Scientists are incentivised not to buck the consensus ... unless they have a really good idea that they can prove, in which case they are incentivised to. That's what you get nobels for. The idea that cientific consensus is an entirely conservative force is wrong in theory and practice.
So you're saying that consensuses are reliable because a very small number of scientists are incentivised to disagree with them? That's a strong argument in *favor* of markets, because all actors in a market are incentivised to make correct decisions, not just a handful with the potential to achieve winner take all prestiege.
The market can move before the consensus does, and this is actually a likely scenario due to the factors mentioned above around scientists being incentivized to reinforce the status quo. Markets aren't subject to the same politics and sociology.
This isn't true either. One way to determine winning without consensus is to demonstrate something is true in a way that gives one control over reality, in a way that enhances one's chance at survival. The one person in a band of hunter-gatherers who understands animal behavior well enough to lay traps in the right place will eat when others starve. The few people who understand gunpowder and metalsmithing well enough to build guns will outfight the masses who stick to spears and bows.
It may even be that a minority nation that embraces the scientific method will (eventually) outlive a larger nation that eschews it, although "eschew" is a strong enough term that I cannot think of a historical example that qualifies.
Scientific consensus is not something different from empirical demonstration. Unrepeatable results (such as cold fusion) quite rightly fail to move the consensus
You're assuming scientific consensus involves only reproducible experiments.
Are you saying that we need not ascribe that label to claims made in science-y circles that aren't reproducible?
I'm open to that, since it would let us assign a more lofty connotation to the term. But it also means that we should deny that term to certain claims made in fields such as string theory, but also the claims in immunology and climate science that tend to get the most news. How are we going to correct all those accounts?
They are also taught at length that the contrarian ideas they find really good are unlikely to actually be really good, especially if they are multiple. And among those who don't learn that lesson, most become crackpots. Even if Nobels weren't limited to "one best this year", few scientists would have something that qualified for them, many just "get by", e.g. by going in detail on some small aspect and deepening its understanding without bucking the consensus.
Yes. This is a bit of an inside-outside view problem, I guess? If you already taught at length that most contrarians are wrong, you are more likely to judge yourself as wrong when you actually step on a correct contrarian idea - or, even if not wrong, to judge your idea as being unlikely to be accepted by the community (perhaps correctly, as status quo bias is powerful). So, there are certainly strong incentives for scientists to be conservative even in their research.
Depending on your definition of good, that’s not necessarily the case. Making money is a clear objective but is that always in alignment with making predictions in any useful sense?
The context of Scott's comment was prediction markets, where you get money if your prediction turns out to be correct. The only distorted incentive I can think of is if you believe that by betting on something happening you can make it more likely to happen through the effect of your bet on the price and of the price on other people. That could be true in is an election market, where if you make your favored candidate appear more likely to win other people will be more willing to vote for him or contribute money to his campaign, and similarly in the other direction.
The angle I am wondering about is the dynamic of gaming a market, depending on how it’s structured. In a very liquid and efficient market a lot of profitable trades can be made that are not very predictive.
I understand, but I don’t get how that translates into any useful tool for social policy/ war and peace planning etc. Profitable decisions in the financial markets are made and realized in milliseconds or decades.
If this structure is optimal then I don’t see what prediction markets bring to the party; we have good financial markets for signals.
I f one wishes for a market free of arbitrage then it’s not a good model.
Financial markets are only indirectly prediction markets on many specific topics, or not prediction markets at all. Financial markets are subject to all kinds of information, and the most pertinent factor or combination of factors will dictate market terms, meaning they offer only tenuous signals for many individual factors.
There's little need to prediction markets over what interest rates will be in 5 years time. But a prdiction market for the next presidential election tells us a hell of a lot more about who the next president will be than any financial market will.
>Financial markets are subject to all kinds of information, and the most pertinent factor or combination of factors will dictate market terms, meaning they offer only tenuous signals for many individual factors.
How do prediction markets differ?
>But a prdiction market for the next presidential election tells us a hell of a lot more about who the next president will be than any financial market will.
Perhaps I’m dense, but I don’t see why this would be the case.
Scientists are incentivised not to buck the consensus ... unless they have a really good idea that they can prove, in which case they are incentivised to. That's what you get nobels for. The idea that cientific consensus is an entirely conservative force is wrong in theory and practice.
