On this day five years ago, Scott made a list of graded predictions for how the next five years would pan out. How did he do?
He correctly predicted that Democrats would win the presidency in 2020. He correctly predicted that the UK would leave the EU and that no other country would vote to leave. He seemed under the impression that Ted Cruz would rise up to take Trump's mantle, but to my mind the only person in the Republican party who has a meaningful chance of opposing Trump is DeSantis. I think a lot of the technological predictions were too optimistic (specifically the bits about space travel and self-driving vehicles) but I don't work in tech and amn't really qualified to comment.
Near the end of the article, in a self-deprecating moment, he predicts with 80% confidence that "Whatever the most important trend of the next five years is, I totally miss it". To my mind, the most significant "trend" (or "event") of the last five years was Covid, and I think he actually did okay on this front: the second-last section of the article is a section on global existential risks:
Global existential risks will hopefully not be a big part of the 2018-2023 period. If they are, it will be because somebody did something incredibly stupid or awful with infectious diseases. Even a small scare with this will provoke a massive response, which will be implemented in a panic and with all the finesse of post-9/11 America determining airport security.
- Bioengineering project kills at least five people: 20%
- …at least five thousand people: 5%
Whether you think those two predictions cames to pass naturally depends where you sit on the lab leak hypothesis.
Jump in the discussion.
No email address required.
Notes -
Zero people predicted a pandemic
About zero people predicted double-digit inflation (except for the usual people like Peter Schiff who make this prediction every year)
A lot of people predicted that the GOP would return to pre-Trump normalcy
No one predicted Dall-E, Chat GPT, etc.
No one predicted Putin making a major move
...goes to show how hard predicting is.
A pandemic is the prototypical black swan event. If you asked people what are the odds of a largely asymptomatic highly contagious coronavirus mutation that crosses over from bats/other reservoirs or is engineered in lab spreads into a worldwide pandemic over the next 30 years, I'm sure many people would have given sensible probability estimates.
Pandemics are largely excluded under force majeure clauses, but take Wimbledon, for example, that following the SARS outbreak in 2002 bought pandemic insurance under an infectious disease clause. They paid roughly 31.7mm USD over 17 years in premiums for 142 mm USD payout. Back of the napkin math (assuming that the payout would be the same every year, i.e. Wimbledon's profits don't increase corresponding to premium increases, which of course in reality they do) suggests that the risk analysts estimated that a Covid-scale pandemic happens at most every 78 years or so. Since they probably factored in losses due to more probable minor diseases, the Covid-19 pandemic was roughly a once in a century event. This passes a cursory sniff check with the last global pandemic being the 1918 Spanish flu.
Safe to say people were thinking about, and even buying insurance against, the risk pandemics posed, so your first point is wrong.
How about conditioning on the aforementioned black swan event and stipulating that US M0 practically doubled and M2 grew by about 5T USD (25% from 15T in 2020) whilst US treasury yields were kept at historic lows? The low interest low inflation free money paradigm that has dominated the past decade, under a historically axed lens, seems more like an exception than a new rule.
The GPT model was novel, sure, but also came out in 2018, so roughly 5 years ago. It was designed specifically to avoid the problems of supervised learning in NLP. What's surprising is perhaps how far it scaled?
EDIT: amusingly, we both forgot to reread Scott's post
This is basically GPT-3 and DALL-E. ChatGPT is GPT-3.5 fined-tuned with RLHF. So Scott did essentially envision this event, although maybe he didn't assign a high enough probability on it for your taste.
This has been discussed in the other threads, but again, once faced with the question, would people really ascribe that low a probability to it? Putin and the other minds in the war rooms cannot have been aiming for a protracted land war, this is more a political decapitation with a puppet substitution gone horribly wrong. If you rephrase war to political/military intervention that aims to replace the Ukrainian head of state, the odds might climb further, and now you can just ask what are the odds of that surgical strike devolving into full-scale war. The Russians were what, 100 miles from Kiev? There is an other timeline where Zelenskyy is either dead or in exile, and Ukraine is not in rubble.
