= 27d567bc2d48d9f40e9ccc20ea370b15)
gwern comments on Rationality Power Tools - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (65)
In October I will be writing a Ruby on rails web app for experimenting with control markets to control a real world organisation.
Simple control markets* are used in Learning Classifier systems to make the system goal oriented. I'm heavily modifying it from that, but the basic idea remains the same, control of parts of the system is based on bidding a currency which is gained from feedback. I'm curious to see if they will work in the real world, or whether aspects of human psychology will make them untenable.
The theory behind them needs to be tightened up a lot as well. What you need to do is prove that the Evolutionary Stable Strategies for your control markets are non-malicious. But that is hard for me, at the moment.
*Warning - Neologism. I chose it for the allusion to prediction markets. Related: Agorics and market-based control.
Edit: An article not designed for lesswrong, can be found here please put any notes of what you would like in a lesswrong article there as well.
How has progress been?
Heh. I've switched track; I'm working on another idea. Having code modified at load time to make exploiting it harder (due to return-oriented code not knowing where to exploit). That is going slow, mainly due to having to understand other peoples code and machine code, which I am rusty on. I've been meaning to contact developers of as and the elf-loaders.
The reasoning behind swtiching track. It'll look better on my CV. It is easier to get people to adopt. And may help reduce existential risk/ world takeover of computers. It'll take up less of my time when it is done, so it makes sense to do it first.
I'll get back on the web-app once that is done.