Bert Gunter writes:
I have no opinion on this (insufficient expertise to judge), but thought you and colleagues might find it interesting if you are not already familiar with the notion.
Couple of quotes:
“Academia is in some ways nearly ideally suited to produce the wrong kinds of expertise.”
“The British government in 2020 started a website that invites individuals to make predictions and ranks them based on accuracy; in future crises, it could consult the best forecasters.”
OK, I [Gunter] will an express an opinion on this latter—it’s crazy. A website to make predictions on the lottery would also identify such “expert predictors.” Same as stock market pundits.
The article in question is an op-ed by political commentator Richard Hanania, and the central example is the bad advice given over the past 20 years for the U.S. to send thousands of troops and spend zillions of dollars dropping bombs, flying planes, and shooting people in Afghanistan and Iraq. He writes that within the U.S., these disastrous plans were formulated by credentialed authorities within the military and that, in Afghanistan, the last president of the American-backed government “has a Ph.D. from Columbia and was even a co-author of a book titled “Fixing Failed States.” And here he is! Ouch.
Here’s what Hanania writes:
As radical as it sounds, just because someone has a Ph.D. in political science or speaks Pashto does not make that person more likely to be able to predict what is going to happen in Afghanistan than an equally intelligent person with knowledge that appears less directly relevant. Anthropology, economics and other fields may offer insight . . .
Not so fast, pal.
I googled, and it appears that the former Afghan president’s Columbia Ph.D. was in . . . anthropology! So this example does not support the claim that political science was a problem, nor does it support the claim that “anthropology, economics and other fields may offer insight.” It’s not a good sign when your example contradicts your claim.
Beyond this, though, my main problem with the proposal in the op-ed is that it seems to be replacing open expertise with closed authority. It says, “Government should set up forecasting tournaments and remove regulatory barriers to establishing prediction markets, in addition to funding them through programs like DARPA and the National Science Foundation”—and that’s fine, I like prediction markets too, we talk about them on the blog all the time—but I also recall that the Department of Defense ran a terrorism prediction market headed by war criminal John Poindexter. Having an actual terrorist running a terrorism prediction market—talk about a conflict of interest. I’m completely serious here when I say this bothers me a lot, especially when you’re using the Afghanistan/Iraq war as an example of bad judgment. It’s hard to imagine Poindexter’s organization being used to exercise policy restraint; rather, I’m guessing it would’ve just been one more tool used to justify the latest military “surge” or whatever. What next—should we put John Yoo in charge of a torture prediction market, have a prediction market in election fraud run by Ted Cruz or a criminal justice prediction market headed by Al Sharpton?
The other thing is that I feel it’s missing the point to place the blame on academia. The examples given in the op-ed are two U.S. army generals and a policy guy who, if I’m counting correctly from his wikipedia page, taught at universities for 14 years. It says that he was considered as a possible secretary general of the United Nations. These aren’t academics in the mold of, say, Samuel Huntington or Robert Putnam; they’re more like the kind of military/government/military figures who spend some time in academia to bring their real-world expertise onto campus. One of the people criticized in the op-ed was a coauthor of a book on “Fixing Failed States”: in retrospect, that does seem laughable or sad, depending on how you look at it—but certainly you can’t tag him for relying on “credentials” and “narrow forms of knowledge.” His Ph.D. in anthropology is not that much of a credential, and “Fixing Failed States” is hardly an example of a “narrow form of knowledge”—actually it sounds like the kind of interdisciplinary research that’s not narrow at all—; and I’d guess that it was his real-life experience, not his “highly specialized knowledge” or “credentials,” which got him his political support.
Summary
I agree with the main point of that op-ed not with its specifics. Or maybe I agree with the specifics but not the main point. I’m not sure.
The place I agree is that I don’t think we should let authority figures suppress debate, which is done in part through intimidation (acting like only they have expertise) and in part through establishing “facts on the ground” (in the military situation this would be actual troops or political or logistical commitments; in an intellectual debate this can involve the use of strategic contacts in the news media or academia who can push a particular agenda, as for example we saw with Fox news promoting unfounded claims of U.S. election fraud). I think the op-ed is right that the ability of authority figures to suppress and channel debate is a problem, and this is a legitimately hard problem. We have legitimate distrust of purported experts but at the same time we need real expertise.
