(2022-03-20) Hoel Why We Stopped Making Einsteins

Erik Hoel: Why we stopped making Einsteins. (With published responses following.) I think the most depressing fact about humanity is that during the 2000s most of the world was handed essentially free access to the entirety of knowledge (WorldWideWeb) and that didn’t trigger a golden age.

If genius were just a matter of genetic ability, then in the past century, as the world’s population increased dramatically, and as mass education skyrocketed, and as racial and gender barriers came thundering down across the globe, and particularly in the last few decades as free information saturated our society, we should have seen a genius boom—an efflorescence of the best mathematicians, the greatest scientists, the most awe-inspiring artists.

this great real-world experiment has seen, not just no effect, but perhaps the exact opposite effect of a decline of genius

In “Where Have All the Great Books Gone?” Tanner Greer uses Olaf Spengler, the original chronicler of the decline of genius back in 1914, to point out our current genius downturn. [Spengler] repeatedly describes Tolstoy (d. 1910), Ibsen (d. 1906), Nietzsche (d. 1900), Hertz (d. 1894), Dostoevsky (d. 1881), Marx (d. 1883), and Maxwell (1879) as figures of defining “world-historical” importance: in other words, as working on the same plane as Plato, Archimedes, Ovid, Shakespeare, and Newton

Ten years later John Erskine would teach his course on the great works of the Western tradition—which was the granddaddy of the Columbia Common Core, the St. John’s curriculum, and the Great Books of the Western World series—and it included all of the names mentioned above as well. To this Erskine would add the names William James, Sigmund Freud, Thomas Hardy, and Charles Darwin. . .

Is there anyone who died in the last decade you could make that sort of claim for?

If you disagree, I’ll certainly admit that finding irrefutable evidence for a decline of genius is difficult—intellectual contributions are extremely hard to quantify, the definition of genius is always up for debate, and any discussion will necessarily elide all sorts of points and counterpoints. But the numbers, at least at first glance, seem to support the anecdotal.

One might, of course, reply that there are still many Albert Einsteins, they just don’t come off as Einsteins because ideas are so much harder to find now. This “ideas are getting harder to find” argument does indeed have some data to support it, although not everyone agrees

And “ideas are getting harder to find” seems especially unconvincing outside the hard sciences in domains like music or fiction.

the absence of genius is a major problem. Global cultural and intellectual exhaustion are an existential risk to the longterm viability of humanity. Geniuses prevent that from happening; they renew us, rejuvenate and reinvigorate us.

The answer must lie in education (schooling) somewhere. (That's a leap.)

Education researcher and fellow Substack writer Freddie (Fredrik deBoer) points out that. . . winning a lottery to attend a supposedly better school in Chicago makes no difference on educational outcomes.

Many have taken this null effect of schools to be a sign of genetic determinism, wherein some innate ability, like IQ, is all that matters, and education is, at best, just the delivery of a repository of facts.

I don’t think this is the case. For paradoxically there exists an agreed-upon and specific answer to the single best way to educate children, a way that has clear, obvious, and strong effects. The problem is that this answer is unacceptable

It’s an answer that was well-known historically, and is also observed by education researchers today: tutoring.

In education research this effect is sometimes called “Bloom’s 2-sigma problem” because in the 1980s the researcher Benjamin Bloom found that tutored students . . performed two standard deviations better than students who learn via conventional instructional methods

Let us call this past form aristocratic tutoring, to distinguish it from a tutor you meet in a coffeeshop to go over SAT math problems while the clock ticks down

It’s a tradition that goes back as far as one can find

If we fast forward almost 2,000 years we can find Bertrand Russell, one of the undeniable geniuses of the 20th century, who was a classic case of aristocratic tutoring—raised by his rich grandparents, he didn’t even attend school until he was 16

J. Stuart, one of Russell’s tutors, had himself been a student of Lord Kelvin (that “Kelvin”). Russell, thanks to his detailed autobiography, gives us a clear impression of what aristocratic tutoring was like. Here’s from the graphic novel adaption of Russell’s life, Logicomix:

most of the time life as a tutor was essentially a cushy patronage job, wherein you instilled a sense of intellectual discovery into a young child in return for a hefty salary that left most of your free time intact

Perhaps the clearest example in history of a genius constructed by tutoring comes from the case of John Stuart Mill.

You may have heard of the Grand Tour that young European aristocrats took part in, traveling from country to country, visiting universities and partaking of the various cultures and cuisines and sights. But did you know the young aristocrats always took their tutors along with them?

the line between parent and tutor is often a blurry one, and it is precisely in this way that the aristocratic style of tutoring sometimes bled into the lives of the non-aristocratic.

With these examples in mind, it’s likely that at a significant contributing factor for the phenomenon of genius running in families is that genius family members act as aristocratic tutors, encouraging learning, the life of the mind, and inculcating the pursuit of the higher mysteries in the young.

it turns out most of the school stuff is exaggerated or apocryphal, and Einstein had multiple tutors growing up in subjects like mathematics and philosophy, such as his uncle, Jakob Einstein, who taught him algebra

The traditional line for why essentially all intellectuals used to be aristocrats is that they were the only people with the leisure time to pursue the life of the mind. But what if it was never solely about leisure, but also a style of education that has fallen out of favor?

