Encounters With Risk PPP (Perception,Policy and Practice) During a Career in Operations Research Stephen Pollock University of Michigan 1 Oct 3, 2007 A PERSONAL VIEW OF RISK PPP • MY EXPERIENCES WITH RISK PPP WHILE DOING O.R. • CLEARLY A BIASED PERSPECTIVE • A NUMBER OF ANECDOTAL EXAMPLES THAT MIGHT SERVE TO FORESHADOW OR TIE TOGETHER THE DIVERSE ISSUES OF THIS WORKSHOP 2 Oct 3, 2007 O.R./I.E/ ENGINEERING? • OPERATIONAL PROBLEM SOLVING • MOSTLY MATHEMATICAL MODELS • ALSO A WAY OF THINKING: – WATER GLASS – GUILLOTINE – FORK • UNCERTAINTY ALWAYS PRESENT 3 Oct 3, 2007 TYPICAL DECISION ANALYSTS VIEW d3 p1 d1 d2 d4 DECISION 1-p1 p2 1- p2 CONSEQUENCE x1 CONSEQUENCE x2 CONSEQUENCE x3 CONSEQUENCE x4 CHANCE 4 Oct 3, 2007 MORE GENERIC VIEW d f(x|d) CONSEQUENCE x DECISION CHANCE 5 Oct 3, 2007 MORE GENERAL VIEW d DECISION d’(d,x) f(x|d) CHANCE* *”UNCERTAINTY” DECISION f(y|d’,x) CHANCE CONSEQUENCE (X,Y) 6 Oct 3, 2007 WHERE DOES RISK POLICY AND PERCEPTION FIT THIS SCHEMA? d DECISION f(x|d) UNCERTAINTY X CONSEQUENCE 7 Oct 3, 2007 WHERE DOES RISK POLICY AND PERCEPTION FIT THIS SCHEMA? d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK 8 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK RISK HAS TWO NECESSARY ASPECTS: • UNCERTAINTY -- WHAT WILL HAPPEN? • CONSEQUENCES -- WHY DOES ONE CARE? CONFOUNDING THESE ARE • CONCEPTION (WHAT ONE THINKS THE “RISK” ASPECTS ARE) • PERCEPTION (HOW ONE “SEES” THE RISK ASPECTS) • CODIFICATION (HOW ONE “TALKS” ABOUT -- OR SHOULD TALK ABOUT -- RISK ASPECTS) 9 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE THE POLICY COMPONENT (DECISION/MITIGATION) -- WHAT SHOULD ONE DO? THIS ALSO INVOLVES A MIXTURE OF • CONCEPTION (WHERE DO POSSIBLE DECISIONS/OPTIONS/POLICIES COME FROM?) • PERCEPTION (HOW ONE “SEES” OPTIONS) • CODIFICATION (HOW TO DESCRIBE OPTIONS) 10 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE ?? FINALLY (AND PERHAPS MOST IMPORTANT): PRACTICE (WHAT ACTUALLY GETS DONE) 11 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK TO THE “PERSON IN THE STREET”, ARE THE FOLLOWING “RISKY”? •BUNGEE JUMPING •NOT BUCKLING UP •BUCKLING UP •PICKING UP A $20 BILL FROM THE STREET •INOCULATING A CHILD AGAINST MEASLES •NOT INOCULATING A CHILD AGAINST MEASLES •LIVING “EVERYDAY” ANSWERS SHOW ALL SORTS OF COGNITIVE BIASES, BUT NEGATIVE FRAMING SEEMS TO BE PERVASIVE Oct 3, 2007 12 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK SOME “EVERYDAY”* CONCEPTION, PERCEPTION, AND CODIFICATION OF RISK: *NY TIMES, NEW YORKER, NPR, ETC. 13 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK “EVERYDAY” CONCEPTION, PERCEPTION AND CODIFICATION OF RISK Souffle (NPR 7/23) David Denby’s review of “A Fine Romance”: (referring to romantic film comedies) “ …with a married couple, romance is like “…a duel with slingshots at close quarters –- exciting but a little risky” (New Yorker 7/23) 14 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK “EVERYDAY” CONCEPTION, PERCEPTION AND CODIFICATION OF RISK “The Black Swan” popular book dealing with “The role of the unexpected” in financial trading (NYT B.R. 7/29) ”What is “unexpected”? {13 craps in a row} {this particular person dies within the next five years} {this levee fails} {a levee fails} 15 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK “EVERYDAY” CONCEPTION, PERCEPTION AND CODIFICATION OF RISK Louis Menan’s review of Caplan’s “The Myth of the Rational Voter: Why Democracies Choose Bad Politics” (New Yorker 7/19) “You can’t use futures markets for assessing probabilities like you can for guessing the number of jellybeans in a bowl, or odds in sports gambling.” 