One of the iron rules of the world is that if you're successful in any domain you feel you have the right to an opinion in every domain. But one particular form of opinion has been recently trending in the memesphere.
It's the grand memetic challenge of having heterodox opinions by challenging conventional wisdom in fairly simplistic narrative frames.
There are hundreds of others but this is a good example, especially because Balaji does have some really good insights. But this particular point has two key characteristics.
Simple narrative - easily understood
It has a clear protagonist and antagonist
It's the Hegelian dialectic brought to life in opinion format, married to a sort of hero’s journey, perfectly sculpted to fit the shapes we like in our stories. This is a broad generalisation that sweeps away all nuance to fit a predetermined prediction that a future is inevitable.
But why this happens is still a mystery. To wit, what does success look like, if you are in the opinion business? I briefly wrote about the phenomenon in an earlier essay in the context of taking risks, but in this gig it looks like this. When you can proclaim loudly and proudly that you predicted [insert awful thing] is going to happen and when one amongst the many proclaimed predictions comes true, that adds disproportionately to your credibility.
This is true for predicting the course of coronavirus, of predicting a recession, of predicting Russia's invasion of Ukraine, of predicting high inflation based on supply chain bottlenecks, of predicting the outcome of the Yemeni civil war. Well, maybe not the last one, since it didn’t exactly hit the twitter cycles much.
These are all cases where the one correct opinion is shown as the proof of how you're prescient, how your model of the world is accurate, how you're the one worthy of being listened to.
Because in the venture capital model of opinion making it doesn't matter that you're wrong often, only that occasionally you're extraordinarily right. After all if you had one Stripe in your investing portfolio it doesn't matter how many Theranoses you had.
This is because there is a clear metric to measure your relative success in investing. It's the ROI. If you invest in 10 companies and they all do well you do very well (see Berkshire). If you invest in 10, 9 goes bust and 1 becomes Google, you also do very well. The latter is the VC model and that's what we see in the changing opinion guard.
It doesn't matter how many blunders you make in your opinions about [monetary policy, history, fiscal policy, geopolitics, covid, healthcare system] as long as you're right about one big thing. Any one big thing. If you're a goldbug but predicted Ukraine invasion in time, that's enough for credit to ride until posterity.
Exhibit A for this in recent years is perhaps Scott Adams, who both explicitly laid out this strategy, and went crazy on the success of his prediction that Trump would win. He does seem to have run it back into a ditch trying to recreate bottled lightning, but its not for lack of trying to be crazy!
But these are also cases where the counterfactual is never clear. Is the win accidental? Are they right more often than not? Are they right about the right things? Are they right about the right things the right amount?
There's no way to answer these. Because in opinion making, in punditry, there is no metric to judge outsized successes against a portfolio of failures. You fall prey to errors on both sides - of misjudging the successes as luck on the high side and misjudging the failures as predictive of the quality of your mental model on the low side.
And here enters the solution to create just such a global accounting system. Prediction markets. The attempt to convert the inchoate nature of debate into the perfectly tangible metric of capital.
Which in itself sounds great. But.
A bet is a tax on bullshit
Ever since Robin Hanson wrote about prediction markets, way too many people whom I respect intellectually have all agreed that it seems like the smartest way to get things done. They have an unwavering belief in the market, and an unwavering belief that really smart people combined with not so really smart people, armed with economic incentives, will bring all requisite information to make the market more efficient.
So far so good.
However, after about a decade or so of talking and writing about it, and more than three decades after the Iowa Electronic Markets launched, I still don't see them actually used much*.* Sure, there is the occasional example of them in the wild, within the military or some research department or a forward thinking company with a fatter than usual innovation budget, but nowhere does it actually seem to be used for something meaningful.
And I wonder why that might be the case. Because if you gave somebody in corporate world a magic wand that allows them to see the future more clearly, the chances of them not taking it up is approximately zero.
What seems to be happening here is that the theory of prediction markets never seem to be fulfilled in practice.
