DataPacRat comments on Open Thread, Feb 8 - Feb 15, 2016 - Less Wrong

4 Post author: Elo 08 February 2016 04:47AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Article Navigation

Comments (215)
Tags: open_thread

Comments (215)

Sort By: Controversial
Best Popular New Controversial Top Old Leading

You are viewing a single comment's thread.

Comment author: DataPacRat 10 February 2016 11:30:32PM 1 point [-]

Seeking socio-econo-political organizing methods

How many useful ways are there for an uploaded mind, an em, to organize copies of itself to maximize the accuracy of their final predictions?

The few that I've been able to think of:

  • "Strict hierarchy". DPR.2.1 can advise DPR.2, but DPR.2's decision overrides DPR.2.1's.
  • "One em, one vote". DPR.2 gets a vote, and so does DPR.2.
  • "One subjective year, one vote". DPR.2.1 is running twice as fast as DPR.2, and so DPR.2.1 gets twice as many votes.
  • "Prediction market". The DPRs implement some sort of internal currency (which, thanks to blockchains, is fairly easy), and make bets, receiving rewards for accurate predictions.
  • "Human swarm". Based on https://www.singularityweblog.com/unanimous-ai-louis-rosenberg-on-human-swarming/ .

How many reasonably plausible methods am I missing?

Comment author: _rpd 11 February 2016 04:28:36AM 0 points [-]

"Prediction market". The DPRs implement some sort of internal currency (which, thanks to blockchains, is fairly easy), and make bets, receiving rewards for accurate predictions.

Taking this a little further, the final prediction can be a weighted combination of the individual predictions, with the weights corresponding to historical or expected accuracy.

However different individuals will likely specialize to be more accurate with regard to different cognitive tasks (in fact, you may wish to set up the reward economy to encourage such specialization), so that the set of weights will vary by cognitive task, or more generally become a weighting function if you can define some sort of sensible topology for the cognitive task space.

Powered by Reddit Powered by Reddit