Sure? It's a chaotic system, and the conditional probability trees have exponential branching factors? No one is pretending to be prescient or an augur here, the exercise is just in estimating probabilities.
The "prescient" fool confidently spouts "predictions", whilst the wise man merely scribbles down a few numbers and replies, "Wanna bet on it?"
More options
Context Copy link
Peter Zeihan in 2015:
https://zeihan.com/ukraine-just-the-beginning/
https://youtube.com/watch?v=OtdOZEgaFIw
C'mon man.
October 2016:
https://scroll.in/pulse/818562/the-next-black-swan-will-illegal-exotic-meat-cause-the-next-pandemic
Unless the fact that this focused on Africa and not China disqualifies it in your eyes.
If you were at all plugged into X-risk analysts then global pandemics, especially possibly bioengineered ones were pretty high on the list.
Nassim Taleb was warning about international travel sharpening the risk of pandemics for like a decade prior to 2020.
David Sinclair put a decently long section in his book, published 2019, about trying to mitigate global pandemic risks via biotracking.
I dunno if you mean that nobody predicted this exact sort of pandemic to arise from that particular region, but it seems obvious to me that people predicted a nontrivial chance of a pandemic arising well before on actually arose.
For what it's worth there are a lot of people who make a lot of predictions and most of them are wrong. You'd be WAY harder pressed to find an event that somebody DIDN'T predict in at least an obtuse way.
being 7 years early is as good as wrong. this comes off of the Crimea conflict.
Regarding a pandemic, I think you are right. That 2016 article was prescient.
But still, the rationalist community is really obsessed with forecasting and predicting, yet missed so many of these big things.
I have never taken Zeihan that seriously, but will conceded that with respect to Russia I think his record is actually pretty good. I mean his book explicitly stated that he thought Russia would invade sometime during the 2020’s and they did, seemingly for the reasons he thought they would (I.e Russias bad demographics and obtaining a more defensible border with Europe).
Good ?
It's just bullshit.
Moscow would have taken Ukraine perhaps whole if they could, though they'd probably have left Galicia to the Bandera fans because why invite more pain. Azerbaijan ? Armenia ? Moldova ?
Belarus is only close with Russia because the West got fed up with Luka and tried to remove him. Hence he has no choice but to cozy up to Russia.
I follow a scholar who's concerned with China and speaks the language, he doesn't even mention Zeihan. He's not taken seriously at all.
He's really a joke, in his talk on China in an interview with Rogan, 80% of what he said were obvious lies. (China can't build high tech, China can't innovate, etc)
I’d be interested in who you follow on China (I agree Zeihan is a clown on China).
Tanner Greer - @scholars_stage on twitter.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
How do you know those were reasons that actually factored into the decision? That's just Zeihan's own hobby horse, he always explains things with demographics, hell, he predicts AI being essentially a nothingburger because the world will start running out of 20-somethings in 30 years!
For all we know Putin was mainly driven by the impression that his subversion network is successful. Russian demographics are at least better than Ukrainian, and on par/perhaps better than European. Zeihan's «natural borders» story is, uh, I'm not sure it's even a sincere argument – he oscillates between treating it as a realistic consideration and some Russian neurosis from the time of Mongols and warfare dominated by low-tech infantry.
More options
Context Copy link
More options
Context Copy link
"Compared to what?"
All that's required to win the forecasting game is to be somewhat more right somewhat more often than the people you're actually competing with, while also being capable of withstanding the shocks you DON'T see coming.
I don't know of any community in particular that came across as somehow more prescient than the rationalists, and plenty who were less prescient.
Sure, if you're playing to win "the forecasting game." If you're playing to win the more general "prediction game" or the even more general game of life, other things like the latter part (being prepared to withstand shocks you don't see coming) matter a whole lot more. I think this article about it is pretty good. In summary the easiest way to be a good predicter is simply to predict that the status quo will remain the same, which will get you a very high forecasting score will rendering you unprepared for all of life's most significant events.
