of

an invading soldier.

Team Fortress Classic

is released.

Mike Harrington amicably

dissolves his partnership

with Gabe Newell, leaving

Newell as the sole head ofValve Corporation.

Counter-Strike (CS)

is released.

Ricochetis released.

Robin Walker demonstrates

to the mod community how

a game can be created

quickly and easily with

Valves SDK.

CS soon becomes the

worlds #1 premier online

action game.

Half-Life:Deathmatch

Classic is released.

Half-Life:Blue Shift

is released.

HFNE:99:00:01::03VA L

Half-Life: Day One OEM demo is released.

Following a certain Black Mesa Incident,

the world is never the same again.

Half-Life is released.

Released as a demo bundled with the

Voodoo Banshee graphics card, the OEM

release circulates far beyond its original

intended audience. Valve realizes the levelof anticipation for the full game.

1998

TeamFortress Software Pty. Ltd. is acquired.

Creators ofTeam Fortress (TF) join Valve and commence work

on Team Fortress Classic.

Valves first company vacation to Cabo San Lucas, Mexico.

# of employees: 30

# of children: 0

A LV E HFNE:98::02


16/37

Source engine is unveiled.

Counter-Strike: Source (CSS) is released.Years of work on Valves new Source engine

technology finally come to light.

Counter-Strike:Condition Zero

is released.

Half-Life 2 (HL2) is released.

The worlds first (legal) look at the Source engine,

along with the game it powers: HL2.

HL2 appears as the first game available both

through Steam and in retail locations.

HL2 also becomes Valves

second Xbox title.

2004

Half-Life: Source is released.

The originalHL gets a visual upgrade.

HFNE:04::05VA L

Valve outgrows its original Kirkland

office space and moves to down-

town Bellevue, WA.

2002

Steam is announced at GDC.

Valves Steam offers to third parties its

new suite of tools and services, which

it had originally built to service its own

games likeHL and CS.

Valve Anti-Cheat (VAC) is released.In a field where rampant online cheating

ruins the experience for many customers,

Valve aggressively addresses the issue.

2003

Half-Life 2 (HL2) source code

is stolen.

A thief infiltrates Valves network to

steal and disperse the code base for

the still-in-productionHL2.

Years of speculation regarding the

Borealis and Kraken Base begin

Steam is released.

Day of Defeatis released.A popular mod gets full Valve support,

becoming one of its stalwart products.

CS is released as Valves first Xbox title.

A LV E HFNE:02:03::04


17/37

Steamworks is unveiled, making the

business and technical tools of the

Steam platform available to third-

party developers free of charge.

Steam hits over 20 million users

and over 500 games.

2008

TF2 gets major class updates for Medic,

Pyro, and Heavy characters.

These updates are delivered via Steam

to all TF2 customers.

2009

Steam ships its first downloadable

content update for indie game

The Maw.

Steam Cloud is released, offering

seamless online storage of any file

types, including saved games,

configuration files, etc.

LEFT 4 DEAD 2

is released.

Presale numbers are

the biggest yet for a

Valve game.

Steam hits over 25 million users

and over 1,000 games.

TF2 releases The Sniper vs Spy Update

followed by outright WAR!

After this release, the TF2 updatesincrease rapidly: more than 280

have shipped in total.

TF2 ships its first hat.

HFNE:08:09::07 VA L

Left 4 Deadis released.2005 2006

2007

First third-party games are

released on Steam.A landmark in digital

distribution, Steamgives PC developers

an alternative to retail

for their games.

Half-Life 2:Lost Coast

tech demo is released.

Supported by the first

version of Valves popular

developer commentary.

Half-Life 2:Episode One

is released.

Valves first experiment in

episodic storytelling.

Half-Life Deathmatch:

Source is released.

Team Fortress 2 (TF2), the

long-awaited sequel to the

classic multiplayer game.

Half Life 2: Episode Two

raising the bar for emotional

storytelling.

Portalhailed worldwide as

an instant classic.

The Orange Boxis released

with two previously-released

titles and three new products:

Day of Defeat: Source

is released.

Valve hires six students

from DigiPen Institute of

Technology after seeing

their demo of the game,

Narbacular Drop.

Steam Community is released

with the first wave of features

designed to help friends

connect and socialize via

the Steam platform.

