Celebrity Deathmatch: Mencius vs Hanson

I finally got around to seeing the video of Robin Hanson debating Mencius Moldbug about "futarchy".

As some have remarked, that was one nerdy gathering. I think if I showed up there, some of the attendees would have hid under the table.

Most of the questions in the follow-up session were piss-poor (except, of course, DDF's) showing that nerds aren't always that smart about markets and government. Then again, I'm just one of those young doctors that Hanson believes thinks they know everything.

I agree with Hanson that futarchy should be judged against the status quo, not against some hypothetical ideal.

Getting down to the meat of the argument, I don't think liquid prediction markets can be manipulated to any significant degree. We have examples of prediction markets right now: stock markets and sports betting markets. It's nearly impossible to predictably manipulate the stock market, and the only way to manipulate the sports betting markets is for a player to throw a game, something that's rare enough to be a non-factor.

Can decision markets be manipulated? That was the question at the center of the debate. The example most understandable to me was a decision market on whether or not Steve Jobs should be replaced as Apple's CEO by a chimpanzee. Let's look at that more closely, because I'm not sure I understand the mechanisms that well. Futures used in prediction markets, I understand. But I don't know how a decision market would actually work.

I can't find where exactly in the video Hanson mentions the "called off bets", and I don't feel like watching the whole thing over again, but lets say there are two cash for stock markets, one for keeping Jobs as CEO (symbol: KEEP), and one for replacing Jobs with a chimp (symbol: DUMP).

On a given day, KEEP is worth $15 and DUMP is worth $10. What exactly does this mean? What am I buying? As I understand it, the losing side gets their money back. So if the board sees that the price of KEEP is higher than the price of DUMP at some predetermined time in the future, it keeps Jobs as CEO of Apple, the KEEPers win, the DUMPers get their money back (because that would be the called-off bet). What do the KEEPers win? What's the payoff? Is there something that ties KEEP to AAPL?

I can't have any sort of opinion on whether decision markets can be manipulated because I don't understand how they would actually work. I suspect this was also the case with most of the audience of the debate.

Share this

We have a decision market

It's called K Street.

Two things: First, the

Two things: First, the contracts KEEP and DUMP aren't just "whether the ceo gets kept or dumped", but "conditional on KEEP/DUMP, the company improves by some measure (say, profits above Y 2 years hence). The decision part is that whichever contract has the higher price is the action taken. The other contract is nullified and all bets are returned. The winning contract is performed, and pays out if the profits ACTUALLY go above Y on the date specified.

Second, manipulating a market that has only one dimension (price) does happen, but isn't that big a deal. It's when there are two dimensions (price and decision) that it becomes likely. Looking at intrade pricing on political outcomes, there's fairly strong evidence of manipulation - people making bad-money bets in order to influence reported predictions.

Markets and craps

These days there is little difference between the markets and gambling. In the markets, the house edge goes to the guy with the fastest computer and the best software.


Perhaps Mencius has some valid idea rolling around in his head, but his chess analogies were very wrong-headed. I can can tell you exactly what would happen if chess moves were arrived at through a decision market. Contrary to what Mencius thinks, it is now trivial for patzers to produce grandmaster quality moves. Do you think people playing for money aren't going to use their computers to produce moves?

He says such a market won't produce good enough moves because the market won't be able to produce long term plans, he scoffs at the idea of running a company algorithmically. He scoffs for precisely the same reason that so many scoffed at the idea that algorithmically produced chess could ever surpass a human grandmaster. But grandmasters can no longer compete with algorithms.

The fact that he is so wrong about what he considers such a strong analogy suggests to me that he may not have a very firm grip on the topic.

Who does he expect to hire his proposed monarch? I didn't follow the answer.