3 ms·

I'm a fan of liquid democarcy on a conceptual level so this result is surprising to me. Started a side project to test it out in a game environment, basically a

by POiNTx 4y ago

I'm a fan of liquid democarcy on a conceptual level so this result is surprising to me. Started a side project to test it out in a game environment, basically an online Nomic[1] although I have abandoned it for now. Reading through some of the paper makes me realize I didn't think it through as deep as the authors and I'm probably not smart enough. I understand they used a perceptual test to figure out who the experts are. And 'over delegation' results in losing benefits of delegates who have information that might be useful? This is not clear to me: "...it (over delegation) reduces the variety of independent information sources" In a perceptual task where one is found to be better than the other participants, why wouldn't it be the case that voting for the best participent (the expert) gives you the best result? I think I'm missing something [1] https://en.wikipedia.org/wiki/Nomic https://en.wikipedia.org/wiki/Nomic

heisenzombie 4y ago
Assume we're deciding on a binary choice. Also assume that there is one correct and one wrong answer, and the only reason people choose wrong is a lack of information. Next assume each participant in the vote has a particular chance of getting the right answer. Lets take an extreme case where you have a million people each with a 55% chance of getting the right answer, and one expert with an 80% chance of getting the right answer. If you have a normal vote, the right answer is EXTREMELY likely to win. But if everyone delegates to the expert (after all 80% is better than 55%) then the right answer only has an 80% chance of winning. Worse outcome.
- POiNTx 4y ago
  Ah this is helpful thank you. So it's a variance thing? So by this logic, expanding the amount of experts who are delegates would probably fix the issue? Let's say you have a 100 experts in charge who have an average of 70% right answer, it would still be extremely likely the right answer gets chosen. It seems more experts would be selected in a real world system as it doesn't have all the assumptions you mentioned. - No binary choice (multiple solutions to one problem) - Not always a right or wrong answer (a lot of issues are preferential)
  - heisenzombie 4y ago
    That's kind of what the experiment was about. They wanted to know: If we run this simple game of chance and delegation with real people, do people delegate more or less than the optimal amount? They found that people delegated more than would be optimal. Now -- whether that says anything about liquid democracy applied to real world situations is quite a different question, of course. As you started to point out.

by POiNTx 4y ago

heisenzombie 4y ago

Assume we're deciding on a binary choice. Also assume that there is one correct and one wrong answer, and the only reason people choose wrong is a lack of information. Next assume each participant in the vote has a particular chance of getting the right answer. Lets take an extreme case where you have a million people each with a 55% chance of getting the right answer, and one expert with an 80% chance of getting the right answer. If you have a normal vote, the right answer is EXTREMELY likely to win. But if everyone delegates to the expert (after all 80% is better than 55%) then the right answer only has an 80% chance of winning. Worse outcome.

POiNTx 4y ago
Ah this is helpful thank you. So it's a variance thing? So by this logic, expanding the amount of experts who are delegates would probably fix the issue? Let's say you have a 100 experts in charge who have an average of 70% right answer, it would still be extremely likely the right answer gets chosen. It seems more experts would be selected in a real world system as it doesn't have all the assumptions you mentioned. - No binary choice (multiple solutions to one problem) - Not always a right or wrong answer (a lot of issues are preferential)
- heisenzombie 4y ago
  That's kind of what the experiment was about. They wanted to know: If we run this simple game of chance and delegation with real people, do people delegate more or less than the optimal amount? They found that people delegated more than would be optimal. Now -- whether that says anything about liquid democracy applied to real world situations is quite a different question, of course. As you started to point out.