Steam reaches 15 million

active users, playing over

200 games.

A LV E HFNE:05:06:07::06


18/37

In 2012, Valve heads to the

Big Island of Hawaii for its

10th company vacation.

# of employees: 293

# of children: 185

2010 2011 2012

Valve moves to a more

expansive location in

Bellevue, WA.

Valve announces that

Steam and Source will be

available for Macintosh.

Portal 2 debuts on multiple

platforms to critical acclaim.

Valves 44th international hire

clears immigrationthis time

from Germany.

Q1: New employee handbook

rolls off press.

Valve announcesPortal 2

is launching in 2011.

Valve begins development

ofDota 2.

Dota 2 premieres at

Gamescom in Cologne,

Germany, with the

first annualDota 2

championship.

Whats next? You tell us

A LV E HFNE:10:11:12::08 19

SETTLING IN

issue with whomever you eel would help. Dina loves to orce

people to take vacations, so you can make her your rst stop.

The ofce

Sometimes things around the oce can seem a little too

good to be true. I you nd yoursel walking down the

hall one morning with a bowl o resh ruit and Stump-

town-roasted espresso, dropping o your laundry to be

washed, and heading into one o the massage rooms, dont

reak out. All these things are here or you to actually use.

And dont worry that somebodys going to judge you or

taking advantage o itrelax! And i you stop on the way

back rom your massage to play darts or work out in the

Valve gym or whatever, its not a sign that this place is going

to come crumbling down like some 1999-era dot-com start-

up. I we ever institute caviar-catered lunches, though, then

maybe somethings wrong. Denitely panic i theres caviar.


19/37


20 21

SETTLING IN

Fig. 2-3

Risks

What i I screw up?

Nobody has ever been red at Valve or making a mistake.

It wouldnt make sense or us to operate that way. Providing

the reedom to ail is an important trait o the company

we couldnt expect so much o individuals i we also penal-

ized people or errors. Even expensive mistakes, or ones

which result in a very public ailure, are genuinely looked at

as opportunities to learn. We can always repair the mistakeor make up or it.

Screwing up is a great way to nd out that your assump-

tions were wrong or that your model o the world was a

little bit o. As long as you update your model and move

orward with a better picture, youre doing it right. Look

or ways to test your belies. Never be araid to run an ex-

periment or to collect more data.

It helps to make predictions and anticipate nasty out-

comes. Ask yoursel what would I expect to see i Im

right? Ask yoursel what would I expect to see i Im

wrong? Then ask yoursel what do I see? I something

totally unexpected happens, try to gure out why.

There are still some bad ways to ail. Repeating the same

mistake over and over is one. Not listening to customers or

peers beore or ater a ailure is another. Never ignore the

evidence; particularly when it says youre wrong.


20/37

23

SETTLING IN

Fig. 2-5

But what i we ALL screw up?

So i every employee is autonomously making his or

her own decisions, how is that not chaos? How does

Valve make sure that the company is heading in the

right direction? When everyone is sharing the steering

wheel, it seems natural to ear that one o us is going

to veer Valves car o the road.

Over time, we have learned that our collective ability

to meet challenges, take advantage o opportunity, and

respond to threats is ar greater when the responsibilityor doing so is distributed as widely as possible. Namely,

to every individual at the company.

We are all stewards o our long-term relationship with

our customers. They watch us, sometimes very publicly,

Fig. 2-4 Methods to nd out whats going on

step 1. Talk to someone in a meetingstep 2. Talk to someone in the elevator

step 3. Talk to someone in the kitchen

step 4. Talk to someone in the bathroom

1.

3.

2.

4.



21/37


24

How Am I Doing?

3

make mistakes. Sometimes they get angry with us. But

because we always have their best interests at heart, theres

aith that were going to make things better, and that i

weve screwed up today, it wasnt because we were trying

to take advantage o anyone.


22/37


26 27

HOW AM I DOING?

prescriptive, and designed to be put to use by the person

youre talking about.

The eedback is then gathered, collated, anonymized,

and delivered to each reviewee. Making the eedback

anonymous denitely has pros and cons, but we think its

the best way to get the most useul inormation to each

person. Theres no reason to keep your eedback about

someone to yoursel until peer review time i youd like to

deliver it sooner. In act, its much better i you do so oten,

and outside the constraints o ocial peer reviews.