The place I half-agree is regarding prediction markets. I agree that prediction markets are a good idea and I agree that open predictions are helpful. But I disagree with any implication that markets are the only way or even the best way to share and discuss predictions. During the 2020 election campaign, we at the Economist posted a probabilistic forecast, our friends at Fivethirtyeight.com posted their forecast, and we had some discussions and disagreements. No market was necessary. There were also prediction markets for the election, and that’s fine, but these markets had their problems too. I think it was good to have many open predictions, some market-based and some otherwise. Markets are just one aggregation mechanism, and we should be aware of their limitations.
The place I disagree with fully disagree with the op-ed is in its focus on academia as the bad guy here. Don’t get me wrong—regular readers know I hate academic pseudo-expertise, not just people like the sleep guy (who misrepresented the research literature) and the pizzagate guy (who described experiments that may never have occurred), but also people who do B.S. research like the voodoo-doll study or the ovulation-and-voting study. Some of these researchers are probably wonderful human beings, but remember that honesty and transparency are not enuf. But I digress. Yes, I have lots of problems with B.S. academic pseudo-expertise, but the problems discussed not coming from B.S. academic pseudo-expertise; they’re coming from B.S. military pseudo-expertise and B.S. global-elite pseudo-expertise. Remember that terrorist in the U.S. government who was running a terrorism futures program! And don’t get me started on B.S. rich-guy pseudo-expertise. Academia is a soft target—but in this case it’s the wrong target. I bang on this point not to defend academia but because I think it’s a major mistake to let military/government/corporate/globetrotter B.S. pseudo-expertise off the hook.
I agree that think tankers, etc. can produce nonsense just as often as academics. There are I’d say at least three reasons I focus mostly on academia
1) I came out of academia, so it’s what I know best
2) I have a sense people expect academics to be better, while saying the foreign policy blob is completely incompetent is sort of normal, as is criticism of say corporate fluff. Truth is supposed to be academics’ only job, while we know think tankers are often salesmen.
3) Finally, I do think academia is much worse if you’re judging by the standard of “what is the worst BS produced?” Think tankers, etc., can be wrong, but they’re almost never as disconnected from reality as people who write the worst things that show up on the New Real Peer Review account. To have something like the Sokal Hoax be possible indicates problems much deeper than “these people are often wrong.”
Richard:
I get that if you have an academic background, you’ll be particularly bothered by bad things going on in academia. I do a lot of work in social science, so I get particularly annoyed by bad social science. I get annoyed by the researchers on beauty and sex ratio, ovulation and voting, himmicanes, etc., even though there are lots worse things out there (as we can see from the headlines every day), because these annoying social scientists are close to me in some sense. Oddly enough, I don’t really care about stupid stuff done by postmodernists or deconstructionists or whoever it was that Sokal was mocking, I guess because I’m not particularly interested in humanities research as an institution.
The place I disagree with you is where you say, “Academia is in some ways nearly ideally suited to produce the wrong kinds of expertise”—and then you give examples where the errors were coming from non-academics.
+1 to both of you for engaging in this debate openly.
Two related points that strike me about this phenomenon (skepticism about expertise):
1) The general thread I am noticing across these diverse disciplines – postmodern literary theory, IR theories etc. – is that their are no penalties for getting things wrong. There are no clawback provisions on citations, even work that fails to replicate (in case of postmodern theory, replication isn’t even a goal) gets highly cited. Unlike a business that goes bankrupt if it fails to offer competitive goods/services, the academia-government-think tank revolving door is very forgiving of those who repeatedly get things wrong.
2) An added dynamic seems to be the self-referential / group-think that operates when confirmation bias is in play. I’m reminded of Col. David H. Hackworth’s quote about the Vietnam war: “politicians only listened to these generals, and these generals only listened to themselves.”
I note that Hanania has a Ph.D in political science, so if we’re not trusting academic experts, perhaps we should start with him?
