4 ms·

Would it be possible to build a consensus algorithm for factual information? Sort of a decentralized fact-checker system

alehul 8y ago
That seems like it would be very trivial to game, and additionally while facts are hard to confirm, a consensus of users who are biased and relatively uneducated on the topic would do a particularly bad job anyway. The best similar use case I can imagine is building a system is to evaluate how much of the media is in agreement on a given topic, and use that to combat fake news. If a similar headline isn't found elsewhere, give it a low truth rating.
- hobofan 8y ago
  > a consensus of users who are biased and relatively uneducated on the topic would do a particularly bad job anyway That's why you build the system with an incentive scheme with the contributors having something at stake, that over time weeds out the incompetent contributors, similarly to a PoS scheme.
  - douglaswlance 8y ago
    That's what I was thinking too. If you have PoS, then the your stake is basically how trusted you have been throughout the history of your wallet.
- douglaswlance 8y ago
  So we'd need a debate forum that is deeply integrated into the system that has a forcing function for truth.
- douglaswlance 8y ago
  Media is a terrible metric. Their incentive is attention, not truth. The consensus of media is worse than the consensus of the public.
  - alehul 8y ago
    I agree. This wouldn't expose hyperbole and bias, but merely be able to advocate for whether the event is being widely reported to protect against completely false news (think celebrity death as an example)
wmf 8y ago
This is called prediction markets and a bunch of people are working on it; e.g. http://bitcoinhivemind.com/ http://bitcoinhivemind.com/ and https://www.augur.net/ https://www.augur.net/
hobofan 8y ago
Yes, that's more or less what we are doing with Rlay (https://rlay.com https://rlay.com).

alehul 8y ago

That seems like it would be very trivial to game, and additionally while facts are hard to confirm, a consensus of users who are biased and relatively uneducated on the topic would do a particularly bad job anyway. The best similar use case I can imagine is building a system is to evaluate how much of the media is in agreement on a given topic, and use that to combat fake news. If a similar headline isn't found elsewhere, give it a low truth rating.

hobofan 8y ago
> a consensus of users who are biased and relatively uneducated on the topic would do a particularly bad job anyway That's why you build the system with an incentive scheme with the contributors having something at stake, that over time weeds out the incompetent contributors, similarly to a PoS scheme.
- douglaswlance 8y ago
  That's what I was thinking too. If you have PoS, then the your stake is basically how trusted you have been throughout the history of your wallet.
douglaswlance 8y ago
So we'd need a debate forum that is deeply integrated into the system that has a forcing function for truth.
douglaswlance 8y ago
Media is a terrible metric. Their incentive is attention, not truth. The consensus of media is worse than the consensus of the public.
- alehul 8y ago
  I agree. This wouldn't expose hyperbole and bias, but merely be able to advocate for whether the event is being widely reported to protect against completely false news (think celebrity death as an example)

wmf 8y ago

This is called prediction markets and a bunch of people are working on it; e.g. http://bitcoinhivemind.com/ http://bitcoinhivemind.com/ and https://www.augur.net/ https://www.augur.net/

hobofan 8y ago

Yes, that's more or less what we are doing with Rlay (https://rlay.com https://rlay.com).