When delivering peer review eedback, its useul to keep

in mind the same categories used in stack ranking because

they concretely measure how valuable we think someone is.

Stack ranking (and compensation)

The other evaluation we do annually is to rank each other

against our peers. Unlike peer reviews, which generate

inormation or each individual, stack ranking is done in

order to gain insight into whos providing the most value at

the company and to thereby adjust each persons compen-

sation to be commensurate with his or her actual value.Valve pays people very well compared to industry norms.

Our protability per employee is higher than that o

Google or Amazon or Microsot, and we believe strongly

that the right thing to do in that case is to put a maximum

Your Peers and Your Perormance

We have two ormalized methods o evaluating each other:

peer reviews and stack ranking. Peer reviews are done in

order to give each other useul eedback on how to best

grow as individual contributors. Stack ranking is done

primarily as a method o adjusting compensation. Both

processes are driven by inormation gathered rom each

otheryour peers.

Peer reviews

We all need eedback about our perormancein order

to improve, and in order to know were not ailing. Once

a year we all give each other eedback about our work.

Outside o these ormalized peer reviews, the expectation

is that well just pull eedback rom those around us when-

ever we need to.

There is a ramework or how we give this eedback to

each other. A set o people (the set changes each time)

interviews everyone in the whole company, asking who

each person has worked with since the last round o peer

reviews and how the experience o working with each

person was. The purpose o the eedback is to provide

people with inormation that will help them grow. That

means that the best quality eedback is directive and


23/37

29

HOW AM I DOING?

amount o money back into each employees pocket. Valve

does not win i youre paid less than the value you create.

And people who work here ultimately dont win i they get

paid more than the value they create.

So Valves goal is to get your compensation to be cor-

rect. We tend to be very fexible when new employees are

joining the company, listening to their salary requirements

and doing what we can or them. Over time, compensation

gets adjusted to t an employees internal peer-driven valu-

ation. Thats what we mean by correctpaying someone

what theyre worth (as best we can tell using the opinions

o peers).

The removal o bias is o the utmost importance to Valve in

this process. We believe that our peers are the best judgeso our value as individuals. Our fat structure eliminates

some o the bias that would be present in a peer-ranking

system elsewhere. The design o our stack-ranking process

is meant to eliminate as much as possible o the remainder.

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

I you think your compensation isnt right or the work you do, then

you should raise the issue. At Valve, these conversations are surprisingly

easy and straightorward. Adjustments to compensation usually occur

within the process described here. But talking about it is always the

right thing i theres any issue. Fretting about your level o compensa-

tion without any outside inormation about how i t got set is expensive

or you and or Valve.

Fig. 3-1 Method to working without a boss

step 1. Come up with a bright ideastep 2. Tell a coworker about it

step 3. Work on it together

step 4. Ship it!

1.

3.

2.

4.



24/37


30 Fig. 3-2

Each project/product group is asked to rank its own

members. (People are not asked to rank themselves, so we

split groups into parts, and then each part ranks people

other than themselves.) The ranking itsel is based on the

ollowing our metrics:

1. Skill Level/Technical Ability

How dicult and valuable are the kinds o problems

you solve? How important/critical o a problem can you

be given? Are you uniquely capable (in the company?

industry?) o solving a certain class o problem, deliver-

ing a certain type o art asset, contributing to design,

writing, or music, etc.?

2. Productivity/Output

How much shippable (not necessarily shipped to outside

customers), valuable, nished work did you get done?

Working a lot o hours is generally not related to produc-

tivity and, ater a certain point, indicates ineciency.

It is more valuable i you are able to maintain a sensible

work/lie balance and use your time in the oce e-ciently, rather than working around the clock.


25/37


32 33

HOW AM I DOING?

By choosing these categories and basing the stack ranking

on them, the company is explicitly stating, This is what

is valuable. We think that these categories oer a broad

range o ways you can contribute value to the company.

Once the intra-group ranking is done, the inormation

gets pooled to be company-wide. We wont go into that

methodology here. There is a wiki page about peer eedback

and stack ranking with some more detail on each process.

