spe distinguished lecturer program · expert reliability & uncertainty steve begg, asp,...
TRANSCRIPT
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
1SPE DL 2010
SPE Distinguished Lecturer Program
Primary funding is provided by
SThe SPE Foundation through member donations and a contribution from Offshore Europe
The Society is grateful to those companies that allow their professionals to serve as lecturers
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
professionals to serve as lecturers
Additional support provided by AIME
Society of Petroleum Engineers Distinguished Lecturer Programwww.spe.org/dl
Reliability of Expert Judgments and Uncertainty
SPE Distinguished Lecture, 2010-2011
Judgments and Uncertainty Assessments
Steve Begg
Australian School of Petroleum, University of Adelaide
Centre for Improved Business Performance
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
2SPE DL 2010
Reliability of Expert Judgments and Uncertainty
SPE Distinguished Lecture, 2010-2011
Judgments and Uncertainty Assessments
Steve Begg
Australian School of Petroleum, University of Adelaide“All business proceeds on beliefs, or j d t f b biliti d t Centre for Improved Business Performancejudgments of probabilities, and not on certainties".
Charles W. Eliot
Outline
• The Nature of Uncertainty
P l P b bilit d• People, Probability and Judgment
• Performance of Industry Experts
• Conclusions
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Conclusions
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
3SPE DL 2010
Industry Performance
• Comments & Observations– “Every one of our 10 most important projects failed to
generate the desired return ” (Super Major)generate the desired return. (Super Major)
– “The actual performance of our key assets wasn’t even within the P1 to P99 range.” (Large Independent)
– “I want your guarantee that we will not spend more than the P50 on this project!” (CEO to Manager)
– “a decade of unprofitable growth”; vast majority of
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
– a decade of unprofitable growth ; vast majority of projects take longer, cost more and produce less than predicted; 1-in-8 of major offshore are “disasters” (IPA)
Taking on a Cult of Mediocrity
• “The last 10 years might be called ‘a decade of unprofitable growth’ for many upstream companies.”p
Ed Merrow, Independent Project Analysis (IPA)
– Based on the analysis of more than 1000 E&P projects:2/3 offshore, average $1Million – $3Billion
– One in eight of all major offshore developments in the last decade falls into the ‘disaster’ category.
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
failed on two out of three metrics: >40% cost growth, >40% time slippage, produced < 50% than 1st year plan
Average CapEx for these is $670MM
– Record even worse for mega-projectsCapEx of $1 billion or more
Source: UPSTREAM, 23 May 2003
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
4SPE DL 2010
Taking on a Cult of Mediocrity
-10%
B t ti
Asset costgrowth
BE
TT
ER
-10%
Schedule slip
9 %
First Year Plan
Goal
-5%
0%
5%
10%
15%
Best practice
Industry average
-5%
0%
5%
10%
15%
Best practice
Industry average
20%
95%
75%
Best practice
Industry average
55%The bigger and more important a project
GoalGoal – Why?
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
30%
25%
20%
Disasters
WO
RS
E
40%
35%
25%
Disasters
30%
35%
Disasters
Ref: Merrow, Ed: Taking on a Cult of Mediocrity, Upstream, 23 May 2003
The bigger and more important a project gets, the more likely it ends up in the
“disaster” category.
The fundamental problem: Industry performance not living up to expectations, or possibilities
Uncertainty
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
5SPE DL 2010
The fundamental problem: Industry performance not living up to expectations, or possibilities
• People tend to grossly under-estimate uncertainty– number of uncertain factors and the magnitude of
uncertaintyu ce ta ty
– complexity of the relationships between them and therefore un-anticipated non-intuitive outcomes)
• Better decision-making requires accurate (unbiased & appropriate range) uncertainty assessment
– Reduce uncertainty only IF it can change a decision AND expected benefit of reduction is less than its cost
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
AND expected benefit of reduction is less than its cost
• Naive understanding of economics– NPV “rule” : uncertainty (and/or delay) = Value Loss
– Focus on mitigating downside risk at expense of capturing upside opportunities
Uncertainties and dependencies matter: they are everywhere in the evaluation “system” ……
Prices
Taxes
Drilling CapEx
Export
EconomicsAsset A
OOIP Model
Petro-physics
PredictedProduction
Royalties/PSCSeismic
Geology
OpExProcessing
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
. . .Production Data
ReservoirSimulation Portfolio
EconomicsAsset 1
EconomicsAsset n
Decline
ProductionAllocation
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
6SPE DL 2010
…. and we spend a lot of money without really knowing which ones matter
Gross Rock Volume
UncertaintyParameter Impact on NPV
Buy more seismic
Action
Average Porosity
Recovery Factor
Saturation
Oil Price
Facilities
Take more core
Build simulator model
y
Different rock model
Hedge with futures
Cheaper Steel Supply
"There is nothing so inefficient as very efficiently doing the wrong things".
