In Today’s Financial Times, Tim Harford reviews the promise and problems of prizes, quoting me along the way: Champions of prizes see them as a component of a wider system to promote innovation, rather than as an outright replacement either for grants or patents. Instead, the hope is that prizes will help to compensate for the specific weaknesses of those alternatives. … In an ideal world, prizes could replace patents. … But to explain that idea is to see its limitations. How could the government know enough about the costs and benefits – and even the very possibility – of an innovation to put a price tag on it and write the terms of reference for a prize competition? …
Robin, there are these things called 'suggestion boxes' which are meant for anonymous contributions. Surely you've heard of them?
Now, maybe it's just me, but if those suggestion boxes are never opened, if, in fact, no one can put any more suggestions in because the box is too full, but I would think that would imply that those suggestions aren't being looked at.
Have you ever worked in a shop with equipment that actually made things? Your suggestion seems . . . theoretical. And it also leads me to believe that you think the utility of 'prediction markets' is prediction. Isn't the standard belief that their strength is in allocating risk?
I think prizes for things the Clay Millennium problems are very useful in the long term, but not because they motivate researchers. Almost everybody who already has the capability and interest to work on such a problem is probably working on it or something similar already.
The key benefits of the prizes, in my opinion, are in raising public awareness, which will increase long-term funding and public support, and in inspiring the smart and talented children of today to become the discoverers of 2020, drawing them into mathematics or physics or theoretical computer science (WOW, computer science isn't about programming and web pages!) or ... rather than engineering or law or....
The sciences are a tough sell: most pre-university level teachers are terrible, and there are few role models. I believe things like the Clay prizes will help in this regard.
Scent, there are many cases where "the guys on the floor" would not give you an honest answer to a direct question, but would be honest with anonymous trading.
The point I was making Richard, is that to the extent these markets work, it's because the guys on the floor are participating; it would have been simpler just to listen to those guys on the floor in the first place. But apparently, that's out these days, and prediction markets are in.
Yes, Phil, that's 'apparently'. These arguments against grants were made long ago. We don't need yet more warmed-over libertarianism here. That ideology has been thoroughly exploded, I would think.
Since Scott Aaronson is a professional theoretical computer scientist, I asked him to say more about the motivations of mathematicians. He gave me permission to publish his reply:
"I meant that within the relevant mathematical communities, these problems are already so well-known that anyone with a promising idea for how to solve them would almost certainly pursue it, without the added inducement of a monetary reward. In fact, we already have empirical data on this question: Grisha Perelman, who proved the Poincare Conjecture, has expressed no interest whatsoever in collecting the million-dollar prize he's now entitled to. His motivations, like those of most mathematicians at his level, are much more internal (even if Perelman is admittedly an extreme case)."
Yet another solution in search of a problem (and one that ignores history to boot; there are reasons why prizes are not the preferred funding mechanism they used to be.)You apparently skipped the two paragraphs on the reasons why prizes are not the preferred funding mechanism they used to be.
What is to prevent the crew on the floor from playing the prediction market?
It's almost comical to imagine a serious mathematician who wouldn't otherwise work on a Clay Millennium problem, deciding to do so because of the $1M prize. At the same time, that prize singlehandedly did a huge amount to increase public interest in mathematical research. As Harford points out, there were similar publicity benefits from the X Prize and Netflix prize (as well as an example he didn't mention, the DARPA grand challenge for autonomous vehicles). So I think the more the merrier with prizes, although they're certainly not yet a replacement for the grant system.
The problem is when the prediction markets start saying "fire the pointy-haired boss..."
I'm only suggesting that 'predictive' markets, to the extent that they are said to predict, derive their benefits from giving weight to data that would otherwise not have been considered. So the canonical pointy-haired boss won't listen to the crew on the floor, but he will point to his use of 'prediction markets' - whose effectiveness solely derives from it's consideration of information he has explicitly rejected - as a sign of his cutting edge decider capabilities.
Iow, it's largely a gimmick to sell to the rubes at the top.
In fact, in many cases it seems to suggest that if the bosses just listened to what their employees thought, their businesses would be much better run.
You're forgetting one of Putt's Laws. "Decisions are justified by considering benefits to the company; decisions are made by considering benefits to the decision makers." In other words, better decision making procedures never get adopted because that would take power, influence, and salary from those with those who currently have the power to make decisions, and those with the power to make decisions get to choose what decision making procedures to adopt.
Yet another solution in search of a problem (and one that ignores history to boot; there are reasons why prizes are not the preferred funding mechanism they used to be.)
'Prediction Markets' - If all you're talking about is allocating risks, yes, there seems to be some place for them as a tool. But if you're talking about using them to make predictions, well, the consensus seems to be that if nothing else works you can try them, hope that somebody out there knows something relevant (iow, they might be better aggregators of data in some circumstances.) Other than that, they seem pretty spotty. Witness for example the poor predictions in the primaries this year.
Iow, another solution in search of a problem - in fact, in many cases it seems to suggest that if the bosses just listened to what their employees thought, their businesses would be much better run.