For the decrease in genius sure does seem to coincide with the end of the aristocracy

if I were forced at gunpoint to name the two greatest minds of the 20th century, I’d pick Bertrand Russell and John von Neumann. Is it really a coincidence that both were basically aristocrats?

Now consider our current situation. Despite all the language professing otherwise, in general the education system of the United States is based entirely on genetic determinism.

they are thrown into the school system, a competitive academic meritocracy wrapped in an obtuse hierarchical bureaucracy, a structure in which they will spend most of their young adult life, forced to learn mostly from their peers, who know as little as they do.

The first real intellectuals that most children meet in person are their college professors

Is it any surprise that such methods don’t reliably produce geniuses? Is it not anathema to how humans normally become interested in things?

Today, tutoring is seen mostly as a corrective to failures within the bureaucratic structures of eduction, like an intervention... In general, those doing well in school don’t get tutoring

Might technology swoop in to the rescue once more? Perhaps aristocratic tutoring doesn’t have to be solely for aristocrats. Recent research has shown the two-sigma effect of tutoring using AI tutors compared to traditional online courses.

Returning to now, while online courses are growing in popularity, the grand experiment that was the introduction of the internet tells us that access to information counts for little in producing genius—perhaps almost nothing. As great as YouTube math tutorials are, they themselves haven’t triggered a golden age of mathematicians.

what’s necessary for genius historically is early engagement with, not access to, intellectual subjects. And, for humans, engagement is a social phenomenon.

Could you hire a Max Talmud for your own family?

There has already been a significant rise in homeschooling and even “unschooling”

Yet, for such a start-up the problem is obvious: tutoring highlights economic privilege.

So, even if costs were reachable for the upper-middle class, would such a system be allowed to exist?

Not everything gets better with time. There are a number of things that people did better in the past, both because of lost wisdom but also simply because in the past things weren’t mass-produced. Beautiful older dresses, hand-stitched rugs, even kitchen appliances used to be sturdier and last longer.

In turning education into a system of mass production we created a superbly democratic system that made the majority of people, and the world as a whole, much better off

But we lost the most elegant and beautiful minds, those mental Stradivari, who were created via an artisanal process. (If he's correct, why don't more smart families use tutors? Maybe because you have to be really rich to afford a many-hours tutor?)

SlimeMoldTimeMold: Three Angles on Erik Hoel’s Aristocratic Tutoring

Erik Hoel, concerned that we’re not getting our fair share of geniuses, suggests that aristocratic tutoring is what’s missing:

We think Erik is right that historical tutoring was better than education today. But we don’t think being aristocratic is what made it better. So here are three other angles on the same idea:

It’s not that tutoring is good, it’s that mechanized schooling is really bad. If we got rid of formal 20th century K-12 education, and did homeschooling / unschooling / let kids work at the costco, we would get most of the benefits of tutoring.

But another possibility is that mechanized schooling is net neutral, and the problem is that we’ve lost some active ingredient that makes tutoring effective.

Education no longer includes moral instruction

Tutoring worked because tutors inspired their pupils. Modern education is a lot of things, but “inspiring” ain’t one of them.

the Manhattan Project was led by a group of strangely brilliant Hungarian scientists. Not only did most of them come from Budapest, many of them went to the same high school, and some of them had the same math teacher, László Rátz.

our point is just that these Hungarians lived in a time when high school math teachers could still inspire former students to describe them as “miraculous”. This seems to be an aspect of the educational system that we have lost.

Other things that are also inspiring / socially encouraging would work just as well — see for example the amazing progress of the speedrunning community, a bunch of teenage nerds bootstrapping a scene by inspiring each another to insane degrees of precision.

Erik hints at this by mentioning the social element

Individual attention is good, but we also think kids are good at teaching themselves. The active ingredient to us is showing kids “what serious intellectual engagement looks like”, and most kids today don’t see that until college (if ever).

The real problem is segregating children. Tutoring worked because you exposed children to people practicing a real skill.

Modern education exposes them only to teachers.

So the problem is not the lack of tutoring per se, as much as the lack of giving children any sense of the real world at all.

Teenagers dream of being youtubers and influencers. This isn’t their fault — these are some of the only professions where they actually understand what is involved.

tutoring isn’t the only way to expose children to real skills. So did working in the family business, and so did apprenticeships

Paul Graham says: As far as I can tell, the concept of the hormone-crazed teenager is coeval with suburbia. I don’t think this is a coincidence. I think teenagers are driven crazy by the life they’re made to lead. Teenage apprentices in the Renaissance were working dogs. Teenagers now are neurotic lapdogs. Their craziness is the craziness of the idle everywhere.

This may be another element common to the cases Erik mentions — most of the geniuses he names seem to have had very little contact with children outside their immediate family. Whether or not this is good for children psychologically is a separate question, but it does seem to lead to very skilled adults.

In fact, the number of children in a family might also be a factor. There was a time when most families were pretty large, so a lot of children had several older siblings.