3. Group Contribution

How much do you contribute to studio process, hiring,

integrating people into the team, improving workfow,

ampliying your colleagues, or writing tools used by

others? Generally, being a group contributor means

that you are making a tradeo versus an individual

contribution. Stepping up and acting in a leadership

role can be good or your group contribution score,

but being a leader does not impart or guarantee a

higher stack rank. It is just a role that people adopt

rom time to time.

4. Product Contribution

How much do you contribute at a larger scope than your

core skill? How much o your work matters to the prod-

uct? How much did you infuence correct prioritization

o work or resource trade-os by others? Are you good

at predicting how customers are going to react to deci-

sions were making? Things like being a good playtester

or bug nder during the shipping cycle would all into

this category.


26/37

Choose YourOwn Adventure

4

step 1. Find someone to watch your catsstep 2. Board our chartered fight

step 3. Relax by the pool

step 4. Relax by the pool some more

Fig. 3-3 Method to taking the company trip

1.

3.

2.

4.



27/37


36 37

CHOOSE YOUR OWN ADVENTURE

who interact with others outside the company call them-

selves by various titles because doing so makes it easier to

get their jobs done.

Inside the company, though, we all take on the role that

suits the work in ront o us. Everyone is a designer. Every-

one can question each others work. Anyone can recruit

someone onto his or her project. Everyone has to unction

as a strategist, which really means guring out how to do

whats right or our customers. We all engage in analysis,

measurement, predictions, evaluations.

One outward expression o these ideals is the list o

credits that we put in our gamesits simply a long list o

names, sorted alphabetically. Thats it. This was intentional

when we shipped Half-Life, and were proud to continue

the tradition today.

Advancement vs. growth

Because Valve doesnt have a traditional hierarchical

structure, it can be conusing to gure out how Valve ts

into your career plans. Beore Valve, I was an assistant

technical second animation director in Hollywood. I hadplanned to be a director in ve years. How am I supposed

to keep moving orward here?

Working at Valve provides an opportunity or extremely

ecient and, in many cases, very accelerated, career

Fig. 4-1

Your First Six Months

Youve solved the nuts-and-bolts issues. Now youre moving

beyond wanting to just be productive day to dayyoure

ready to help shape your uture, and Valves. Your own

proessional development and Valves growth are both now

under your control. Here are some thoughts on steering

both toward success.

Roles

By now its obvious that roles at Valve are fuid. Tradition-

ally at Valve, nobody has an actual title. This is by design, to

remove organizational constraints. Instead we have things

we call ourselves, or convenience. In particular, people


28/37


38 39

CHOOSE YOUR OWN ADVENTURE

Most people who t well at Valve will be better-

positioned ater their time spent here than they could

have been i theyd spent their time pretty much

anywhere else.

Putting more tools in your toolbox

The most successul people at Valve are both (1) highly

skilled at a broad set o things and (2) world-class experts

within a more narrow discipline. (See T-shaped people on

page 46.)Because o the talent diversity here at Valve, its

oten easier to become stronger at things that arent your

core skill set.

Engineers: code is only the beginning

I you were hired as a sotware engineer, youre now sur-

rounded by a multidisciplinary group o experts in all kinds

o eldscreative, legal, nancial, even psychological.

Many o these people are probably sitting in the same room

as you every day, so the opportunities or learning are huge.

Take advantage o this act whenever possible: the more

you can learn about the mechanics, vocabulary, and analysiswithin other disciplines, the more valuable you become.

Non-Engineers: program or be programmed

Valves core competency is making sotware. Obviously,

growth. In particular, it provides an opportunity to broaden

ones skill set well outside o the narrow constraints that

careers can have at most other companies.

So the growth ladder is tailored to you. It operates

exactly as ast as you can manage to grow. Youre in charge

o your track, and you can elicit help with it anytime rom

those around you. FYI, we usually dont do any ormalizedemployee development (course work, mentor assign-

ment), because or senior people its mostly not eective.

We believe that high-perormance people are generally

sel-improving.

