Chap 1 – Humans try to guess the pattern, and in the process allow ourselves to be outperformed by a rat.
Echos of Kahneman. Rats do fine choosing on the basis of frequency, versus finding the patterns for humans. Human outcomes not much better.
Regression to the mean: praise makes things worse, a chewing out makes things better.
p. 14 so typical patterns of randomness ... are routinely misinterpreted & worse…
Does wisdom of the crowd disappear when people collaborate? p. 26 The ability to see meaningful connections ...may ... be worth a few mirages
Hawthorne effect -- any change to the environment improves performance.
Innovation results from looking for patterns, can be for good or ill, need to have some “adventurous” types. - Value of diversity.
p. 40 With respect to DNA evidence, "...even when attorneys & judges don't quote explicit probabilities ... often employ this sort of reasoning.”
Probability without Bayes is misleading
Lab error rate totally dominates all other errors/probabilities. If you don’t understand the math, it becomes irrelevant (at best).
Jury selection based on patterns that may or may not be relevant!
Chap 3, loc 971 Monty Hall Problem
I'm not sure anyone can describe this, all those Math PhD's who disagree may be right
No, they’re not! Consider the 100 door case, where you pick one, then I reveal 98.
Wight’s summary: there are two cases to consider: (1) Lucky Guess scenario where your initial choice was correct (probability of 1 in 3), and (2) Wrong Guess scenario where your initial choice was incorrect (probability of 2 in 3). Chap 3 - Only a computer simulation would convince Erdös or Jeff or Pat that switching increased the odds of winning Sample space is key.
Chap 3, loc 821 Invention of Equal Sign: Didn't Algebra require = ?
Yes, it did, that’s why it was a breakthrough. But .EQ. will do.
Chap 4- Pascal’s Wager - Expected payoff on piety is infinity/2 (your gain if God exists) minus small number/2 (your loss if he doesn't exist)
If he’d known there were something north of 2500 practiced religions, would he still feel that way.
Chap 5 – Bernoulli’s golden theorem calculates how many samples are “enough,” based any personal definition of almost certain “You can be as certain as you are prepared to pay for.”
Loc 1813 - Confusing (1) the probability that a series of events would happen if there were a huge conspiracy with (2) the probability that a huge conspiracy exists if a series of events occurs. (or even if not)
We expect big effects to have big causes.
He didn’t respond because he’s getting ready to fire me. (Assumptions)
Confusing cause with effect
Our intuition is terrible.
“One should not appraise human action on the basis of its results.” - Jakob Bernoulli Location 1706 of 4091 We assume agency and intentionality.
Loc 2111 – In close elections, we do recounts, as if our 2nd or 3rd count of the votes will be less affected by random errors than our first.
There are always random errors in any measurement. The error analysis is key. Measure twice, cut once. "any measurement … is susceptible to random variance and error"
Location 2091 of 4091 Different measurements have different error profiles
“Merely because the number has changed it doesn’t necessarily mean that a thing itself has changed. For example, any time the unemployment rate moves by a tenth of a percentage point…that is a change that is so small, there is no way to tell whether there really was a change.” - Gene Epstein, the economics editor of Barron’s Location 2252 of 4091
Does the difference really make a difference? Errors and noise in every measurement.
Pay attention to the 7-day moving average, forget the daily numbers
Loc 2882 - The longer the sequence, the greater the probability that you’ll find every pattern imaginable—purely by chance.
Chapter 9, Loc 3076 - Fundamental clash between our need to feel we are in control -- Agile "Self Organized"?
Misinterpreting what we can control/predict
Self-organizing teams - “we control ourselves”
When I was on one… what made it work was a well-defined division of responsibilities
“I don’t care how you do it.” Implies there is faith that you will get it done, or ask for help if you’re stuck.
Stringing a few winners (or losers) doesn’t really mean much. Working with customers -- something happens when you’re working with the ultimate end-users that provides a rich source of errors.