two guesses better than one

a new study finds that the average of two of someone's guesses is more accurate than the first or second guess alone. the authors posit that we draw from an internal probability distribution instead of having one best guess. further research could help us make a better estimates, critical for professions such as medicine and the military.

this indicates that the "wisdom of crowds" idea, roughly that averaging a crowd's answers is better than asking one person, applies to individuals as well. as surowiecki notes, this has many caveats. however, scienceblogs is curiously attempting to falsify it using this guessing study.

how is that? well, they erroneously claim that under w-o-c each person must give stable answers over time, their "best guess." they then misinterpret this guessing study to say that second guesses were more accurate (the study clearly states the reverse). ergo, w-o-c is false because people don't give best guesses first.

rather, this study may bolster woc: even though people are possibly drawing from some internal distribution of answers, the woc effect persists. this study simply illustrates the benefits of "benefits from polling the 'crowd' within" when making your own estimates, leaving the idea of polling from the crowd outside to stand on its own merit.

Labels: decision making