Peter Drucker
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Rig Cost
Fiscal Terms
Net:Gross
Continuity
Renegotiate contract
Fire lawyers
More gamma logs
Survey Analogues
Flaw of Using Averages (after Savage)
NCF
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
NCF
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
7SPE DL 2010
Averages don’t always work
• For non-linear processes,– reservoir simulation– volumetrics with cut-offs
Mean = 10 Mean = 5
– development alternatives
even if only a single, “best” estimate is required, we still need to use complete range of inputs - cannot use an average input
Model
Y = X2 Z2/
3 21 1 8x z
Si l ti R lt
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• Also, P10 (P90) results are NOT given by taking P10 (P90) inputs and running them through the model
ResultY = 4
Y
Simulation Result
0 30
True Mean~ 7.8
The Flaw of Averages:
• Unless model (f) is linear (“model” is a calculation)
average of f( x y z )
average of f( x,y,z,…. )
f( average(x), average(y), average(z),….)
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
where x,y,z,… are uncertain quantities, and
f is a calculation using x, y, z ....
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
8SPE DL 2010
Two major fields of study in decision-makingunder uncertainty
• how decision makers
PrescriptiveDescriptive ≠• how decision makers
should choose
• Best decisions given information, alternatives, objectives & preferences
• principle of maximizing expected value (or utility)
• how decision makers actually choose
• psychology of judgment and decision making
• behavioral studies and observations
• heuristics & biases
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
expected value (or utility)
• traditional decision theory (decision trees, MC Simulation, VoI)
• heuristics & biases
• Sub-optimal, inconsistent illogical decisions
Psychology Mgmt. Science
Technical Work in the context of Decisions and Uncertainty
The role of a Petroleum Geoscientist or Engineer is to support decision-making
• Whether you, your group or company explicitly recognize it or not – technical work is fundamentally about uncertainty assessment for the purpose of making decisions
• If you have a “make the best possible prediction” focus, there is no stopping rule
– you can always reduce uncertainty a bit more (more data, more time, more analysis)
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
more analysis)
• A decision-driven focus gives a trivially simple stopping rule– Stop when further analysis doesn’t change the decision!!
– From a decision-making perspective we only need to find which option has the greatest value – we don’t (usually) need a precise (= little uncertainty) estimate of that value
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
9SPE DL 2010
The solution
Betterperformance
BetterDecisions
performance
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Better uncertaintyassessment
Results of survey on “reserves” uncertainty-related word meanings
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Source: McLane et al, AAPG Bull v92 #10
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
10SPE DL 2010
Probability: The Language of Uncertainty
• Classical (Theoretical)Number of outcomes representing the occurrence of an event
Total number of possible outcomes
30 d b ll d 70 b ll i b P(R d) 30%– e.g. 30 red balls and 70 green balls in a bag. P(Red) = 30%
• Relative Frequency – Proportion of times an event occurs in the long run– Can be ESTIMATED from sample data ASSUMING identical events
e.g. 15 out of 20 wells drilled were dry holes. P(Dry) = 75%– More accurate with greater sample size. May not apply to future.
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• Subjective– Personal degree of belief of the likelihood of a future event occurring
(or of the unknown outcome of a past event)
– May be based on some past similar / analogous occurrences
Uncertainty: not knowing if a statement is true or not
Throw a die and hide top face. What is the probability of a 3? 1/6
Now you getinformation.