On a less snarky note, has anyone ever been really good at predicting what was going to happen, credentialed or not? One of the enduring tropes of human history has been “well, that was a bad decision” and it occurs in all eras and all areas.
Hari Seldon
That Asimov could do an entire series about the amazing-ness of someone predicting the future shows my point rather nicely.
Greg Cochran is known for having “creepy oracular powers”.
https://slatestarcodex.com/2020/04/14/a-failure-but-not-of-prediction/
Bryan Caplan has an impressive record of making bets on future events and winning them: https://www.econlib.org/my-complete-bet-wiki/
Thanks. Looking at Caplan makes me realize how hard it is not only to predict the future, but to figure out how effectively people are doing. Caplan’s big thing is that he’s 20/20 on futures bets, which is impressive. The problem is that some of the bets are remarkably easy to win (Rand Paul won’t be elected President in 2008; the Republicans will regain at least one branch of government between 2009 and 2017 (!)) and some are harder. I find it hard to give credit to the easy ones and that makes me start thinking about how to evaluate people.
Matt:
Caplan pointed me to a page linking me to his list of bets, and I actually wrote something on it. I think it’s in a scheduled post and should appear in a few months. The short answer is that he seems to be pretty savvy about taking good betting opportunities when they come and at avoiding bets where he has a good chance of losing.
> On a less snarky note, has anyone ever been really good at predicting what was going to happen, credentialed or not?
Me.
I filled out an application to participate in Tetlock’s latest forecasting tournament (https://forum.effectivealtruism.org/posts/Dm5eNgyvEwF9ibvzj/participate-in-the-hybrid-forecasting-persuasion-tournament) about existential risk. The application required a fairly lengthy questionnaire assessing a variety of existential risks – the evaluation of applications would rest partially on how close your assessments are to various panels of “experts.” Many of the questions were of the sort: what do you think the probability is of an “extinction event” (defined as reducing the human population to 5% or less of its current level) due to various hazards (e.g., a virus, nuclear war, etc.).
What struck me in answering these questions was that it wasn’t clear to me whether I should think about what my answer would be or what I believe the “experts” would say – and those are two different things in my mine. For example, I find it hard to believe that nuclear war would result in human extinction, while I find the possibility of nuclear war considerable. Now, what do I think that people who do research about nuclear war would say? Do the researchers end up over-estimating the probability of the things they study, or does their expertise result in more accurate assessments? And, how could they possibly produce an accurate assessment of the probability of such an event (that has not happened)?
No doubt, Tetlock will get more research papers out of this effort – just the application data should produce some “publishable” work. But I was left feeling that the questions were so misspecified that I’m not sure what the value is of such research. Wouldn’t it be better to assess the likelihood of nuclear war than the likelihood of nuclear war reducing the human population by 95%?
Along the same lines were many questions about how you would allocate $X (say $100 billion) among different hazard research areas (nuclear conflict, global warming, pandemics, etc.). The problem with such questions, in my view, is that they are compound questions: first, how serious is the problem, and second, how effective is money at reducing the risk? I view these as two distinct issues and collecting information about the compound question seems less valuable than separately evaluating the two parts.
But who am I to say?
“What struck me in answering these questions was that it wasn’t clear to me whether I should think about what my answer would be or what I believe the “experts” would say”
If you’re applying to a forecasting / prediction contest, presumably you’re supposed to answer what you think.
I read Tetlock’s book. Its a great book. What he observed makes sense to me: bombastic “Big Predictors” and “experts” usually fail (Jim Cramer), while people who are more tentative and more willing to revise their views and take a broader view are better forecasters.
Tragically however in the modern world, neutrality (“climate change might be a disaster or it might be not that big of deal”) don’t draw the newspaper sales or clicks that sell adds, so their theynesses in the media – or university PR departments, or other PR people – are loathe to publish them.
If the application specifically asked what you yourself think, surely you should provide your own answer rather than guessing at what other people would say. If you think it’s unlikely nuclear war will produce human extinction, then give a probability reflecting that judgment.
