3 ms·
One of the wiki's sources might have been a better linked article. The MIT summary [1] notes "The researchers first derived their result mathematically, then a
by tofof 7y ago
One of the wiki's sources might have been a better linked article.
The MIT summary [1] notes "The researchers first derived their result mathematically, then assessed how it works in practice, through surveys spanning a range of subjects, including U.S. state capitols, general knowledge, medical diagnoses by dermatologists, and art auction estimates." Across all those areas, this technique had error rates about 20% lower than other competing techniques. Those techniques included simple majority vote to two different kinds of confidence-weighted scoring.
The paper earned a prestigious publication in Nature.
1: http://news.mit.edu/2017/algorithm-better-wisdom-crowds-0125 http://news.mit.edu/2017/algorithm-better-wisdom-crowds-0125
- carbocation 7y agoAwesome, thank you. And 'nacc found the link to the Nature article as well: https://news.ycombinator.com/item?id=20547787 https://news.ycombinator.com/item?id=20547787
- bscphil 7y agoOne interesting thing about the paper is that it seems that the Wikipedia article incorrectly describes the procedure: respondents were not asked to guess whether the majority would agree with their position. They were asked to guess what per cent of other respondents would agree. I think that's a pretty severe difference in the method.