16 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK “EVERYDAY” CONCEPTION, PERCEPTION AND CODIFICATION OF RISK Louis Menan’s (continued): “People exaggerate the risk of loss; they like the status quo and tend to regard it as a norm; they overreact to sensational but unrepresentative information (shark attack phenomenon) … Most people, even if you explained …the economically rational choice … would be reluctant to make it, because … in particular, they want to protect themselves from the downside of change. They would rather feel good about themselves than to maximize ... profit.” 17 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK “EVERYDAY” CONCEPTION, PERCEPTION AND CODIFICATION OF RISK Levitt and Dubner’s article “The Jane Fonda Effect” “... in 1916 … the legendary economist Frank Knight made a distinction between two key factors in decision making: risk and uncertainty. The cardinal difference, he declared, is that risk — however great — can be measured, whereas uncertainty cannot.” (NYT Magazine 9/16) 18 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK “EVERYDAY” CONCEPTION, PERCEPTION AND CODIFICATION OF RISK Levitt and Dubner’s* article “The Jane Fonda Effect” “… Has fear of a [nuclear] meltdown subsided, or has it merely been replaced by the fear of global warming? …. in 1916 … the legendary economist Frank Knight made a distinction between two key factors in decision making: risk and uncertainty. The cardinal difference, he declared, is that risk — however great — can be measured, whereas uncertainty cannot.” (NYT Magazine 9/16) * Authors of “Freakonomics” 19 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK “EVERYDAY” CONCEPTION, PERCEPTION AND CODIFICATION OF RISK AND POLICY HELMET WEARING BY NHL PLAYERS (1979-80 SEASON REQUIREMENT) 20 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY HOW MUCH TIME TO CARRY A GARBAGE CAN FROM A BACK YARD TO THE CURB? WHICH CUTTER HEADS WERE THE DEFECTIVE ONES? WILL A SUB PASS BY DURING AN ASWEX? 21 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY ELLSBERG PARADOX BAG A HAS 500 RED BALLS AND 500 GREEN BALLS BAG B HAS 1000 RED AND GREEN BALLS YOU CHOSE A BAG, DRAW A BALL AND WIN $100 IF IT IS RED. WHICH BAG DO YOU PREFER? MOST PEOPLE PREFER A 22 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY ELLSBERG PARADOX DEMONSTRATES MANY PEOPLE’S PREFERENCE FOR “CRISP” PROBABILITIES VS. “AMBIGUOUS” ONES 23 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY ELLSBERG PARADOX -- PREFERENCE FOR “CRISP” PROBABILITIES VS. “AMBIGUOUS” ONES BAG A HAS 450 RED BALLS AND 550 GREEN BALLS BAG B HAS 1000 RED AND GREEN BALLS YOU CHOSE A BAG, DRAW A BALL AND WIN $100 IF IT IS RED. NOW WHICH BAG DO YOU PREFER? 24 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY ELLSBERG PARADOX -- PREFERENCE (?) FOR “CRISP” PROBABILITIES VS. “AMBIGUOUS” ONES BAG A HAS 200 RED BALLS AND 800 GREEN BALLS BAG B HAS 1000 RED AND GREEN BALLS YOU CHOSE A BAG, DRAW A BALL AND WIN $100 IF IT IS RED. NOW WHICH BAG DO YOU PREFER? 