And people could vote on which of the re-phrasings (a) minimize semantic noise, and (b) reasonably capture the original phrase's meaning/intent. That latter part might not be possible, and should be dropped as design goal.
- could AI/ML do the former, as in Brin's Earth?
- probably need a WebOfTrust to filter out the crazy others
And then, of the highest-scoring (for minimum noise) re-phrasings, users could vote on the Accuracy of the statement. That's not binary or even a spectrum, see the categories at Nature of Truth.
Users could attach sources to their votes.
Those sources might have statements needing re-phrasing... :)
Statements that are predictions should make good quality Bets. (Idea Futures)
- update: this might be the best place to start, because it puts skin in the game, and provides monetization... Downside is this only supports predictions about the future, not descriptions of the present/past....
- and/or/yes/and.... maybe web3 makes it easy to turn everything into a bet, chasing out bullshit and bad-faith? (see thinking in bets framing at bottom)
- do you still need an oracle either way?
Statements which are conditional predictions could only be settled if the precursor condition became true. (Futarchy)
Chains of logic of rephrasings A/B/C/D could also be evaluated: (A & B & C) -> D?
- Why? Because the "atomic" claims are usually "data points" supporting a larger claim
- So we need a WebOfTruth to balance the weights.
One benefit to this idea is that well-factored claims can be pointed/linked to, and given more weight-votes (convergence), vs the infinite variety of noise.
cf (2017-03-13) Udell Teaching Students To Marshal Evidence And Evaluate Claims, (2012-12-03) Journalism Service Lists Outlines And Data (Journalism), Disputation Arena, Schroeder Degrees Of Freedom, e-prime