Fig. 4-2


29/37

Valve Is Growing

5


40

dierent disciplines are part o making our products, but

were still an engineering-centric company. Thats

because the core o the sotware-building process is

engineering. As in, writing code. I your expertise is

not in writing code, then every bit o energy you put

into understanding the code-writing part o making

sotware is to your (and Valves) benet. You dont

need to become an engineer, and theres nothing

that says an engineer is more valuable than you. But

broadening your awareness in a highly technical

direction is never a bad thing. Itll either increase

the quality or quantity o bits you can put into boxes,

which means aecting customers more, which means

youre valuable.


30/37


42 43

VALV E IS GR OWI NG

We do not have a growth goal. We intend to continue

hiring the best people as ast as we can, and to continue

scaling up our business as ast as we can, given our existing

sta. Fortunately, we dont have to make growth decisions

based on any external pressuresonly our own business

goals. And were always ree to temper those goals with the

long-term vision or our success as a company. Ultimately,

we win by keeping the hiring bar very high.

Hiring

Fig. 5-1

Your Most Important Role

Concepts discussed in this book sound like they might work

well at a tiny start-up, but not at a hundreds-o-people-plus-

billions-in-revenue company. The big question is: Does all

this stu scale?

Well, so ar, yes. And we believe that i were careul, it

will work better and better the larger we get. This might

seem counterintuitive, but its a direct consequence o

hiring great, accomplished, capable people. Getting thisto work right is a tricky proposition, though, and depends

highly on our continued vigilance in recruiting/hiring.

I we start adding people to the company who arent as

capable as we are at operating as high-powered, sel-

directed, senior decision makers, then lots o the stu

discussed in this book will stop working.

One thing thats changing as we grow is that were not

great at disseminating inormation to everyone anymore

(see What is Valvenotgood at?, on page 52).

On the positive side, our protability per employee is

going up, so by that measure, were certainly scaling correctly.

Our rate o hiring growth hovered between 10 and 15

percent per year, or years. In 2010, we sped up, but only to

about 20 percent per year. 2011 kept up this new pace,

largely due to a wave o hiring in Support.


31/37


44 45

VALV E IS GR OWI NG

adding a great person can create value across the whole

company. Missing out on hiring that great person is likely

the most expensive kind o mistake we can make.

Usually, its immediately obvious whether or not weve

done a great job hiring someone. However, we dont have

the usual checks and balances that come with having

managers, so occasionally it can take a while to understand

whether a new person is tting in. This is one downside o

the organic design o the companya poor hiring decision

can cause lots o damage, and can sometimes go unchecked

or too long. Ultimately, people who cause damage always

get weeded out, but the harm they do can still be signicant.

How do we choose the right people to hire?

An exhaustive how-to on hiring would be a handbook o

its own. Probably one worth writing. Itd be tough or us to

capture because we eel like were constantly learning really

important things about how we hire people. In the mean-

time, here are some questions we always ask ourselves when

evaluating candidates:

Would I want this person to be my boss? Would I learn a signicant amount rom him or her?

What i this person went to work or our competition?

Across the board, we value highly collaborative people.

That means people who are skilled in all the things that are

Hiring well is the most important thing in the universe.

Nothing else comes close. Its more important than breath-

ing. So when youre working on hiringparticipating in

an interview loop or innovating in the general area o

recruitingeverything else you could be doing is stupid

and should be ignored!

When youre new to Valve, its super valuable to start

being involved in the interview process. Ride shotgun with

people whove been doing it a long time. In some ways, our

interview process is similar to those o other companies,

but we have our own take on the process that requires

practice to learn. We wont go into all the nuts and bolts in

this bookask others or details, and start being included

in interview loops.

Why is hiring well so important at Valve?

At Valve, adding individuals to the organization can infu-

ence our success ar more than it does at other companies

either in a positive or negative direction. Since theres

no organizational compartmentalization o people here,

Bring your riends. One o the most valuable things you can do as a

new employee is tell us who else you think we should hire. Assuming

that you agree with us that Valve is the best place to work on Earth,

then tell us about who the best people are on Earth, so we can bring

them here. I you dont agree yet, then wait six months and ask

yoursel this question again.

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =


32/37


46 47

VALV E IS GR OWI NG

Were looking or people stronger than ourselves.

When unchecked, people have a tendency to hire others

who are lower-powered than themselves. The questions

listed above are designed to help ensure that we dont

start hiring people who are useul but not as powerul

as we are. We should hire people more capable than

ourselves, not less.