25 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY “CRISP” VS. “AMBIGUOUS” PROBABILITIES •TEST OF CONCEPT OF TOTAL RE-DESIGN OF A GLOBAL LOGISTIC CHAIN VIA M.C. SIMULATION •USED PREVIOUS YEAR’S DEMAND DISTRIBUTION FOR TO PROVE OUT RE-DESIGN CONCEPT •MASSIVE CORPORATE PRESSURE AGAINST USING DISTRIBUTION OVER SIMULATION MODEL’S PARAMETERS -- SINCE “WE WON’T KNOW WHAT THEY MIGHT BE” 26 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY “CRISP” VS. “AMBIGUOUS” PROBABILITIES •ORDER SIZE FOR CRITICAL MATERIAL BASED ON PROJECTED PRODUCT SALES AND MATERIAL PRICE •SALES BASED UPON “TARGETS” •MATERIAL PRICE BASED ON SPECIALIST’S FORECASTS •MASSIVE CORPORATE PRESSURE AGAINST USING DISTRIBUTION OVER EITHER SALES OR PRICES “THESE ARE EXPERTS, THEY SHOULD KNOW THE ANSWER” 27 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY AMBIGUITY EXPRESSED AS PROBABILITY DISTRIBUTIONS OVER PROBABILITIES (REF: H. KUNREUTHER) uncertainty in Probability uncertainty in Loss Probability {loss > v} 95% “mean” prob. 5% v ($) 28 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY WHAT IF ADVERSARIES CHOOSE THE PROBABILITIES? 29 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE ADVERSARIAL RISK ANALYSIS FOOTBALL ANALOGY [OFFENSE’S VIEW] d OFFENSE PLAYER PERSONNEL d’(x) f(x|d) DEFENSIVE ALIGNMENT AUDIBLE PLAY CALL f(y|x,d’(x)) PLAY OUTCOME (y) 30 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF POLICY WHAT IF THE DECISION MAKER CHOOSES THE PROBABILITIES? CONSIDER THE MATHEMATICAL PROGRAMMING PROBLEM: min f(x) s.t. g(x, z) ≥ 0 31 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF POLICY “CHANCE CONSTRAINED PROGRAMMING” min f(x) s.t. Prob. { g(x, Z) ≥ 0 } ≥ p WHERE Z IS NOW A RANDOM VARIABLE 32 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE EXAMPLE: DETERMINE d = BANK’S CASH RESERVES BANK WANTS TO MINIMIZE d REGULATORS REQUIRE SMALL PROBABILITY OF RUNNING OUT OF CASH; Z = DEMAND FOR CASH (A R.V.) min d s.t. Prob. { Z ≥ d } ≥ .90 .10 p(z) z x = 2M 33 Oct 3, 2007 d f(x|d) POLICY X CONSEQUENCE UNCERTAINTY HOW ABOUT A RANDOMIZED POLICY? min x s.t. Prob. {x ≥ Z} ≥ .90 .55 .01 z x = .6M Prob. = .2 x = 2.1M .8 Prob. { x > Z} = .2(.55) +.8(.99) = .901 (OK) E(X) = .2(.6M) + .8(2.1M) = 1.93 ( better than 2) BUT -- WOULD THE REGULATORS APPROVE?? 34 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF POLICY •CAMSHAFT HARDENING •SEWAGE TREATMENT IN *** •1979 NHL HELMET REGULATION (REVISITED) 35 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY A US FEDERAL DEPARTMENT REPORT USES, WITH LITTLE DIFFERENTIATION: PROBABILITY LIKELIHOOD CHANCE FREQUENCY RELATIVE PROBABILITY STOCHASTIC PROBABILITY 36 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY A US FEDERAL DEPARTMENT, IN A PRA REPORT, USES, AS SYNONYMS: MEAN AVERAGE 37 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY ANOTHER US FEDERAL DEPARTMENT USES, AS SYNONYMS: DISTRIBUTION FUNCTION DENSITY FUNCTION PROBABILITY FUNCTION PROBABILITY DISTRIBUTION DISTRIBUTION 38 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF RISK (computation) FEDERAL DEPARTMENT REPORT: RISK = "the probability or frequency of an event multiplied by the consequences of the event” 39 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK CONCEPTION, PERCEPTION AND CODIFICATION OF RISK Society for Risk Analysis (SRA) Glossary (http://sra.org/resources_glossary.php) RISK = “The potential for unwanted, adverse consequences to human life, health, property, or the environment; BUT THEN SRA GOES ON TO SAY: “estimation of risk is usually based on the expected value of the conditional probability of the event occurring times the consequence of the event given that it has occurred." 