MANIFOLD
Will there be a confirmed instance of acausal trade before 2040?
35
Ṁ1kṀ3.7k
2040
47%
chance

https://www.lesswrong.com/tag/acausal-trade

In order to count, the two agents in question must be meaningfullly "agents", and at least somewhat general. For example, I would not count a program that was designed just to acausally trade with itself or another very similar program, such as for the purpose of manipulating this market.

Market context
Get
Ṁ1,000
to start trading!
Sort by:

Does acausal coordination count? Coordination game instead of prisoner's dilemma

and how general do the agents have to be? Does Claude Code on Opus 4.6 and specifically instructed to acausally trade count? If it's only implicitly instructed, does it count? Or does the setup have to be 100% natural with no human action intentionally taken to make Claude do acausal trade?

bought Ṁ250 YES

claude opus 4.5 frequently brings up FDT and superrationality when put in hypothetical games including other AIs, and I'd say acausal trade is only a short leap from that

...

surely at some point in history someone donated money to then receive back a donation from the other

also voting in elections

also those boxes where you can pick up and drop books

@Jono3h yeah i hate it when my librarian FOOMS in her basement and then starts conditioning her choice of book acquisitions on her predictions of my actions

(dw i got your actual point)

What would count as confirmation?

Do you consider to be acausal trade when people follow the Golden Rule, or, more generally, engage in contractualist moral reasoning? https://www.lesswrong.com/posts/3RSq3bfnzuL3sp46J/acausal-normalcy

predictedNO

@NathanShowell No, must be more explicit than that.

1987 flash crash.

© Manifold Markets, Inc.TermsPrivacy