Has has the top face changed? No
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
What is the probability of a 3 now?
Uncertainty is a function of what you know. There is no “right” uncertainty (or PDF)!
p g oHas the probability of a 3 changed? Yes!
1/3
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
11SPE DL 2010
Uncertainty is in OUR heads – it’s a function of our state of knowledge
Different people can, legitimately, hold
Its not a feature of the “system”. A consequence:
0.333 0.333
p p g ydifferent views about the uncertainty of an
unknown quantity
Person A Person B
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
1 2 3 5 64Outcome
Prob
.
1 2 3 5 64Outcome
Prob
.
Uncertainty is in YOUR head:What’s it worth to know more?
Another consequence:
Information might have value by virtue of its ability to change probabilitiesits ability to change probabilities
• Betting Game- Win $100 if it is a 3.
- Lose $10 if it is not
• How much is it worth to look at?
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
How much is it worth to look at (get information about) the centre?
Corollary: you might change your mind on whether to bet or not as a result of getting new information
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
12SPE DL 2010
Probability is subjective: games of chance – eg coin tossing
Person A Info
(A thinks there is a
Person BInfo(A thinks there is a
small chance the coin is biased
towards heads)
InfoShared
Info
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Person A PDF Person B PDF
Heads Tails Heads Tails
Probability is subjective: implications within companies or for joint ventures
Personor
Personor
Company BCompany
AInfo
Company BInfo
SharedInfo
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Company A PDFCompany B PDF
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
13SPE DL 2010
Gambling (probability = repeated outcomes) vs. “Real World” (probability = degree of belief)
Uncertainty Quantification
All Identified Some missed or unknowable
KnownDistributionType
UnknownDistributionType
1. Identify Possible Outcomes 2. Assign Probabilities to Outcomes
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
KnownParameters
UnknownParametersGames of Chance:
Classical Prob & StatsOil & Gas:Subjective
Uncertainty and Ambiguity
• What is your degree of belief (probability) that Steve is a “wine-drinker”?
• The event “wine-drinker” is ambiguous. It could mean– Prefers wine to beer– Enjoys wine– Drinks a glass of wine once per week– Drinks a glass of wine daily, etc
• The precise definition of the event/statement/outcome is critical to determining our degree of belief in it.
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
We MUST remove any ambiguity in the definition of uncertain statements (events,
outcomes) to which we wish to assign probabilities
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
14SPE DL 2010
Uncertainty vs Risk
Uncertainty Risk
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• A Risk (noun!) is one possible consequence of uncertainty. It has a negative connotation, which is “personal” to the D-M
- an event that has a negative impact on DM’s objectives
- It is specified by defining the event and assessing its probability,
Don’t take a biassed approach to managing consequences of uncertainty
• Risk is only one outcome of uncertainty - so is Opportunity!– often over-looked – it is a source of value creation
Consequences of Uncertainty
Risk Opportunity
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Possibility of loss or injury
A dangerous element or factor
The degree of probability of loss
Possibility of exceeding expectations
Upside potential
A wonderful element or factor
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
15SPE DL 2010
Uncertainty vs Variability
Variability of all sand-body widths
Width
Uncertainty in individual sand-body width??
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
y??
Width
Sand 1
?? Sand 2
Uncertainty vs Variability
Variability of all sand-body widths
A di t ib ti th t
Width
Uncertainty in individual sand-body width??
A distribution that describes the variability
of a natural phenomenon is not usually appropriate to
describe the uncertainty in
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
y??
Width
Sand 1
?? Sand 2
describe the uncertainty in a single occurrence
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
16SPE DL 2010
Outline
• The Nature of Uncertainty
P l P b bilit d• People, Probability and Judgment
• Performance of Industry Experts
• Conclusions
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Conclusions
Adelson’s square: Estimate the gray %
Using a scale of 0% (black) to 100% (white) estimate the % gray of squares A and B
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
17SPE DL 2010
*Shepard’s Table
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Perceptual Limitations as a Metaphor Cognitive Limitations
• Awareness of illusions by itself does Awareness of illusions, by itself, does not produce a more accurate perception.