Now there's an answer, which is as convenient as it is ideologically consistent, that it is being held back due to regulations. Even if that is true, in places like the US, it is worth noting even in the last half a decade that there exists a multi hundred billion industry, maybe even a trillion dollar industry by now, that has sprung up trading unregulated assets, and hundred to one levered contracts. Recently one of its stalwarts, Coinbase, even went public to great fanfare!
Clearly regulations by itself did not stop it. Even Binance, which has gotten into some legal trouble recently, got to the same size as Coinbase or more before any consequences caught up with them. But I don't see a single prediction market that even comes close to even the error margins of one of these. In fact, funnily enough, one of the closest ones might be Augur, which combines the craziness of crypto and the theory of prediction market, and despite not working well enough at all seems to be posting a decent sized market cap at least.
And even if you think it's nothing but gambling, it should at least work in markets where gambling is legal, like the UK.
All of this is very strange, if disheartening. Will we ever solve the scourge that is the plague of VC style opinion making?
Hopefully, but, a few reasons that people have put forth on why we don't have an opinion utopia below.
The most valuable predictions are highly sensitive
This is Hal Varian's response, which essentially says any interesting prediction has a non trivial and privately valuable answer, which means that those with information don't want to make the answers to those topics public.
This doesn't fully make sense to me, because the question of "who will win the election" isn't all that sensitive. You don't need election officials to weigh in to figure that Donald Trump having a 13% chance of winning after the election was called was ridiculous.
They are too easy to game
This is the opposite end, that prediction markets only work for trivial things, and since "insider trading" is so easy most markets would be rigged in some insiders' favour.
The same argument works here too. Because if the insiders are rigging the presidential election market to say there's a 13% chance of Trump winning, surely others can notice this and bet accordingly, to reduce the alpha? That's how the actual equity markets work after all. If a company's CEO buys its stock, we credibly see that as a signal regarding the quality.
The costs of trading are too high
If you consider Betfair a market, they take a percentage of betting, mostly focused on Sports since that's where the enthusiasts currently are. IEM takes a $5 fee for an account opening and PredictIt takes 10% of trading profits.
These are amazing spreads, though Coinbase charging 0.5% of crypto trades and a whole bunch of other fees comes close.
The trades we're interested in aren't the ones with demand
Not surprisingly, sports betting is one of the major areas that sees high volume. The global sports betting is supposedly in the hundreds of billions of dollars, and FanDuel and DraftKings both have around $30m of bets, which seems fairly trivial.
Similarly, this is also why we don't seem to get enough people interested or betting on whether the GDP will rise by a particular percentage point, or who will get elected, or what a company's CEO will do next.
There's not enough liquidity in these markets
This is also my favourite theory, that liquidity is a quintessential wicked problem. My experience as a trader (I used to run a small hedge fund) is that liquidity really really matters. FX markets are very deep, as does NYSE and NASDAQ. But even in these markets, FX gets harder to trade if you're starting with Nigerian Naira, and the equity markets, even in that deep pool that is the US, you still have the OTC market where you would be lucky if you could sell more than $100,000 worth a day. And these are for real companies with real management teams and real products and real revenues, trading in the real market.
(And just to prove how smart I am, here's Phil Tetlock saying the opposite more than a decade ago, albeit for one particular sports betting market. He argues that liquidity doesn't improve calibration of prices and the forecasting actually gets worse with increases in liquidity, mostly because the behaviour of the traders show they aren't "rational". Whether this extends from sports to other arenas is a question for smarter people than I.)
This is a hard problem. If you start trading esoteric instruments, liquidity becomes an even more bigger issue, because you don't have the backstop any longer of the systematized and centrally mandated disclosures that equity trading requires. SEC is integral to help us figure out how to value a company and how to trade it.
The market ecology isn't diverse enough
This is another one of my favourite ones, where the market participants aren't large and diverse enough to truly capture the heterogeneity of information, and there aren't enough "true experts" participating where their knowledge differential gets captured in the form of a probabilistic price.