I discussed that on my blog many years ago in the context of something Nordhaus wrote:
https://daviddfriedman.blogspot.com/2014/03/through-rose-colored-glasses.html
‘Scientits’ is a keeper
So you're saying that consensuses are reliable because a very small number of scientists are incentivised to disagree with them? That's a strong argument in *favor* of markets, because all actors in a market are incentivised to make correct decisions, not just a handful with the potential to achieve winner take all prestiege.
I'm saying that consensus is not broken in the way Friedman suggests.
You can't actually replace scientific consensus with a market , because there is no way to determine winning other than consensus
The market can move before the consensus does, and this is actually a likely scenario due to the factors mentioned above around scientists being incentivized to reinforce the status quo. Markets aren't subject to the same politics and sociology.
What market? What's the market in science?
This isn't true either. One way to determine winning without consensus is to demonstrate something is true in a way that gives one control over reality, in a way that enhances one's chance at survival. The one person in a band of hunter-gatherers who understands animal behavior well enough to lay traps in the right place will eat when others starve. The few people who understand gunpowder and metalsmithing well enough to build guns will outfight the masses who stick to spears and bows.
It may even be that a minority nation that embraces the scientific method will (eventually) outlive a larger nation that eschews it, although "eschew" is a strong enough term that I cannot think of a historical example that qualifies.
Scientific consensus is not something different from empirical demonstration. Unrepeatable results (such as cold fusion) quite rightly fail to move the consensus
You're assuming scientific consensus involves only reproducible experiments.
Are you saying that we need not ascribe that label to claims made in science-y circles that aren't reproducible?
I'm open to that, since it would let us assign a more lofty connotation to the term. But it also means that we should deny that term to certain claims made in fields such as string theory, but also the claims in immunology and climate science that tend to get the most news. How are we going to correct all those accounts?
No, I am.assuming it includes reproducible experiments without the "only".
They are also taught at length that the contrarian ideas they find really good are unlikely to actually be really good, especially if they are multiple. And among those who don't learn that lesson, most become crackpots. Even if Nobels weren't limited to "one best this year", few scientists would have something that qualified for them, many just "get by", e.g. by going in detail on some small aspect and deepening its understanding without bucking the consensus.
What's the problem ? Most contrarian ideas aren't new, and aren't good.
Yes. This is a bit of an inside-outside view problem, I guess? If you already taught at length that most contrarians are wrong, you are more likely to judge yourself as wrong when you actually step on a correct contrarian idea - or, even if not wrong, to judge your idea as being unlikely to be accepted by the community (perhaps correctly, as status quo bias is powerful). So, there are certainly strong incentives for scientists to be conservative even in their research.
>because everyone has good incentives in a market
Depending on your definition of good, that’s not necessarily the case. Making money is a clear objective but is that always in alignment with making predictions in any useful sense?
The context of Scott's comment was prediction markets, where you get money if your prediction turns out to be correct. The only distorted incentive I can think of is if you believe that by betting on something happening you can make it more likely to happen through the effect of your bet on the price and of the price on other people. That could be true in is an election market, where if you make your favored candidate appear more likely to win other people will be more willing to vote for him or contribute money to his campaign, and similarly in the other direction.
Understood.
The angle I am wondering about is the dynamic of gaming a market, depending on how it’s structured. In a very liquid and efficient market a lot of profitable trades can be made that are not very predictive.
In a prediction market, practically by definition a money making decision is one that makes correct predictions.
I understand, but I don’t get how that translates into any useful tool for social policy/ war and peace planning etc. Profitable decisions in the financial markets are made and realized in milliseconds or decades.
If this structure is optimal then I don’t see what prediction markets bring to the party; we have good financial markets for signals.
I f one wishes for a market free of arbitrage then it’s not a good model.
Financial markets are only indirectly prediction markets on many specific topics, or not prediction markets at all. Financial markets are subject to all kinds of information, and the most pertinent factor or combination of factors will dictate market terms, meaning they offer only tenuous signals for many individual factors.
There's little need to prediction markets over what interest rates will be in 5 years time. But a prdiction market for the next presidential election tells us a hell of a lot more about who the next president will be than any financial market will.
>Financial markets are subject to all kinds of information, and the most pertinent factor or combination of factors will dictate market terms, meaning they offer only tenuous signals for many individual factors.
How do prediction markets differ?
>But a prdiction market for the next presidential election tells us a hell of a lot more about who the next president will be than any financial market will.
Perhaps I’m dense, but I don’t see why this would be the case.