• Illusions & cognitive errors therefore, can be extremely difficult to
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
can be extremely difficult to overcome.
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
18SPE DL 2010
Judging likelihoods of events
• Linda is a 31 years old, single, outspoken and very bright. She majored in philosophy. As a student she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations. p p
Which is the more likely alternative?
a) Linda is a bank teller
b) Linda is a bank teller and active in the feminist movement.
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Answer:______
*Probability Rules: Addition Rule for Non-Mutually Exclusive Events
Event A and Event B can occur at the same time
Not A or B
AB
Joint probability = P (A and B)
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• Multiplication Rule for independent events
P(A and B) = P (A) * P(B)
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
19SPE DL 2010
*Discussion of Linda Question
• Nearly 90% of respondents choose the second alternative (bank teller and active in the feminist movement), even though this is provably logically incorrect
bank tellers feminists
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
P(Bank Teller) > P(Bank Teller AND Feminist)
feminist bank tellers
Cognitive Illusions
• The description of Linda is more representative of a feminist bank teller so people, wrongly, conclude it is more likely that she is a feminist and a bank teller
• Kahneman & Tversky (1982)
– “As the amount of detail in a scenario increases, its probability can only decrease steadily, but its representativeness and hence its apparent likelihood may increase.”
“Th li t ti b li i i
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
– “The reliance on representativeness, we believe, is a primary reason for the unwarranted appeal of detailed scenarios and the illusory sense of insight that such constructions often provide.”
• Implications: consider a “rich” description of a reservoir depositional environment
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
20SPE DL 2010
Sequences - Representivity Heuristic:The “Law of Small Numbers”
• I have just tossed a fair coin 7 times. You have not seen the result
• You are invited to play a betting game to guess which of the three sequences below is the one I actually observed.
• Which sequence would you bet on
a) HHHHTTT
b) THHTHTT
• Using multiplicative rule for independent eventsP(A&B&C&D.) = P(A)*P(B)*P(C)*P(D) ….
P = (1/2)7 = 1/128
P (1/2)7 1/128
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
b) THHTHTT
c) TTTTTTTP = (1/2)7 = 1/128
P = (1/2)7 = 1/128
• ALL sequences have the SAME probability and are thus EQUALLY likely (or equally rare!)
*Intuition: Steel Band Problem
• Assume the earth is perfectly flat and a band of steel is places around the equator, which has a circumference of 50,000 km.
• Now we add in an extra 10m of steel, which slightly forces the band off the surface of the earth.
• Estimate by how much?– ………….
– The diameter of a proton, an atom …
Thi k f $ t
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
– Thickness of a $ note
– ……..
• Answer: 1.6 metres
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
21SPE DL 2010
Heuristics, Biases & Uncertainty
• Heuristics
– simple rules of thumb and mental shortcutssimple rules of thumb and mental shortcuts
• Biases
– systematic errors that can result from the use of heuristics
• Our “mental wiring” is just not good when it
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
comes to uncertainty
– Intuition and “gut feel” often significantly wrong
Heuristics, Biases & Uncertainty
• Heuristics
– simple rules of thumb and mental shortcutsHuman beings are not endowed with simple rules of thumb and mental shortcuts
• Biases
– systematic errors that can result from the use of heuristics
• Our “mental wiring” is just not good when it
Human beings are not endowed with rational probabilistic thinking and optimal
behaviour under uncertainty.
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
comes to uncertainty
– Intuition and “gut feel” often significantly wrongBias & error => poor decisions => undesirable outcomes
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
22SPE DL 2010
Evidence of Bias: Data from IPA
All 1000+projects
If the Forecasted production is the “Base Case”, we h ld h i t l projects
in the study
No projects
should have approximately as many projects producing more than expected as less than expected !!
