What does "How do we prove 8 billion people agree to a constitution" mean? Does it mean how do we secure a big referendum that decides whether a constitution is applied with 50%+1 of the vote?

No. It would require 100% of the vote. Maybe read this.

https://manifold.markets/Krantz/if-the-work-between-anthropic-and-t?r=S3JhbnR6

In general, you appear to be imagining a constitution similar to something like the US constitution.

I'm talking about a constitution in a technical sense for computers to operate on.

@strutheo 'How does this resolve?'

I think that's an important question society should consider.

(you should answer it as the market creator)

other wise this is going to get marked unranked or whalebait or something

That's a very unfortunate consequence of a broken system.

I hope they actually take the time to listen to what I'm trying to say.

whatever you're trying to say, i dont think a multiple choice market really makes sense here

(nobody is going to trade if it isnt clear what they're trading on. and any trades you get before you make the resolution clear will be noise that tells you nothing)

Ok. Here's a possible resolution criteria that might address your concerns while preserving the intent behind listing the prediction. (which is to teach people that instead of wagering their money on predictive markets, they could be getting paid to align AI instead without needing to put down any capital at all).

If Anthropic is still pursuing constitutional alignment on August 1st 2025, which principles will exist on Claude's constitution?

https://www.anthropic.com/news/claudes-constitution

This would resolve according the the state of Claude's constitution (or equivalent if renamed) on August 1st, 2025.

Thoughts?

could be pretty cool actually, you should make that a separate market i think some ai people would dig it

how does this resolve?

I believe that the criteria for saying something is 'resolved' is similar to the criteria for saying something is 'objectively true'.

As a philosopher, I believe I can never know whether something is objectively true.

I think prediction markets are flawed because of this.

I think it's important to talk about that.

It's a principle I want AI to understand.

>As a philosopher, I believe I can never know whether something is objectively true.

You should be posting some essays about this on substack or something, participating in Manifold like this is just wasting all of our time.

I believe that the criteria for saying something is 'resolved' is similar to the criteria for saying something is 'objectively true'.

Why not instead resolve options at a sufficiently high non-absolute credence of a given third-party-verifiable criterion, so that you can harness the incentives of a prediction market to produce actual information?

Other than Hume's problem of induction?

Because you need to trust the third party verifier.

Do you trust nobody at all?