I do find it odd, though, that a catastrophe reducing population by 95% is called an “extinction event.” Extinction, surely, refers to population being reduced by 100%. (Either way, I presume this tournament can’t be monetized. No one would expect to get paid after correctly predicting this sort of catastrophe.)
The problem is not (primarily) poor forecasts and bad decisions. The world is complicated enough that we should expect them. What’s upsetting is how little people learn from these experiences. That takes us to theories of ideology, etc.
I agree the main culprits are people with “real” power and not academics, but the academic world could potentially promote forecasting protocols that would aid learning. Suggestion: rather than summary, composite predictions, I’d like to see the demand for disaggregated predictions that pull apart the various strands of the issue. If the question is, what will be the result of the US using military force to prop up a client government in Afghanistan, we could reasonably ask the “experts” to break it out into the likelihood that the power of local warlords will dissolve, that public support for the Taliban will diminish to the point at which they can no longer operate, the role of the Pakistani military, etc. Then you can look at what actually happened and see which particular expectations were misguided and, hopefully, why.
Not exactly what I was expecting when I opened this blog up this morning. The fuss over the Census PES and the remarkable commentary by the AZ Republic, WSJ, NYT and the eminent senator from Illinois seem topical. I’m interested in your thoughts on that. Don’t think much about Poindexter, but surprised by terrorist remarks. One thing I’m pretty sure about is he wasn’t an army general. The Peter Principle is applicable here as opposed to the great man principle. It’s a miracle any big organization succeeds. Some succeed despite their leader. My favorite top dog was the guy who ran NYU back in the 1970s. I asked him what his management style was and he said: “I manage the same way as I did when I was a platoon leader in the Marines. I tell people what to do and then follow up to see that they did it.” He saved NYU from bankruptcy. It was a manageable job involving triage. Right up a Marine’s alley. The Census involves herding an impossibly big number of cats, yet the Bureau seems to do a pretty good job of counting us even when our leader is intent on creating chaos.
Gilligan:
That’s right, Poindexter was not an army general. He was an admiral in the navy.
Paul Meehl’s 1997 piece on credentialed persons vs credentialed knowledge is apposite:
Meehl, P. E. (1997). Credentialed persons, credentialed knowledge. Clinical Psychology: Science and Practice, 4(2), 91.
https://1drv.ms/b/s!AiU3z2ipXdAWibdHCOmvAdGx3YLNoQ
Every Meehl is a good Meehl.
Two points here. First, credentialed experts in military and government get their positions in part due to telling even higher officials just what they want to hear. If George W. Bush had surveyed all PhD’s in political science to decide whether to invade Iraq, the invasion would almost certainly never have happened.
Second, and maybe more interestingly, I don’t buy Gunter’s belief that it’s crazy to look for especially skilled predictors via a website or any other means. It’s true, of course, that some people will make exceptionally good predictions just due to blind luck. But it should be possible, if nontrivial, to identify predictors who exceed what you’d get from mere luck.
So, for example, suppose you have a thousand would-be predictors who are asked to make binary yes-or-no predictions about twenty outcomes. If the predictors are split fifty-fifty on each question, you still wouldn’t expect the luckiest of them to get all twenty questions right. If someone does get all twenty right, you can be confident that person really is an excellent predictor and not just lucky.
In the real world, of course, it will get a lot more complicated. We’re talking about probabilistic predictions, not binary yes-or-no predictions; and a lot of the probabilities will be obviously well off from fifty percent. Still, it should be possible to disentangle luck from skill after enough predictions get made.
One last point: if you’re really interested in policy, it’s conditional predictions that matter most. When you’re trying to decide between A and B, you need to consider what will happen conditional on choosing A, and conditional on choosing B. Testing for exceptionally good conditional predictors may be doable, but it seems to raise new challenges. What if one type of condition never gets met, or gets met only rarely? What if some of the would-be predictors are great at dealing with some kinds of conditions, but terrible at dealing with others? I’m not sure if much of the research on prediction markets has delved into these issues, but they seem well worth looking into.