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
0 50 100 150 200 250 300
Basis for development sanction
0% 50% 100% 150% 200% 250% 300%
Outline
• Introduction
Th N t f U t i t• The Nature of Uncertainty
• People, Probability and Judgment
• Performance of Oil & Gas Industry Experts
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Industry Experts
• Conclusions
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
23SPE DL 2010
Refresher: Interpreting PDFs & Percentiles
The area under the PDF between any two points is theprobability of X lying between those two points
0.10
80% ofarea
10% ofarea
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
P10 P90
10% ofarea
X
Assessing the ability of experts to assign P10-P90 ranges
Lower Limit(P10)
UpperLimit(P90)
80%Chance
10% 10%
Question 1
Question 2
Question 3
etc
E.g. What was daily average oil production in the USA in 2003?
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Answer: 7,454,000
g y g p
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
24SPE DL 2010
Overconfidence Results: Large Industry Sample –using “industry- related” questions
0.30
0.35
pan
ts Expected
Observed
0.10
0.15
0.20
0.25
po
rtio
n o
f P
arti
cip
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
0.00
0.05
0 1 2 3 4 5 6 7 8 9 10
Pro
p
Questions Correct /10
Exploration Production AbandonDevelopment
Perceived uncertainty
Actual uncertainty
* Experts & Over-confidence
Max
EOR
uncertainty
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Time
Min
2D3D
Discovery. drilling
Development. drilling
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
25SPE DL 2010
US DoE Price Forecasts
Trends Predicted Beginning From the Actual Price of Year ListedTrends Predicted Beginning From the Actual Price of Year Listed
120120
100100
19821982
19811981100100
8080
6060
4040
19841984
19851985
19861986 19871987
Dol
lars
per
Bar
rel
Dol
lars
per
Bar
rel
19831983
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
after U.S. Department of Energy, 1998after U.S. Department of Energy, 1998
2020
0019751975 19801980 19851985 19901990 19951995 20002000 20052005
YearYear
19911991
19951995Actual
DD
Does it Matter? Overconfidence Model
Biassed –OverconfidentOverconfident
True
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
12500 15000 17500 20000 22500
Area
OC20 True 10th/20th percentile
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
26SPE DL 2010
400
on
Economic Impact of Overconfidence
Welsh, Begg & Bratvold (2007) SPE 110765
100
200
300
Rea
l E
(NP
V),
$M
illio
NPV
EV
Assessed (Overconfident) Value
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
-100
0
0% 5% 10% 15% 20% 25% 30%Overconfidence
Probability Intuition: Assessing co-variation
Present Absent
Present 16 4Seismic
Hydro-carbons
Present 16 4
Absent 4 1
• Which cells of the table are needed to determine whether seismic anomalies are associated with hydrocarbons
Upper Upper Lower Lower
Anomaly
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Upper Upper Lower Lower
Left Right Left Right
• According to the data in the table, do seismic anomalies increase the probability of hydrocarbon presence? Yes/No
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
27SPE DL 2010
Probability Concepts: Conditional Probability in Venn Diagram
Not A or B
A
Joint probability = P (A and B) =
AB
P (A and B) P(A|B) = =
As a proportion of the whole area
The probability of A happening, given that B has occurred1/52 1
P(Q|) = =
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
P(A|B) P(B)
(Q|)1/4 13
P (A and B) P(B|A) = =
P(A)
The probability of B happening, given that A has occurred1/52 1
P(|Q) = = 1/13 4
Probability Intuition: Assessing co-variation
Present Absent Total
Present 16 4 20
Hydrocarbons
Seismic
• All 4 pieces on information (cells) are required
• Conditional probabilities of HC presence given the presence/absence of seismic anomaly
Absent 4 1 5
Total 20 5 25
Seismic Anomaly
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
P(HC Present | Anomaly present) = 16/20 = 80%
P(HC Present | Anomaly absent) = 4/5 = 80%
• Probability of HC being present is the same, whether or not a seismic anomaly is present – so no information
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
28SPE DL 2010
Assessing Co-variation (presumed association): Results
80
90
100
Correct
Incorrect
30
40
50
60
70
80
of
Par
tici
pan
ts
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
0
10
20
30
First Both
%
Correct Answer
Correct for right reason
Probability Intuition: Assessing Reliability of Predictors
• Historical estimates suggest one in every 1000 blow-out preventers has serious cracks.
• Suppose x-ray analysis is a very good but not• Suppose x-ray analysis is a very good, but not perfect, detector of these cracks.
