I.

Nostradamus was a 16th century French physician who claimed to be able to see the future.

(never trust doctors who dabble in futurology, that’s my advice)

His method was: read books of other people’s prophecies and calculate some astrological charts, until he felt like he had a pretty good idea what would happen in the future. Then write it down in the form of obscure allusions and multilingual semi-gibberish, to placate religious authorities (who apparently hated prophecies, but loved prophecies phrased as obscure allusions and multilingual semi-gibberish).

In 1559, he got his big break. During a jousting match, a count killed King Henry II of France with a lance through the visor of his helmet. Years earlier, Nostradamus had written:

The young lion will overcome the older one,
  On the field of combat in a single battle;
  He will pierce his eyes through a golden cage,
  Two wounds made one, then he dies a cruel death

The nobleman was a bit younger than the king, supposedly they both had lions on their shield (false), maybe King Henry was wearing a golden helmet (I can’t find evidence for this, but as a consolation prize please accept this picture of his amazing parade armor), and his slow agonizing death over ten days from his wounds was pretty cruel. Seems like a match, sort of. Anyway, for the next five hundred years lots of people were really into Nostradamus and spent goodness knows how many brain cycles trying to interpret his incomprehensible quatrains.

The basic Nostradamic method was:

1. Write 942 vague and incomprehensible quatrains, out of order and without any dates.

2. Whenever something happens, say “that sounds a lot like quatrain #143!” or “quatrain #558 predicted that”

3. Prophet

For example, prophecy 106:

Near the gates and within two cities
There will be two scourges the like of which was never seen,
Famine within plague, people put out by steel,
Crying to the great immortal God for relief

This is an okay match for the atomic bombs, in the sense that there were two cities where something really bad happened. But read on to prophecy 107:

Amongst several transported to the isles,
One to be born with two teeth in his mouth
They will die of famine the trees stripped,
For them a new King issues a new edict.

…and it starts to sound like he’s just kind of saying random stuff and some of it’s sticking by sheer luck.

A few prophecies sound more impressive than this, eg:

The lost thing is discovered, hidden for many centuries.
Pasteur will be celebrated almost as a god-like figure.
This is when the moon completes her great cycle,
but by other rumours he shall be dishonoured

This seems to name Pasteur, who was indeed a celebrated discoverer of things. And Nostradamus scholars note that a historian accused Pasteur of plagiarism in 1995, which is a kind of dishonorable rumor. But the work here is being done by the translator: Pasteur is just French for “pastor”, and an honest translation would have just said “the pastor will be celebrated…”, which is in tune with all his other vague allusions to things happening.

The blood of the just will be demanded of London
Burnt by fire in the year '66
The ancient Lady will fall from her high place
And many of the same sect will be killed.

Seems like a match for the London fire of 1666. But again checking the original French and the commentators, the second line is more properly “burnt by fire in 23 the 6”, which a fanciful translator rounded off to 20 * 3 + 6 = 66 and then assumed was a year. The experts say that this is really a coded reference to 23 Protestants being burned, in groups of six, during Nostradamus’ lifetime (many of his quatrains are references to past or present events, for some reason). This sounds more compatible with the “many of the same sect will be killed” ending.

I had a weird experience writing the end of this first part of the post. When I was a kid, reading through my parents’ old books, I came across an weird almanac from the 70s that had a section on Nostradamus. It listed some of his most famous prophecies, including the ones above, but also (reconstructing from memory and probably getting some things wrong, sorry):

The way of life according to Thomas More
Will give way to another more sweet and seductive
In the land of cold winds that first gave it birth
Without strife, without a war it will fall

…and the 70s almanac interpreted this as meaning Soviet communism would fall peacefully. Reading this in 1995 or whenever it was I read it, a few years after Soviet communism did fall peacefully, I was really impressed: this is the only example I know where someone used a Nostradamus quatrain to predict something before it happened.

