Korva is a scroll of the internet’s best ideas.
Fast feedback doesn't mean fast learning
Feedback without learning
Learning is understood to happen when there is plenty of feedback tied closely in time to decisions and actions. Poker has exactly that, and learning still wasn't happening very well. The missing piece is that the outcomes of a decision are only loosely connected to how good the decision was.
Two sources of uncertainty
Uncertainty comes from two places. One is hidden information: in poker the cards are face down, and in life there is plenty that can't be known. The other is luck. Even with a coin whose odds are known and no hidden information about it, the next flip can't be predicted. Over 10,000 flips the pattern shows up, but most decisions never get run that way.
One trial at a time
Real decisions arrive one trial at a time, and the results are absorbed one at a time. Waiting for enough trials is impractical anyway: nobody collects thousands of trials before deciding whom to marry.
When does fast feedback actually build expertise?
In a 2009 paper in American Psychologist, Daniel Kahneman and Gary Klein examined when expert intuition can be trusted. They concluded that skilled intuition develops in environments with stable regularities and a chance to learn them through prompt, clear feedback. Environments where outcomes depend heavily on chance or hidden information fail that test even when feedback arrives quickly. The two authors had long disagreed about intuition and used the paper to map where they agreed.
Why do 10,000 coin flips reveal the pattern?
The law of large numbers, first proved by Jacob Bernoulli in a work published in 1713, says that results of repeated trials drift toward their true probability. For a fair coin flipped 10,000 times, the standard deviation of the number of heads is 50. That means the count lands between 4,900 and 5,100 about 95 percent of the time. Ten flips, by contrast, can easily produce seven heads.
Why do small samples mislead so much?
In a 1971 paper titled Belief in the Law of Small Numbers, Amos Tversky and Daniel Kahneman showed that people expect small samples to look like the population they came from. Even trained researchers made this error in choosing sample sizes. A short run of results can show streaks and lopsided splits purely by chance. The paper became an early landmark in the study of judgment errors.
More from Talks at Google