40 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK CODIFICATION OF RISK IN A FORTHCOMING PROPOSED LEXICON RISK = The potential for unwanted, adverse consequences. It is important to distinguish between the term "risk,” which involves uncertainties, consequences and conditioning statements, and "expected risk" [q.v.] which combines these factors using a linear additive operation. PROBABILITY = One of a set of numerical values between 0 and 1 assigned to a collection of random events (which are subsets of a sample space) in such a way that the assigned numbers obey axioms [ …] 41 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE MORE CODIFICATION FROM THE PROPOSED LEXICON CONSEQUENCE (OUTCOME) = A description of a scenario, in terms of measurable factors, that a decision-maker may consider when assessing preferences over different scenarios; these factors are often random variables. EXPECTED RISK = A summary measure of risk for an event, scenario, etc., as expressed by the expected value of any one of the measurable consequences associated with the risk. 42 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE RISK CODIFICATION OF UNCERTAINTY IN A FORTHCOMING PROPOSED LEXICON •ALEATORY PROBABILITY = A measure of the uncertainty of an unknown event whose occurrence is governed by some random physical phenomena that are either: a) predictable, in principle, with sufficient information (e.g., tossing a die); or b) phenomena which are essentially unpredictable (radioactive decay). •EPISTEMIC PROBABILITY = A representation of uncertainty about propositions due to incomplete knowledge. Such propositions may be about either past or future events. Oct 3, 2007 43 (ANECDOTE) “NEWSVENDOR” STADIUM HOTDOG DEMAND NORMAL SEASON CONTENDER 44 Oct 3, 2007 (ANECDOTE) “NEWSVENDOR” STADIUM HOTDOGS NORMAL SEASON PROBABILITY = ? CONTENDER PROB = 1-? 45 Oct 3, 2007 (ANECDOTE) “NEWSVENDOR” STADIUM HOTDOGS OVERALL SALES 46 Oct 3, 2007 (ANECDOTE) “NEWSVENDOR” STADIUM HOTDOGS WHICH PROBS ARE ALEATORY WHICH ARE EPISTEMIC? NORMAL SEASON, CONTENDER PROBABILITIES OR CONDITIONAL DEMAND DISTRIBUTIONS? 47 Oct 3, 2007 (ANECDOTE) “NEWSVENDOR” STADIUM HOTDOGS DIFFERENCE BETWEEN IGNORANCE AND APATHY? 48 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF CONSEQUENCES WE’RE REALLY TALKING ABOUT APPROPRIATE PERFORMANCE MEASURES THIS IS DIFFICULT ENOUGH TO DO IN DETERMINISTIC O.R. -- THE UNCERTAINTY ASPECT ONLY MAKES THINGS MORE “INTERESTING” 49 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF CONSEQUENCES SUBMARINE SEARCH: MINIMIZE EXPECTED TIME TO DETECT ? MAXIMIZE PROB. {DETECT TIME ≤ Tcritical} ? 50 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF CONSEQUENCES ALLOCATION OF POLICE PATROLS: MINIMIZE EXPECTED TIME TO RESPOND ? MINIMIZE VARIANCE OF RESPONSE TIME ? 51 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF CONSEQUENCES QUALITY CONTROL AND 6-SIGMA “TAGUCHI” LOSS FUNCTION ESSENTIALLY QUADRATIC: DECISION IS d, RANDOM VARIABLE IS X 2 LOSS = CONST. (d - x) 52 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF CONSEQUENCES 2 “TAGUCHI” LOSS FUNCTION C(d - x) x d 53 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF CONSEQUENCES WITH “TAGUCHI” LOSS FUNCTION BEST DECISION IS d* = E(X) WHICH RESULTS IN MINIMUM EXPECTED LOSS = CONST•VAR(X) MINIMUM EXPECTED RISK = CONST•S.D.