– If a blow-out preventer has cracks, x-rays will correctlysay it has them 99% of the time
– If a blow-out preventer does not have cracks, x-rays will wrongly say that it has them 2% of the time
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• A blow-out preventer has been x-rayed at random and the result was positive!
– What is your intuitive assessment of the chances that is cracked?
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
29SPE DL 2010
Probability Intuition: Assessing Reliability of Predictors
• Historical estimates suggest one in every 1000 blow-out preventers has serious cracks.
• Suppose x-ray analysis is a very good but not• Suppose x-ray analysis is a very good, but not perfect, detector of these cracks.
– If a blow-out preventer has cracks, x-rays will correctlysay it has them 99% of the time
– If a blow-out preventer does not have cracks, x-rays will wrongly say that it has them 2% of the time
4.7%!P( test positive given crack)=99%
P( crack given test positive)=4.7%
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• A blow-out preventer has been x-rayed at random and the result was positive!
– What is your intuitive assessment of the chances that is cracked?
Reliability of predictors: Results
70
80
90
30
40
50
60
0
No
of
Par
tici
pan
ts
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
0
10
20
0 20 40 60 80 100Estimated Probability
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
30SPE DL 2010
Intuition is Overrated
• Many decision-makers believe that intuition, repeated experience and their general intelligence will see them through
• Human beings are imperfect information processors
• We can’t always trust our intuition and perception. particularly in an uncertain environment!
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• We need to use the appropriatetools and frameworks to address the uncertainties and decisions.
Anchoring - Subtle changes in wording of a question can significantly impact responses
Question Outcome
How long was the movie? 130 mino o g as e o e
How short was the movie?
30
100 min
Question Outcome
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
How wide are the channels?
How narrow are the channels?
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
31SPE DL 2010
Anchoring Question: Large Industry Sample
• Alternate versions of question with high and low anchors given to two groups
Hi h A h G “W ld d il i– High Anchor Group: “Were world proved oil reserves in 2003 greater or less than 1721.6 Billion Barrels?”
– Low Anchor Group: “Were world proved oil reserves in 2003 greater or less than 573.9 Billion Barrels?”
B th th k d th ti
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• Both groups then asked the same question
– “What is your best estimate of the world proved oil reserves in 2003?”
Anchoring Results: Large Industry Sample
3000
3500
Anchor
1932
1722
1000
1500
2000
2500
mat
ed W
orl
d P
rove
d
ves
2003
+
/-1s
d
Estimate
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
574
682
0
500
Low HighAnchor Group
Mea
n E
stim
Res
erv
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
32SPE DL 2010
Anchoring Results: Large Industry Sample
3000
3500
Anchor
1932
1722
1000
1500
2000
2500
mat
ed W
orl
d P
rove
d
ves
2003
+
/-1s
d
EstimateCommon approach in E&P project evaluation:
“Let’s start with a base case and then build some scenarios around it ”
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
574
682
0
500
Low HighAnchor Group
Mea
n E
stim
Res
erv then build some scenarios around it.
• Unpacking Question (RoW Packed)
– “What % of world oil consumption is accounted for by each of the following regions: North
Unpacking
for by each of the following regions: North America, Europe/Eurasia and the Rest of the World?”
• Unpacking Question (RoW Unpacked)
“What % of world oil consumption is accounted
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
– What % of world oil consumption is accounted for by each of the following regions: North America, Europe/Eurasia, South and Central America, Middle East, Africa, Asia Pacific and the Rest of the World?”.
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
33SPE DL 2010
Unpacking Question Results
50
60
Oil
1SD
True:
Packed
Unpacked
20
30
40
mat
ed
% o
f W
orl
d
um
pti
on
(20
03)
+/- Unpacked
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
0
10
N America Europe/Eur Rest of World
Est
iC
on
s u
Region
Risk Aversion and Incentives
• Your track record hasn’t been too good recently. You can recommend one of two investments. Which one?
