Consensus Prediction Markets

A consensus prediction market is a mechanism where a prediction market is created to predict the outcome of a vote, and then settled by the outcome of the vote.

This can be interesting from the perspective of content curation because there is generally a large group of users who want to casually consume content, but a much smaller subset of users who are willing to spend time sorting through and curating new content. For example, in a curation community like reddit, where posts are organized into popular, trending, and new views. Users who filter through the spam and low quality content on in the “new” section add tremendous value but are generally not rewarded. By introducing a consensus prediction market, users trying to promote spammy or low quality content to the front page have to put skin in the game, and users who sort through spammy content can make money by betting against them. The net result is that more people will curate content, and the content that reaches the wider community should be higher quality.

This mechanism has been discussed within the curation markets community quite a bit, and is currently being implemented by the RelevantFeed team.

Consensus Prediction Markets can also be applied to DAO governance, which is what DAOStack is exploring with the Holographic Consensus protocol.

For consensus prediction markets to work the following need to be true:

  1. participants are confident in their ability to predict the preferences of the voting group.
  2. participation in the market doesn’t significantly alter the preferences of the voting group.

One instance where this might not be the case If the voting group contains a small number of influential voters (or coordinated voting blocks), as these influential voters will be highly confident in their ability swing votes, at some point they may have 100 percent certainty their vote will be decisive, at which point they can settle the prediction market since they will be the only person/group that knows that the outcome is certain.

If one of these decisive voters/voting blocks does not have a strong preference, their position as a decisive voter will give them the incentive to swing the vote towards the minority preference, simply because it is more profitable to settle the market in favor of the “under” position compared to the “over” position.

If decisive voters do emerge, and have an incentive to flip the outcome, then participants won’t be confident in their ability to predict the preferences of the voting group and wont participate in the prediction market.

Agreed that consensus-based prediction markets won’t work if market participants can manipulate oracle processes. Augur, the leading dApp PM, safeguards against that problem by ultimately resolving disputes of fact with a hard fork. Since by far most token holders will want to be associated with a platform that respects consensus reality, they will tend to outvote lying manipulators.

The Gnosis and Bodhi dApp PMs, in contrast, ultimately rely on “beauty contests” resolved by the faction willing to bet the most on its favored claim. That approach, though probably not as resistant to manipulation as Augur’s, tends to steer markets toward consensus results in most cases.