Rationalists are pretty great at this too, but I'd say there are other groups (top government officials, anyone in a cutting-edge tech industry, anyone in a cutthroat business such as hedge fund trading) who are as good or better.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
John Mearsheimer? https://youtube.com/watch?v=aedKFv_ABr4
More options
Context Copy link
More options
Context Copy link
Actually that last paragraph is excellent. Scott is a bit of a technological optimist. I would have bet against him on a self driving car
More options
Context Copy link
Nice post but I protest the use of “amn’t”. Yes I see why it makes sense but I still hate coming across it. I acknowledge you may discard this opinion.
I was going to comment about what a strong move it was to write in common vernacular American English and including an ain't.
And to think my mom would chastise me for saying "ain't" as a child.
More options
Context Copy link
That's Hiberno-English for you, I say if we have to put up with "ain't" you should tolerate this.
Ain't and amn't ain't words and I amn't tolerating either.
I bet y'all'd've'd more respect for these words if you'd been raised right.
More options
Context Copy link
More options
Context Copy link
Alright, I respect that
All right
More options
Context Copy link
More options
Context Copy link
Haha, in my experience, most of us across the pond from you don't really say "ain't" unless it's in jest, or trying to match a specific style.
More options
Context Copy link
More options
Context Copy link
Contraction of "am not". What's the issue
Probably just because I'm is the usual contraction, this is a special case where you can drop the second I that would have gobbled up that am - I'm not positive on whether it's even grammatically correct.
fully written out the whole statement would be
Using both tricks at once to produce "amn't" just strikes me as strange, It implies the existence of a I'mn't which is truly an abomination.
Just guessing, but it's probably not grammatical in English because a lot of how English is spoken in Ireland is down to imported Irish grammer.
'I'm not' can be one word in Irish (níl mé/nílim), so 'amn't' probably sounded natural when English was becoming the dominant language.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
I don't think there is much difference between a person predicting, and an ML algorithm predicting. In other words, start with a good training set. And fine-tune with an even better validation set.
I consider myself largely vindicated in my very early (around February/March 2020, extrapolating from Diamon Princecss data) prediction that most covid restrictions would be largely useless and will have massively outsized costs because of their damage on economic and social institutions. Thank God I stuck to primary sources for covid news, If I was getting my information through any discussion forum, I would probably been derailed.
Continuing the ML analogy, Prune liberally (and add nodes judiciously). I strongly reduced the weightage I apply to the opinions of rationalists a la Scott Alexander post covid, given their initial hypochondria.
It's Random Forests all the way down.
More options
Context Copy link
fits pretty nicely with
But if COVID was the result of gain-of-function research (which seems to me pretty likely), then yeah, maybe it's the bioengineering one.
I don't think anybody, even pro-lifers, really expected any change on Roe vs Wade. "Republicans run on doing something about it; Democrats run on 'Republicans are gonna do something about it'; nobody does anything about it no matter who wins" is how it went for the recent past. That we got the Trump Supreme Court and they did do something about it when he was out of office was a big surprise to everyone.
I certainly didn't expect it. I thought it's a bad decision (original RvW, not the reversal), but I didn't think SCOTUS would dare to revert it. I underestimated them. Since I am by nature rather skeptical and pessimistic, being wrong in that direction once in a while makes me actually feel better.
More options
Context Copy link
That's not really the full picture, from the American pro-life perspective. As a grassroots movement, it was very far from fringe: solid majority position in the Republican party, and a minority position among Democrats--until the fallout from Obamacare hit, there was a small caucus of pro-life House Democrats, though it's gone now. Prior to Dobbs, the controlling precedent was Planned Parenthood vs. Casey, which functionally replaced RvW--of note, the "Casey" that PP was suing was the pro-life Democrat Governor of Pennsylvania.
When you're talking about the Supreme Court, it's got few enough members that you might as well refer to the individuals rather than aggregating trends. Roe/Casey would have been overturned decades ago, if not for a string of Republican appointees that refused to pull the trigger--Kennedy, O'Connor, Souter, and Roberts. At least publicly, their appointing presidents (Reagan, GHWB, GWB) made common cause with the pro-life movement, and promised to appoint originalist justices, except oops...oops...oops.... Even then, each of those presidents did appoint justices who either were part of the Dobbs majority, or would have been if they'd still been on the Court (Rehnquist, Scalia, Thomas, Alito). Trump was unusual in that all of his appointees joined Dobbs, instead of just half.
