Conversation

Replying to
Analogous to why the thought that some massive gradient-descent based algorithm is going to be the best kind of AI we might ever invent 🤬😬🤢🤮😖
5
25
Interesting distortions include: reason (eg planning), urgent temporal forcing functions (climate change), low liquidity regimes and “art” pricing, ‘economics of pricelessness’, bureaucracy (= human GOFAI), smart contracts, HFT, exotic securitized instruments...
2
33
Problem with all these distortion ideas is that they are almost always easier to hack than use as intended. Cf the Lebowski theorem
Quote Tweet
The Lebowski theorem: No superintelligent AI is going to bother with a task that is harder than hacking its reward function
Show this thread
1
24
Still, out of every 100 ideas to do better than today’s market, if 99 are easier to hack than use, 1 is a genuine advance within or outside the ‘market’ paradigm. So I like the idea of an eternal arms race between markets and imaginative attempts to distort them.
1
15
You’ll always end up with one of 3 things: A new kind of criminal (99%) A new kind of market (0.9%) Something smarter than a market (0.1%)
1
28
Dumb view: both markets and distortions are eternal patterns. Free-market fundamentalist view: markets evolve, distortions don’t Reactionary view (both left and right): markets are static but my new kind of distortion will deliver utopia Ze Truth: Arms race. Both evolve.
4
32
Replying to
Well, it’d be disastrous as well as depressing (and incorrect) to claim progress was over in *any* area. One can’t predict improvements that haven’t been conceived of yet. And culture is hugely broader than markets. This interaction itself is an example of cashless cooperation.
1
2
Show replies
Replying to
I don’t think there is a non-distorted platonic market. Therefore the practical question is what kinds of distortions are going to give us the results we want. Semi related - one of the things old towns in England always brag about is the date the monarch granted them a market