In some ways, hiring lower-powered people is a naturalresponse to having so much work to get done. In these

conditions, hiring someone who is at least capable seems

(in the short term) to be smarter than not hiring anyone at

all. But thats actually a huge mistake. We can always bring

Fig. 5-2

integral to high-bandwidth collaborationpeople who can

deconstruct problems on the fy, and talk to others as they

do so, simultaneously being inventive, iterative, creative,

talkative, and reactive. These things actually matter ar more

than deep domain-specic knowledge or highly developed

skills in narrow areas. This is why well oten pass on candi-

dates who, narrowly dened, are the best at their chosen

discipline.

O course its not quite enough to say that a candidate

should collaborate wellwe also reer to the same our

metrics that we rely on when evaluating each other to evalu-

ate potential employees (See Stack ranking, on page 27).

We value T-shaped people.

That is, people who are both generalists (highly skilled at

a broad set o valuable thingsthe top o the T) and also

experts (among the best in their eld within a narrow disci-

plinethe vertical leg o the T).

This recipe is important or success at Valve. We oten

have to pass on people who are very strong generalists with-

out expertise, or vice versa. An expert who is too narrow hasdiculty collaborating. A generalist who doesnt go deep

enough in a single area ends up on the margins, not really

contributing as an individual.


33/37

48 49

VALV E IS GR OWI NG

Q: I all this stu has worked well or us, why doesnt every company

work this way?

A: Well, its really hard. Mainly because, rom day one, it requires a

commitment to hiring in a way thats very dierent rom the way most

companies hire. It also requires the discipline to make the design o

the company more important than any one short-term business goal.

And it requires a great deal o reedom rom outside pressurebeing

sel-unded was key. And having a ounder who was condent enough

to build this kind o place is rare, indeed.

Another reason that its hard to run a company this way is that it

requires vigilance. Its a one-way trip i the core values change, and

maintaining them requires the ull commitment o everyone

especially those whove been here the longest. For senior people

at most companies, accumulating more power and/or money over

time happens by adopting a more hierarchical culture.

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

on temporary/contract help to get us through tough spots,

but we should never lower the hiring bar. The other reason

people start to hire downhill is a political one. At most

organizations, its benecial to have an army o people

doing your bidding. At Valve, though, its not. Youd

damage the company and saddle yoursel with a broken

organization. Good times!

Hiring is undamentally the same across all disciplines.

There are not dierent sets o rules or criteria or engi-

neers, artists, animators, and accountants. Some details are

dierentlike, artists and writers show us some o their

work beore coming in or an interview. But the actual

interview process is undamentally the same no matter who

were talking to.

With the bar this high, would I be hired today? Thats

a good question. The answer might be no, but thats actu-

ally awesome or us, and we should all celebrate i its true

because it means were growing correctly. As long as youre

continuing to be valuable and having un, its a moot

point, really.



34/37

Epilogue

6


35/37


52 53

EPILOGUE

What Happens When All This StuDoesnt Work?

Sometimes, the philosophy and methods outlined in this

book dont match perectly with how things are going day

to day. But were condent that even when problems persist

or a while, Valve roots them out.

As you see it, are there areas o the company in which

the ideals in this book are realized more ully than others?

What should we do about that? Are those dierences a

good thing? What would you change? This handbook

describes the goals we believe in. I you nd yoursel in

a group or project that you eel isnt meeting these goals,

be an agent o change. Help bring the group around.

Talk about these goals with the team and/or others.

What Is ValveNotGood At?

The design o the company has some downsides. We usu-

ally think theyre worth the cost, but its worth noting that

there are a number o things we wish we were better at:

Helping new people nd their way. We wrote this

book to help, but as we said above, a book can only

go so ar.

Mentoring people. Not just helping new people gure

things out, but proactively helping people to growin areas where they need help is something were

organizationally not great at. Peer reviews help, but

they can only go so ar.

Disseminating inormation internally.

Finding and hiring people in completely new

disciplines (e.g., economists! industrial designers!).

Making predictions longer than a ew months out.

We miss out on hiring talented people who preer to

work within a more traditional structure. Again, this

comes with the territory and isnt something we should

change, but its worth recognizing as a sel-imposed

limitation.