> In the real world, of course, it will get a lot more complicated. We’re talking about probabilistic predictions, not binary yes-or-no predictions; and a lot of the probabilities will be obviously well off from fifty percent. Still, it should be possible to disentangle luck from skill after enough predictions get made.
I think the bigger issue is that any set of questions you produce will probably be correlated, and if you generate an uncorrelated set of 20 outcomes, then you’ve probably covered so broad a range of topics you can’t reasonably expect any expert to do well at all of them.
Good point, that’s definitely an issue. I think you could partly get around it by asking people to make forecasts about quantities, rather than just about whether some event will happen. E.g., you could ask them to forecast the Democratic vote-share in Wisconsin and Minnesota, rather than just asking which of those states the Democrats will carry. With enough questions of that sort, you could take account of the overall correlation.
“If someone does get all twenty right, you can be confident that person really is an excellent predictor and not just lucky.”
Uh, actually, no. You would expect that some, just by random chance, got all 20 right.
Why do you say that? The probability of getting all twenty right by random chance is roughly one in a million. (More exactly, it’s one in 1,048,576.) With a thousand would-be predictors, the probability of one of them getting that lucky is roughly one in a thousand.
Of course, it’s vital to keep track of how many people are participating in the prediction contest. The more participants you have, the higher the bar will become.
I was thinking, somewhat fuzzily, of Zhou Feng’s point, which is that it’s unlikely that all 20 questions are uncorrelated, and thus the actual chance that someone could do it randomly is significantly higher than 1 in 1 m.
*Zhou Fang, sorry.
Didn’t the 2020 election essentially invalidate the idea of prediction markets as being effective – at least in prediction of outcomes with large partisan skew? Furthermore there’s large financial incentives at work, so it would be very easy for, say, a weapons company to swamp a prediction market on war, or oil companies a prediction market on climate change. Credentialed academic experts have their flaws, but so do other systems. It’s fallacious to cherry pick a few failures and pick whatever other method. Perhaps if it wasn’t for the experts, the US would have nuked Afghanistan.
I didn’t really get that impression. I participated in one of the prediction markets myself, and made $4k on an investment of $9k. It took a lot of work–if the prediction markets had been so bad, it would have been a lot easier.
If there’s been any rigorous retrospective analysis, though, I’d definitely be curious to see it. The markets certainly weren’t perfect. One thing I noticed is that they had a clear “long-shot bias.” That is, they overstated the probability of unlikely events. Meanwhile, they understated the probability of events that lacked emotional appeal to partisans on either side. So, for example, they understated the probability that the House and Senate would be controlled by opposite parties. They also understated the probability that Biden would fall significantly short of his polls but still win, which is what actually wound up happening. (Maybe that’s what you’re thinking of when you mention the prediction markets as having been invalidated?)
Anyway, because of the correlation problem you mentioned earlier, I think it’s awfully hard to put prediction markets to the test based on a single election. Avoiding machinations by some deep-pocketed manipulator is a little easier. You just need caps on the amount of money people can put in, and you need to ensure no single individual controls a large share of the market. A sufficiently determined organization could still get around those limits, of course, but I doubt it could keep the scheme secret for long.
> I didn’t really get that impression. I participated in one of the prediction markets myself, and made $4k on an investment of $9k. It took a lot of work–if the prediction markets had been so bad, it would have been a lot easier.
Well that’s hard to quantify, but I think being able to make an almost 50% profit points to a vastly inefficient market.
> (Maybe that’s what you’re thinking of when you mention the prediction markets as having been invalidated?)
No, I am thinking of the large probability assigned to Trump winning even after the election results came in and made his victory impossible. The core assumption of the prediction markets is that putting on a dollar incentive would make participants honest, but clearly partisan skew overwhelmed that.
I guess that depends on what you mean by “large probability.” After the results in the key states were certified, the markets consistently had the probability at well under 10% as I recall. I’d characterize this as an example of the long-shot bias.
The participants were honest enough. It’s just that some of them were driven by wishful thinking.
I recall that some markets had the probability in excess of 15% even after Supreme Court judges indicated they were not going to intervene.