This problem is also simultaneous with the liquidity problem because if there isn't enough liquidity then there won't be room for more participants, and vice versa. How you kickstart it is a hard problem, albeit one with some pretty interesting technical solutions that Robin Hanson has offered. But the more complex a new offering is, the less its takeup is likely to be, especially when its actual ability to offer better insights is unknown.
So while I agree that it would be great if more people used prediction markets to hold their own beliefs accountable and help the commonwealth by contributing to the knowledge pool, I also wonder why we haven't seen it happen much.
Could we actually still use them? Perhaps. After all the only way to help calibrate individual expert opinions to know whom to pay attention to, and by how much, is to create some basic metric to judge everyone equally. To know who's playing the VC game (trying to get the one big thing right) Vs the Berkshire game (trying to get many things right).
This isn’t a choice between good and bad. They’re both extraordinarily valuable. Just as in the markets you need all risk appetites and return curves, in expert opinions too you need all possible correctness curves. If not for absolutely everyone including Gawker, at least for those who want to be taken a tad more seriously. Unfortunately for everyone its a Moloch-ian problem that makes it harder to take the first step.
The costs of being wrong might start to not get overshadowed by sheer volume too. Just like losing on every trade but making up in volume doesn't work in the markets, it should really stop working in the opinion sphere too.
Back to the original question, were we to try and create a prediction market that hits the Siskind Trilemma, we might be able to break out of the problem of opinion writers playing a venture capital game, or rather VC pundits playing the same game on twitter as in their day job.
I’m not as optimistic on its existence as yet, but at the very least recognising that most of the opinionati are playing the VC game at least can help calibrate our emotional response to seeing seemingly contrarian-for-the-sake-of-it opinions or far-out-there insane takes. And perhaps answer the question of why some lack all conviction, while the rest are full of passionate intensity.
Austin from https://manifold.markets here -- great post! At Manifold, one of our goals is to be "Twitter for prediction markets" - hence our emphasis on user-created markets, allowing anyone to set up a market for their hot take.
(aside: I don't know that "trilemma" is exactly the right word here; there's no fundamental reason that a PM platform couldn't have all three of user-created, easy-to-use, and real-money. We're currently working on several initiatives to make our fake currency more like real money, such as allowing withdraws to charities, and hosting cash-prize tournaments.)
I share your puzzlement at the observation that intellectual support for PMs far outweighs their real-world usage. Here are a few other explanations that I like:
- For the corporate use case: any single market conveys very few bits of information (just a probability from 0 to 1). This is good for letting two sides know what they disagree on, but bad for searching through the total space of ideas. E.g., at the point when I ask a question "If we implement dark mode for our site, will our DAU increase by 15% in the week afterwards?" I've already done a ton of work to narrow down the space of all possible features to ask, and all possible metrics to track. Getting a more accurate probability estimate of that question is only like 5-10% of the whole value of the question.
(Incidentally: we're trying to solve this problem with our Free Response markets! This lets someone pose a question "What features will be valuable for Manifold" and users to respond and vote on suggestions StackOverflow-style, except their votes are bets.)
See also: https://forum.effectivealtruism.org/posts/dQhjwHA7LhfE8YpYF/prediction-markets-in-the-corporate-setting
- For operationalizing Twitter disagreements: it's actually quite difficult and time-intensive to arrive at a single statement that two different sides are comfortable betting on! I believe this is called a "crux" by LessWrong folks. For example, on this public disagreement https://www.lesswrong.com/posts/sWLLdG6DWJEy3CH7n/imo-challenge-bet-with-eliezer, Eliezer and Paul apparently spent a while to arrive on a concrete disagreement here over their differing intuitions of how the AI scenario will play out, discarding a few other candidate proposals throughout the way.
Finally: I'm always happy to chat more in-depth on anything prediction-market related; reach out at firstname.lastname@example.org!
Does it make more economic sense if you replace "prediction market" with "Netflix search" and "pundit" with "recommendations Algo" ?
I think some of the drawbacks of prediction markets go away if you assume they are people-operated but the forecasters are all bots.