(X) = CONST•SIGMA 54 Oct 3, 2007 d POLICY f(x|d) X CONSEQUENCE UNCERTAINTY (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF CONSEQUENCES MORE REALISTIC MANUFACTURING LOSS FUNCTION x d Oct 3, 2007 55 INTENSITY MODULATED RADIATION TREATMENT (IMRT) radiation beam tumor critical (healthy) tissue 56 Oct 3, 2007 typical 2-D slice of 3-D image TUMOR PAROTID GLAND SPINAL CORD 57 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE THE FUNDAMENTAL PROBLEM IN IMRT DETERMINE THE NUMBER, ANGLES AND INTENSITIES OF THE BEAMLETS SO THAT a) THE TUMOR RECEIVES A SUFFICIENT DOSE, BUT b) CRITICAL TISSUES (E.G. NORMAL ORGANS) ARE SPARED HIGH DOSES 58 Oct 3, 2007 CONFLICTING CONSEQUENCES • want to have: – at least a “critical dose” of radiation absorbed by the tumor “target” – as little radiation as possible absorbed by healthy tissue • this is a mathematical programming problem, right? min (radiation to healthy tissue) s.t. (radiation to tumor) critical dose or max (radiation to tumor) s.t. (radiation to healthy tissue) damaging dose 59 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE GRAPHICAL REPRESENTATION OF CONSEQUENCES: DOSE-VOLUME HISTOGRAM (“DVH”) 60 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE ADD UNCERTAINTY TO DVH 61 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE NOTE SIMILARITY TO KUNREUTHER’S REPRESENTATION uncertainty in Probability uncertainty in Loss Probability {loss > v} 95% “mean” prob. 5% v ($) 62 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE REMEMBER PRACTICE? (WHAT ACTUALLY GETS DONE) 63 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X PRACTICE CONSEQUENCE PRACTICE REMOVING SKUS FROM EOQ POLICIES FACULTY RETIREMENT OPTIONS CLASS ACTION SINKING FUND SIMULATED ANNEALING AND IMRT “X BAR” CONTROL CHART DESIGN 64 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X PRACTICE CONSEQUENCE HERE IS A FINAL CHALLENGE: DURING THIS WORKSHOP, TRY TO DISCOVER IMPLIED RESOLUTIONS (USUALLY VIA ASSUMPTIONS) TO THE INHERENTLY PROBLEMATIC NATURE OF RISK Conception, Perception, Policy, and Practce 65 Oct 3, 2007 CLICK TO HEAR NEXT SPEAKER 66 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE PRACTICE (UNLIKELY DERIVATIVES) From the New York Times, 2/23/89 -- p. 1(!): “The underlying rate of inflation is accelerating” If X(t) = PRICE INDEX, then INFLATION = X(t) RATE of INFLATION = X(t) “ ... is accelerating” ==> [X(t) > 0 ==> X(t) >0 From the NYT 2/4/02 “The increase in Chip Speed is Accelerating …” [work it out …] 67 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (EVERYDAY) CONCEPTION, PERCEPTION AND CODIFICATION OF UNCERTAINTY “CRISP” VS. “AMBIGUOUS” PROBABILITIES Building a prison in Berlin NH would produce a revenue increase of $264,000/year (NYT 9/2) 68 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE INTERESTING CONTRAST IN USAGE INTELLIGENCE COMMUNITY “ANALYSIS”: GATHER INFORMATION ABOUT OPPONENT’S INTENTIONS AND CAPABILITIES, “ASSESSMENT”: USE THIS INFORMATION TO PRESENT A STATEMENT OF THE CURRENT SITUATION RISK AND DECISION COMMUNITY “ASSESSMENT”: OBTAIN INFORMATION ABOUT UNCERTAINTY OF EVENTS (AND ALSO CONSEQUENCE) “ANALYSIS”: USE THIS INFORMATION AND COMBINE IT IN SUCH A WAY THAT ONE CAN MAKE BETTER DECISIONS 69 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF POLICY “MORAL HAZARD”?: SEEDING HURRICANES •CONSEQUENCES = WIND SPEED •SPEED NOW IS 70 MPH •WITH NO SEEDING, FORECAST: 140 MPH AT LANDFALL •WITH SEEDING, FORECAST: 100 MPH AT LANDFALL •DECISION IS TO SEED ==> WIND IS 90 MPH AT LANDFALL •SEEDING “CLEARLY” INCREASED THE SPEED, SO SUE THE GOVERNMENT 70 Oct 3, 2007 d POLICY f(x|d) UNCERTAINTY X CONSEQUENCE (ANECDOTAL) CONCEPTION, PERCEPTION AND CODIFICATION OF POLICY “MORAL HAZARD” •DECISION MAKER DOES NOT SUFFER CONSEQUENCES •HURRICANE INSURANCE •SUB-PRIME LENDING INSTITUTIONS •“DONORCYCLES” 71 Oct 3, 2007