Improving Crowd Governance With The Surprisingly Popular Algorithm
At Ardent United, we’re working to create an esports organization that is fully crowd governed. Studies show that groups of people, even people with no domain expertise, make better predictions than experts. The Good Judgement Project, funded by the US government, aggregates the geopolitical predictions of normal people. Together, their predictions are more accurate then CIA analysts with access to classified information.
However, these predictions can become more accurate. MIT recently did a study on crowd wisdom and showed that the Surprisingly Popular algorithm increases prediction accuracy by 21.3%.
So, how does the Surprisingly Popular algorithm work? Instead of just being asked one question (the prediction), the crowd is asked two. What they think the right answer is, and what they think the most popular option will be. Or, as paper co-author Drazen Prelec puts it, “In situations where there is enough information in the crowd to determine the correct answer to a question, that answer will be the one [that] most outperforms expectations.”
Why does this work? In every group, there’s people with more information than average. They would be the “mini-experts”. Even though they’re not well known experts in the field, they still carry small bits of information that could be useful. However, even if they hold a contrarian belief, they would know that it is a contrarian belief. As a result, they would still vote that their belief would not be shared by most people. This means that their vote would be surprisingly popular because their contrarian belief would be more popular than expected.
By making better predictions, a company can also make better decisions. Here’s an example of how Ardent United will use the Surprisingly Popular algorithm. Imagine that we wanted to get into a new game. We would ask our token holders, do you think that Game X will reach 1 million monthly viewers by Date Y.
Now, let’s pretend that 40% said Yes and 60% said No. However, 30% think most people will say Yes and 70% think most people will say No. This could be because some people that voted Yes knew that Game X was going to release a big new update, but they also knew that most people wouldn’t have that information, so they voted that most people would say No.
Even though most people thought that Game X would not reach 1 million monthly viewers, it’s more likely that it will, because the Yes vote was surprisingly popular. Here’s the general formula:
Through using the Surprisingly Popular algorithm, Ardent looks to make decisions based off of more accurate predictions. We’re also looking at more crowd governance methods, like breaking token holders off into small discussion groups.
We’d love to know what you think, so please connect with us on Twitter at @ArdentUnited and on Discord at https://www.discord.gg/aun
Thanks for reading!
