Risk Savvy- Gerd Gigerenzer
Chapter 1
Too much knowledge -> experts
But we shouldn't entirely trust them because they are prone to miscommunication and conflicts of interest
First few rules:
1) Understand the reference class
Weather forecast says there's a 30% chance of rain… meaning it'll rain for 30% of the day? Over 30% of the land? Or that 30% of forecasters believe it will rain… percent of what?
2) Absolute % vs. relative %
UK Pill Scare in the 60s when oral contraceptives were said to "double" risk of thrombosis… from 1/7000 to 2/7000, effect = 13000 extra abortions
3) Don't overestimate the likelihood of a repeat event
Risk of dying from driving 12 miles same as risk of dyin in a nonstop flight… effect of 9/11 = 1600 deaths from traffic accidents
Chapter 2
Certainty is an illusion, risk isn't uncertainty, and more info doesn't mean more accuracy
e.g. False +ve in AIDS tests… if prevalence of disease is 1 in 10000, but FP is 1 in 20000… 1/3 of all positives are false
Chapter 3
Build an accepting error culture, one that doesn’t promote defensive decision making (choosing inferior option B in order to protect oneself)
Compare error cultures of medicine vs aviation, see relevant article:
Chapter 4
Psychological shift towards external over internal goals: graduates favoring "being well off financially" over "developing a meaningful philosophy of life" while the opposite was true in the 60s-70s, accompanied by rises in narcissism and self-perceived falls in control/autonomy
More external goals with less control = greater anxiety and unhappiness, especially in the face of uncertainty
Chapter 5
Bias-variance tradeoff/dilemma: "make everything as simple as possible, but not simpler." Albert Einstein
If complex, make it simple, if simple, make it complex…
Reducible error = bias^2 + variance + noise
Error due to squared bias is introduced at model selection, the amount by which the model prediction differs from the true value over all the data
Error due to variance is how inconsistent predictions are from one another, over different training sets, regardles sof whether they are accurate or not
Small variance + high bias = underfit, vice-versa = overfit
Markowitz's (Nobel) mean-variance portfolio allocation is complex with lots of variance = lots of free elements = requires >500 years of stock performance data
Chapter 6
Much executive decision-making is intuition-based but fear of personal responsibility => defensive decision-making and producing reasons after the fact
Hindsight makes complex methods perform far better as information increases, but pre-event simple methods work far better:
1/n > Markowitz
Hiatus rule: inactive if no purchase in 9 months > complex Pareto NBD (marketing)
Intuition is not the opposite of rationality, it is unconscious intelligence based on personal experience and smart rules of thumb. In order to harness it, middle management should be given more autonomy, and more time.
Chapter 7
If probability fails, use natural frequencies, i.e. results
Monty Hall
Types of decision making:
Maximizing, such people are generally more perfectionist, depressed, and likely to self-blame
Satisficing
Imitating
Habit
Advice taking
(Eg shopping/dining… Satisficing is more efficient, FOMO from not maximizing will dissipate after a few positive experiences)
Chapter 8
Rules of thumb:
One-reason decision making
Leverage rule (10:1)
3:6:3, borrow at 3, lend at 6, be at golf by 3
3:1, 3x raw food price
Gaze heuristic
Hire well and let them do their job
Complex maximization calculations fail horribly when there is risk and uncertainty
37% rule from secretary problem outlines an efficient procedure: (1) set aspiration level (2) stop searching when it's met
Chapter 9
DIC (defensive decision-making, innumeracy, conflicts of interest) syndrome
10m unnecessary pap smears in the US... Women didn't even have cervixes
Healthcare reform in the US requires information revolution: sharing of patient and study information
Chapter 10
More people die with prostate cancer than from it (3%), yet PSA, the screening technique, costs US $12-30B/y
Difference in US/UK survival rates, made famous by Giuliani, are caused by lead time and over-diagnosis bias, in that early screenings lead to better 5 year "survival rates" when in actuality all those affected by progressive prostate cancers die, and that screenings catch non-progressive cancers resulting in unnecessary treatments
The "Great Prostate Mistake", Richard J. Ablin, the man who invented the PSA:
Testing should absolutely not be deployed to screen the entire population of men over the age of 50, the outcome pushed by those who stand to profit. I never dreamed that my discovery four decades ago would lead to such a profit-driven public health disaster. The medical community must confront reality and stop the inappropriate use of PSA screening. Doing so would save billions of dollars and rescue millions of men from unnecessary, debilitating treatments.
Double-tonguing - reporting survival rates and comparing them to mortality rates
E.g. saying mammography leads to 98% 5Y survival rate and reduces mortality rate by 30% (5 in 1000 to 4 in 1000, but it doesn't mention unnecessary treatment caused by false positive)
1m cigarettes = 1 death with decades lead time and $10000 profit (Proctor 2012)
Chapter 11
Risk can be calculated if and only if
Uncertainty is low
There are few components, not too many places where risk can play a part
Large amount of data
Consider components in financial regulation:
1998 basil I 30 pages, 20 in US Legislation
2004 basil II 347 pages
2019(?) basil III 616 pages, 1000+
Also
Avoid fear of dread fear
Only 150 people died in Europe in a decade due to BSE
Similar stories for SARS and swine flu (which WHO over-warned about saying there'd be 2B infections and recommended governments to stockpile Tamiflu which had been shown by the FDA to be ineffective… Because WHO experts had ties to pharma (Cohen and Carter 2010))
Chapter 12
Teach digital, medical, and financial literacy to children… its too late to tell a 15yo to stop smoking
Parents who read daily to 8-16m baby increase language test score by 7 points vs decrease of 17 for every daily hour on DVD (Zimmerman 2007)
Heavy media multitaskers performed worse on a test of task-switching ability, likely due to reduced ability to filter out interference from the irrelevant task set. These results demonstrate that media multitasking, a rapidly growing societal trend, is associated with a distinct approach to fundamental information processing. (Ophir 2009)
Liberty cannot be preserved without a general knowledge among the people. - John Adams
Hard vs. paternalism- coercing people into behaving, defensible if actions are proved to increase welfare … vs nudging certain behaviors, such as via opt-out programs