But I searched for the exact text so I could include the correct version in this essay, and I didn’t find it - this is none of Nostradamus’ 942 prophecies! The almanac authors must have made it up, or unwittingly copied it from someone else who did.

But I remember this very clearly - the almanac was from 1970-something. So how did the faker know Russian communism would collapse?

The moral of the story is: just because Nostradamus wasn’t a real prophet, doesn’t mean nobody else is.

II.

Francis Fukuyama was a political scientist who wrote a book saying nothing would ever happen.

No! Sorry! Not that! That’s the popular misinterpretation! In 1992 he wrote a book called The End Of History And The Last Man, arguing that liberal democracy was the last form of government we would ever need, everything would get more and more liberal democracy over time, and history - in the sense of a constant progression of paradigms and worldviews and political-economic systems - would settle down and stop, in favor of everything just being liberal democracy all the time.

How has this theory fared since 1992?

I think mediocre. On the one hand, nobody has a super-compelling alternative to liberal democracy yet. On the other, China has done better than expected at maintaining its autocracy and uniting it with economic prosperity, other dictatorships have muddled along, and there’s been democratic backsliding in a few countries like Turkey. I would give this prediction maybe a C-.

But I don’t think Fukuyama feels like someone who’s gotten a C-. There is a steady drip of “this proves Fukuyama was more wrong than anyone has been before” takes, which show no sign of running out. The worst was just after 9-11, during the War On Terror, when people were panicking about “the rise of Islamofascism”. See for instance The End Of The End Of History, From The End Of History To The Clash Of Civilizations, The War On Terror: The Retreat Of Liberal Democracy, and many more. Fukuyama himself wrote in October 2001 that “A stream of commentators have been asserting that the tragedy of September 11 proves that I was utterly wrong to have said more than a decade ago that we had reached the end of history”.

With the benefit of hindsight, everything about 9/11 and the War On Terror was a random blip in history with no broader implications. There was not a rising Islamofascism, there was not a clash of civilizations. There were a few guys in some caves doing terrorism, they got lucky once, the US got angry and invaded a few countries, and then everything continued as before. If people were ranking threats to the world order now, Islam and terrorism wouldn’t make the top twenty.

Not that this has done Fukuyama any good. Any time anything happened over the past thirty years, the headlines have been “FUKUYAMA PROVEN WRONG”. The particulars have changed with the crisis du jour - most recently it’s Putin’s Ukrainian Actions Show “The End Of History” Is A Myth, The Russia-Ukraine War Shows History Did Not End, and Ukraine: A Restart Of History. But Twitter-search “end of history” at any time, and you’ll see the Fukuyama-bashing still going strong:

I think of Fukuyama as a sort of anti-Nostradamus. Nostradamus said some meaningless vapid stuff in a way such that everyone insists on interpreting as him being a genius; every time something new happens, it always proves Nostradamus right. Fukuyama said some (no offense) kind of vapid stuff in a way such that everyone insists on interpreting as him being a fool; every time something new happens, it always proves Fukuyama wrong. It’s hard to imagine what series of events could ever debunk the former or vindicate the latter.

At best, you could imagine a century or two where a careful analysis shows events didn’t match Nostradamus’ prophecies better than chance, or where a study finds that liberal democracy is overall rising and upheavals are getting rarer. But this is a bad match for modern attention spans and demands for timely content.

III.

Where does everyone else fall on the Nostradamus - Fukuyama spectrum?

Nassim Taleb has it really good. Every time something weird happens, people write articles saying “If only we had listened to Nassim Taleb, the man who tried to warn us that weird things sometimes happen!”

Of course, this is as much an oversimplification of Taleb’s book as “nothing ever happens” is of Fukuyama’s. In fact, I give Taleb huge credit for trying to fight it. When the coronavirus started, there were so many articles calling it a “black swan” that Taleb pushed back, saying that no, pandemics were pretty predictable - just totally ordinary white swans.