36/37


54 55

Glossary

Jargon. Lingo. Code words.

14-Year-Old BoyI you see one running your project, dont worry. Thatsactually 57-year-old Josh Weier (see Josh Weier). I you have any extra stemcells, give them to him! He bathes in them daily.

AustraliaA place thats either very near or is New Zealand where morethan hal o Valves employees were born.

City o SeattleWhere Valves ounders promised wed locate our ocebeore pulling a massive bait and switch to the Eastside (see also Greg Coomer).

Coee Machine, Right-hand DispenserThe dispenser in all coeemachines at Valve that holds the decaeinated coee beans. To the best oour knowledge, these have never needed to be relled. For all we know, thebeans are decorative plastic.

Company VacationEvery year, the company gathers all the employees andour amilies, fies us somewhere tropical, and gives us a ree weeklongvacation. Popular pastimes include beard contests, snorkeling, ice creamsocials, jet skiing, or just sitting on the beach chatting with the locals abouthow many googly-eyed seashells you should buy rom them. (Your eeling:none. Their counteroer: Just buy ve then.)

Empty Shel on Fith FloorPlace were planning on putting all thoseawards or Ricochetonce the gaming world nally catches up with it.

FishbowlThe conerence room by the lunchroom. The one with a bigglass wall. Dont let the name throw youwe dont actually use it as ashbowl! Except, o course, on Fishbowl Fridays, where we ll it up with tenthousand gallons o putrid saltwater so that all the manta rays and sharks

will have something to breathe while they ght to the death. You wont seeit in your list o benets, not because it isnt un, but because it is illegal.

Freight Elevator(See Method to move your desk, on page 18.)

Gabe NewellO all the people at this company who arent your boss,Gabe is the MOST not your boss, i you get what were saying.

Where Will You Take Us?

Valve will be a dierent company a ew years rom now

because you are going to change it or the better. We cant

wait to see where you take us. The products, eatures, and

experiences that you decide to create or customers are

the things that will dene us.

Whether its a new game, a eature in Steam, a way to

save customers money, a painting that teaches us whats

beautiul, something that protects us rom legal threats,a new typeace, an idea or how to be healthier while we

work, a new hat-making tool or TF2, a spectacular ani-

mation, a new kind o test that lets us be smarter, a game

controller that can tell whether youre scared or a toy that

makes our-year-olds laugh, or (more likely) something

nobodys thought o yetwe cant wait to see what kind

o uture you choose to build at Valve.


37/37

VALV E: HANDBOOK FOR NEW EMPLOYEES

Greg CoomerThe only person who cares or remembers that somebodyonce might have said wed move to Seattle.

KnivesThat which one can never own enough o. A vast collection othem is in no way a Freudian compensation.

ManagerThe kind o people we dont have any o. So i you see one, tellsomebody, because its probably the ghost o whoever was in this buildingbeore us. Whatever you do, dont let him give you a presentation onparadigms in spectral proactivity.

Mann Co.Maker o square, unsae products or men that occasionallycatch on re, and more occasionally, work as advertised. Owned andoperated by Saxton Hale (see Australia).

Parking Garage ElevatorsAutonomous hostage-taking devices with a willo their own. Beware.

PlaytestingWhat we do early and oten. And loudly, i Karen is the tester.

PoniesThe animals most beloved by those away rom their computers,and most despised by people who preer to hear jokes just once.

Scorpions, Poison, QueenRepeated exposure to our bathrooms Pavlov-ian rock block soundtrack will ensure that youll never be able to relieve

yoursel again unless someone hums Rock You like a Hurricane.

Shitty WizardPerson responsible or allDota 2bugs. AkaFinol.

Talk AliasMarc Laidlaws internal blog.

(Un)weighted Companion PillowThe thing Erik Wolpaw carries aroundwith him and covers his mouth with ater others have sat on it.

Valve ActivitiesYou will learn to love blacksmithing.

Josh WeierVariously pronounced Josh Weere, Josh Wire, JoshWoe-Rue, Josh wuhh[trailing o], and Josh Joshington by those o uswho stopped caring. Theyre all equally valid!

WFHWorking From Home. What to do i a single snowfake alls out othe sky.

www.valvesoftware.com company valve handbook lowres

Documents