So, yes, Dobbs was a "big surprise to everyone," but only in the sense that the pro-life movement had decades of experience supporting candidates that ultimately betrayed them, and expecting Trump's vetting to succeed where Reagan's, GHWB's, and GWB's did not looked like starry-eyed optimism at best.
It's especially surprising given Trump is generally terrible at vetting people. A lot of his appointments ended up undermining him, getting into very public fights with him or doing something very different from what he wanted them to do. Maybe this one time he listened to the right advice.
Overall, yes, Trump's vetting was at best a mixed bag--though I'll note that failures in this area tend to be more spectacular than successes. But in the case of judicial nominations in particular, there's more to the story.
When Trump entered the 2016 primaries, it was plainly obvious that he was--on paper--a very bad fit with the sizable chunk of the Republican base that was some combination of Southern, Evangelical, and Conservative. In a way, the obviousness of this problem was an advantage, in that it could not be ignored and required a strategy. This cohort was extremely sensitive to betrayal on judicial matters, as noted above. So Trump did two things--he publicly announced a short list for candidates to the Supreme Court, so that they could be vetted in advance, and he appointed Leonard Leo, the executive VP of the Federalist Society, to oversee the selection of nominees. While Trump made the final calls, Leo was the one preparing the short list.
Two of the three appointments to SCOTUS that Trump made were due to the death in office of Justices Scalia and Ginsberg. The third was the carefully negotiated retirement of Justice Kennedy. Prior to Kennedy's retirement, Trump nominated Neil Gorsuch--a former Kennedy clerk--to replace Scalia, and then appointed Brett Kavanaugh--another former Kennedy clerk--to replace Kennedy himself. Both were considered more conservative than Kennedy, who was often the deciding vote, though Gorsuch more than Kavanaugh.
None of Trump's Supreme Court nominees were close to Trump himself--all three were previously-established, well-respected members of the federal judiciary, with resumes to match. While all three would generally be considered somewhat more conservative than the combination of Kennedy, O'Connor, Souter, and Roberts, they are also on average more centrist than Thomas and Alito. (This is something of an oversimplification--all of the justices mentioned have independent streaks. Indeed, prior to Ginsberg's death, every single one of the conservative justices had formed a five-justice majority with the four liberals in separate cases. The liberal justices vote together more often on average, but it's not uncommon to see splits in any of the "voting blocs" that analysts might describe.)
It's true (though not very often discussed) that SCOTUS judges not always vote in partisan blocks, but as I understood where it happens is mostly technical cases (which usually constitute the bulk of the work and are sparsely covered because they are incomprehensible to a regular person and can bore one to death). For high-stakes politicized cases, unless it's something obvious, usually one can predict the split by partisan lines.
More options
Context Copy link
More options
Context Copy link
Supposedly, Trump's SC picks were from a list created by a group whose whole purpose is figuring out which people would be best for overturning RvW, but then, wasn't that also true for prior Republican SC picks?
Perhaps the highest levels of the Republican party didn't actually want Roe v Wade overturned, because they thought they'd be a dog that catches the car and would just be screwing themselves over electorally. And that changed with Trump since he just does his own thing.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Despite the 5% chance of 5,000 deaths, I would say if you squint - "If they are, it will be because somebody did something incredibly stupid or awful with infectious diseases. Even a small scare with this will provoke a massive response, which will be implemented in a panic and with all the finesse of post-9/11 America determining airport security. Along with the obvious ramifications, there will be weird consequences for censorship and the media, with some outlets discussing other kinds of biorisks and the government wanting them to stop giving people ideas. The world in which this becomes an issue before 2023 is not a very good world for very many reasons." - holds up pretty well.
Good reminder that these sorts of prediction are really hard, perhaps impossible. All in all, I would consider this to be a pretty reasonable effort at a nearly impossible thing.
More options
Context Copy link
More options
Context Copy link