Gary Marcus is doomed. I’m sorry. He has generally been very nice to me, and I am not insulting him, or even calling him wrong. I am just saying he’s doomed. For better or worse, people interpret him as saying that AI won’t have lots of crazy huge advances soon. Now, I think AI will have lots of crazy huge advances soon. But suppose he’s right and I’m wrong. Suppose that of 50 possible crazy huge advances that people are predicting in the next ten years, only one materializes. That should be a victory for Marcus. But in fact what will happen is that when that one materializes, people will shake their heads and say “That Gary Marcus guy’s takes really didn’t age well, it seems naive and ostrich-head-in-the-sand-y to keep denying the power of AI when we’re dealing with $THE_ONE_THING_THAT_MATERIALIZED”. When he argues that - come on, 49/50 of my predictions came true! - everyone will call it “cope”.

There’s an even worse problem. He is arguing that the media is hyping AI advances so much that people are getting overly excited about very minor things. Crucially, he argues they’re doing this successfully - this is why he needs to push back against them. If he’s right about everything, then in the future, we can expect the media to continue to successfully hype AI advances. Every time they succeed will be another chance for people to say “That Gary Marcus guy sure looks like a fool now - he said AI was just media hype, but I just heard on the media today that actually it’s turning out to be a really big deal”.

It sounds stupid when I phrase it like that, but I have seen this happen so many times you wouldn’t believe it. Our entire civilization is built on this happening, again and again and again.

I started feeling a deep kinship with Francis Fukuyama a few years ago.

Around when Trump was elected, I wrote a post saying that the fear that he would be an “open white supremacist” who would destroy the lives of minorities were silly. I said he talked in ways that were offensive to modern liberal sensibilities but that I expected his policies on race to be about the same as any other Republican president’s (rather than, say, putting minorities in concentration camps - which people were literally saying he might do!)

I tried as hard as I could to emphasize I was not saying that Trump wouldn’t say offensive things, that he wasn’t bad in lots of other ways, etc, etc. It didn’t help. Every time Trump did something bad, especially something bad related to race, people would tweet “Looks like Scott’s take really aged poorly” or just link my article with “life comes at you fast”. Every time Trump did something, I would get some emails like “Obviously you have beclowned yourself beyond redemption, there’s no way you can ever recover your destroyed reputation, I am unsubscribing from your blog”.

I don’t want to re-litigate how right or wrong I was (no, really, I don’t want to do that). I will just say that I was left with a strong desire to never end up in that situation again. And part of that process was figuring out the borders of “that situation”. This post is the result. It’s not about how to make good predictions. It’s about how to make predictions that don’t make you miserable and cost you lots of credibility.

IV.

First, be careful about predictions that sounds like they’re claiming things won’t happen, or won’t be interesting, or won’t change very much. Suppose it is true, and in ten whole years, the most spectacular version of the thing only happens once or twice. Every day the thing doesn’t happen, nobody will write “NEWS FLASH: LOCAL PROPHET WAS RIGHT!” But the one or two times the thing does happen spectacularly, everyone will be focusing on it and obsessing over it, people who write takes will be looking on takes about it, and “NEWS FLASH: LOCAL PROPHET PROVEN WRONG” will be extremely available.

Never say we’ll be seeing declining terrorist attacks, or school shootings, or serial murders. Never say a certain technology won’t advance very much. Never say it will be a quiet year, or quiet month, even in a certain domain (“it’ll be a quiet year for Asian geopolitics”). Never say something has reached a final form and won’t change very much.

Second, be careful about predictions that sound like something or somebody will be good, or go well. I realize this is a weird asymmetry - why not be equally careful about saying something will be bad? Partly because we live in a pessimistic age. Partly because when things are going well, we don’t hear about them, but when things are going badly, we hear about them a lot.

