“The wisdom of crowds” has apparently seeped a bit into popular culture, or at least the geekier end of it.
On the heels of British illusionist Derren Brown’s invoking of “the wisdom of crowds” as a (false) part of his explanation of how he appeared to predict winning lottery numbers, last night a character in the American TV show House invoked the wisdom of crowds as part of an explanation for how he obtained a diagnosis of his medical condition.
(The character – a highly intelligent, geeky, successful video game designer – posted his medical symptoms on the internet and offered $25,000 for a successful diagnosis. Then, mentioning “wisdom of crowd” based reasoning, concluded that the most frequent diagnosis appearing in emailed responses was likely correct. As the story turned out, the crowd-sourced diagnosis was incorrect. Instead, the correct diagnosis was submitted by series main character Greg House, working from home after quitting his job at the hospital. The “wisdom of crowds” element doesn’t make it into the official episode summary.)
Although the crowd was wrong (the better to highlight how clever our main character is when, later, he provides the correct diagnosis), at least the basic “wisdom of crowds” logic illustrated in the episode was correct. As a fan of the show, I appreciate that it doesn’t insult my intelligence by dressing up clever cons with misleading science-based patter.
So, six years after the publication of the Wisdom of Crowds, the notion creeps into the public mainstream, as a tool of deception and comedy. Says it all, really …..
“at least the basic “wisdom of crowds” logic illustrated in the episode was correct”
What is the mechanism, here? The more people claim the solution is XXX, the more likely that solution is correct? As in “Who wants to be a millionaire?”, when they poll people in the room where they tap the show.
Just as a thought experiment, what would be a more interesting mechanism? Making people bet (real) money on their preferred solution?
More interesting mechanisms:
If you had a list of N possible diagnoses, you could run independent yes/no prediction markets (more than one possible diagnosis can be correct at the same time). However, there may be interactions between diagnoses such that is one is correct another is more likely or less likely to be correct, and this information may be relevant to the preferred medical treatment.
Would the good doctor Hanson prescribe combinatorial prediction markets?