Korva is a scroll of the internet’s best ideas.
Offering a bet ends certainty
A confident friend
Picture a friend flatly predicting that one party will win the House in the fall, and the reply is an offer to bet on it. The friend suddenly has reasons to doubt. There is a lot of time between now and November, plenty of events can intervene on the luck side, and the opposing view, which the friend may not have read, starts to matter.
Two kinds of uncertainty
A bet brings both forms of uncertainty to the surface. Almost nothing is known with perfect information, and the future stays uncertain no matter how well it is predicted: even a bad outcome with a 1 percent chance can still happen. Putting real time, money, or health behind a claim raises the question of what the other person knows that you don't.
Hunting for information
Once beliefs and predictions are probabilities, information becomes valuable, because refining the odds protects limited resources. That pushes toward opinions that disagree. Certainty came from knowing why you're right, so backing away from it means asking what the other side knows.
Do real-money bets make forecasts better?
The Iowa Electronic Markets, run by the University of Iowa since 1988, let traders buy and sell contracts tied to election results using small real stakes. A review of the market's record by Joyce Berg, Forrest Nelson, and Thomas Rietz found its final-week prices were closer to the actual vote shares than polls in a clear majority of comparisons. Markets of this kind are called prediction markets. The Iowa market is operated for research and teaching.
How good are confident experts at forecasting?
Psychologist Philip Tetlock tracked about 28,000 predictions from 284 experts over roughly two decades. He reported the results in his 2005 book Expert Political Judgment. The experts did little better than simple statistical rules, and those who held one big theory did worse than those who drew on many sources. Confidence in a prediction was not a reliable sign of accuracy.
More from Talks at Google