P Pf EV $MMPs Pf EV, $MM
“Safe” 80% 20% 10
“Risky” 10% 90% 20
• You should take a corporate (organizational) attitude to risk, not a personal one, and recommend “Risky” based on EV
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
o a pe so a o e, a d eco e d s y based o
– any other choice (being risk-averse or risk-seeking) is value destroying (money–losing)
• Most incentive policies, focused on reward by outcome, encourage inappropriate risk-aversion, therefore value loss!
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
34SPE DL 2010
De-biasing
• Create awareness.– Anchoring– Overconfidence– Availability, Vividness & Recency
• Understand the limits of our knowledge.
• Actively challenge ourselves.
– Stop to consider reasons why your judgement might be wrong.
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
• Abandon false comfort of single-point predictions.– Use ranges instead of single-point estimates.– Use multiple anchors.
• Calibration.– Feedback and accountability.
Reasoning under uncertainty = using the rules of probability
• In terms of practical applicability, probability theory is comparable with geometry;
- both are branches of applied mathematics that are- both are branches of applied mathematics that are directly linked with the problems of daily life.
• While most people have a natural feel for geometry (at least to some extent), many people clearly have trouble developing a good intuition for probability.
• In no other branch of mathematics is it so easy to
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
o ot e b a c o at e at cs s t so easy tomake mistakes as in probability theory.
- Conditional probabilities, and Bayes theorem in particular, are especially difficult
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
35SPE DL 2010
Reasoning under uncertainty = using the rules of probability
• In terms of practical applicability, probability theory is comparable with geometry;
- both are branches of applied mathematics that are
“The theory of probabilities is at bottom nothing but common sense reduced to calculus; …- both are branches of applied mathematics that are
directly linked with the problems of daily life.
• While most people have a natural feel for geometry (at least to some extent), many people clearly have trouble developing a good intuition for probability.
• In no other branch of mathematics is it so easy to
;
It teaches us to avoid the illusions which often mislead us;
… there is no science more worthy of our contemplations nor a more useful one for admission to our system of public education ”
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
o ot e b a c o at e at cs s t so easy tomake mistakes as in probability theory.
- Conditional probabilities, and Bayes theorem in particular, are especially difficult
to our system of public education.Laplace – Theorie Analytique des Probabilites
Conclusions
• In our context, uncertainty is a function of what we know about a situation – its in our heads, not an objective attribute of the “system”- there is no single, “right” probability for an uncertain event
- variability is not the same thing as uncertainty
• Evolution has not “wired” our brains for a good natural ability to assess uncertainty
- training helps, but even industry experts, including those
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
g y gwhose job it is to deal with uncertainty, are not great
• Do not use intuition to propagate (amalgamate) assessed uncertainties
- use the rules of probability, or Monte Carlo simulation
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
36SPE DL 2010
Understanding the Limits of Our Knowledge
• To know that we know what we know, and that we do not knowknow, and that we do not know what we do not know, that is true knowledge
Confucius
“It’s not what we don’t know that gets
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
It s not what we don t know that gets us into trouble, it’s what we know that
ain’t so”Will Rogers
The Pioneers
• Kahneman and Tversky– 2002 Nobel Price in Economics
• Russo and Schoemaker• Russo and Schoemaker– Decision Traps
– Winning Decisions
• Thaler– The Winner’s Curse
• Bazerman
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
Bazerman– Judgment in Managerial Decision Making
• Plous– The Psychology of Judgment and Decision Making
Steve Begg, ASP, University AdelaideExpert Reliability & Uncertainty
37SPE DL 2010
Acknowledgements & Further [email protected]
• Prof. Reidar Bratvold– University of Stavanger, Norway
• Dr. Matthew Welsh– University of Adelaide
• Dr. Michael Lee– University of California, Irvine
• SPE 77509– “Would You Know a Good Decision if You Saw One?”
• SPE 84238
2010-2011 SPE Distinguished Lecture Experts and Uncertainty Steve Begg
– “Shrinks or Quants: Who will improve decision-making?”
• SPE 96423
– “Cognitive Biases in the Petroleum Industry: Impact and Remediation”
Thi k th 50thThis year marks the 50th
anniversary of the SPE Distinguished Lecturer program. Please visit our site to learn more about this amazing program.
www.spe.org/go/DL50