But partly it’s because the world is so much worse than we’re used to, that it’s hard not to wilt when confronted with the badness of even something quite a bit better than normal. I think Obama was a better-than-average president. But if someone predicted “Oh yeah, Obama, great guy” - well, I mean, he did drone-bomb a lot of children. He failed to close Guantanamo, failed to end mass incarceration, failed to make significant progress on the climate, and had an important journalistic organization call his administration the most closed, anti-transparency, and hostile to investigation presidency since Nixon. By his own admission, he messed up his handling of the Libyan revolt so badly that the country is now controlled by warlords and re-instituting slavery. I’m not trying to attack Obama here! Probably some of these things are exaggerations others aren’t really his fault. This is my point! It sounds kind of hollow when I say that! Even good presidents will have a lot of things going on that are so bad that it feels cruel or hollow to say they were good.

I think this is part of what happened to Steven Pinker. There are some good arguments against some of his points in Better Angels. But the argument I see most real people making is “it feels offensive to say that things are getting better when bad things still happen”.

Third, be careful saying that certain categories of very extreme thing won’t happen. People who want to argue with you will rules-lawyer the definition of those words until they do. For example, it might have seemed like a fair prediction in 2000 that Britain would not commit genocide in the next twenty years. Genocide is an extreme event that usually happens in dysfunctional countries like Cambodia or Rwanda, and Britain is a developed liberal democracy. But you would have to deal with the people who think Britain committed genocide by being in the Iraq War, that Boris Johnson’s COVID policy constituted genocide (an argument by a professional genocide scholar, no less). Some people argue that various forms of opposition to gay or trans people constitute genocide, in the sense that it keeps them in the closet and prevents them from being a unique culture/population - and isn’t destroying unique cultures/populations genocide by definition? Unless you want all these people using you as a foil and example of a regime dupe, be very careful to clarify exactly what you’re saying (“I don’t think the British government will violently kill tens of thousands of members of its own population” might be safer than “I don’t think Britain will commit genocide”).

It might have once seemed safe to say fascists or communists wouldn’t seize power in the US any time soon - or that if they did, there would at least be violent resistance. But millions of people are very dedicated to the proposition that Obama was a communist or that Trump was a fascist. And even though you might think you have a clear idea of what something is - isn’t genocide where they put people in gas chambers? - usually the dictionary definition is poorly-fleshed out and much weaker, plus people can just invent their own definitions and insist that they’re the real one (ie “racism is prejudice plus power”).

Fourth, be extra careful about predictions of the form “X isn’t really that bad/important, the media is just hyping people into worrying about it”. If you’re right, then media is good at spinning the news to make it look like worrying things have happened. They’ll continue to do that, it will look to everyone else like even more worrying things are happening now than ever, and then all of those viewers will yell at you personally to tell you how wrong you were.

There have been some strong and well-thought-out responses to my piece on ivermectin which I really appreciate. But also, every time someone does another crappy low-standards observational study showing that ivermectin cures COVID with 90% or 100% success rate, people act like this is devastating for the people who rejected previous crappy ivermectin studies.

This wasn’t too unpleasant for me - I was taking the mainstream side of this debate, and being in the majority has nice psychological benefits. But sometimes I’ve had this same thing happen when I’m in the minority and it’s NYT and WaPo instead of The Epoch Times and Truth_Seeker, and then it really sucks.

V.

The science of forecasting is about how to abstract away all these problems and judge forecasters on the truth or falsehood of their statements.

But this science is still in its infancy, and most people don’t use it. In real life, people are judged on their predictions the way Nostradamus and Fukuyama get judged: by people taking little potshots at them every time something happens or doesn’t happen, then adjusting their reputation accordingly.

I’m writing this post mostly because I want to inch my way back into making some predictions that might have this failure mode, and I want to link this at the top to see if maybe it saves me some aggravation.

But also: if you want to become a thought leader in some field - and I think this is a reasonable ambition, some people reading are trying to build reach and influence in AI or biosecurity or land policy or various other important areas - then this is the game you’ll have to play. It’s going to be terrible no matter what, but there are ways to lessen the damage.