So part of the success of tutoring might simply be exposing a child to subjects “before they are ready”, and one way to reliably do that is to have them overhear the lessons of their older siblings, who they are ready to imitate.

Scott Alexander: Contra Hoel On Aristocratic Tutoring

Erik Hoel has an interesting new essay, Why We Stopped Making Einsteins. It argues that an apparent decline in great minds is caused by the replacement of aristocratic tutoring by ordinary education.

I agree that this kind of tutoring sounds great. I wouldn’t be surprised if it has a big effect size. But it’s not the reason we have fewer geniuses

Hoel is making a stronger claim: that there are almost no geniuses today. For aristocratic tutoring to explain that, we would need for almost all past geniuses to be aristocratically tutored. But as far as I can tell, that isn’t true. Probably well below half of them were.

One other argument: Hoel cites Holden Karnofsky’s Where’s Today’s Beethoven?, which suggests that music is a typical case of the genius decline. But aristocratic tutoring in music is alive and well....it seems common across a variety of fields, especially those that aren’t taught in school and where success doesn’t make you too rich to need tutoring money

So what’s my explanation? You will not be surprised to hear it’s the maximally boring one, a combination of:

Good ideas are getting harder to find.

We’re finding lots of ideas anyway, but only by dectupling the number of researchers. More researchers means more distributed progress

More democratic norms / tall poppy syndrome . In the past, people celebrated geniuses

AI seems to have its share of geniuses: for example, people seem very impressed with Geoff Hinton

If people are still working on AI a hundred years from now, I expect them to talk about Hinton in the same way biologists talk about Darwin now

I think this is because AI is new and small(-ish)

I’ll give one even weirder example.

people like Jordan Peterson who became really famous by saying controversial things - and it asked: why aren’t there equally famous figures on the left?

The answer is: they were, we just need to look further back

the modern paradigm of minority rights began around the 60s and 70s, the first few people to operate within it got outsized acclaim, and there’s no easy way to equal them now.

Hoel Follow-up: Why we stopped making Einsteins

Do the rich still do aristocratic tutoring?

from what I vaguely know of the super rich, they mostly want to send their kids to Harvard. And college admissions dominates education so thoroughly that aristocratic tutoring has fallen out of favor, since it’s not explicitly geared toward, say, SAT test prep.

Is aristocratic tutoring possible now, somehow?

Probably the harder aspect is demand. One big problem is that aristocratic tutoring, just in terms of the logistics, often necessitates homeschooling (as it often did in the past). And homeschooling is not in vogue, particularly among the elite.

Some have pointed me to economist and education researcher Bryan Caplan as someone currently doing aristocratic tutoring with his kids (via homeschooling).

Caplan seems to be sneaking in this modern attempt at aristocratic tutoring through the cracks of the education system, saying that: "As far as I can tell, the Real World pays zero attention to what students do in middle school. The Caplan Family School won’t keep my kids out of good high schools; they can re-enter Fairfax County Public School in 9th grade.""

Objections, or, “Contra Scott Alexander”

His objections don’t strike me as particularly strong.

First, Scott attempted to give some evidence that not all historical geniuses were tutored

I’m not sure why he picked these examples, as the majority of names here were tutored. As I said in the piece, it’s actually hard to figure out if people didn’t have tutors and/or governesses

in the essay I never said that the decline of genius is due fully and completely to a lack of tutoring. Personally, I do think that “ideas are getting harder to find.” Indeed, evidence of this hypothesis is linked in the original essay. I just don’t think it explains the full decline, and aristocratic tutoring is the missing puzzle piece.

Finally, Scott also briefly suggests that there are fields in which we still do aristocratic tutoring (he names music and Chess). Scott’s evidence for the case of music is his own brother, who was tutored by a famous jazz musician, and went on to become. . . a famous musician

But Scott seems to think this is so common we should have Mozarts everywhere now, and we don’t, so the overall hypothesis is wrong. A classical musician disagreed, commenting that: "The art has indeed declined on the compositional side, and in fact private tutoring on composing is basically non existent."

Chess (and sports) are the two fields I would name wherein we still regularly do the sort of intense one-on-one tutoring we saw historically in intellectual subjects, and indeed, both chess and sports records have been being shattered decade upon decade very consistently: Magnus Carlsen is a better player than Bobby Fischer, Roger Federer is a better player than Andre Agassi, etc.

Scott Alexander: The Low-Hanging Fruit Argument: Models And Predictions. A followup to Contra Hoel On Aristocratic Tutoring:

Imagine some foragers who have just set up a new camp. The first day, they forage in the immediate vicinity of the camp, leaving the ground bare. The next day, they go a little further, and so on

Let’s add intelligence to this model. Imagine there are fruit trees scattered around, and especially tall people can pick fruits that shorter people can’t reach.

Finally, let’s add the human lifespan. At night, the wolves come out and eat anyone who hasn’t returned to camp.

This model can explain some otherwise confusing observations about the history of science

1: Early scientists should make more (and larger) discoveries than later scientists

The very first forager can walk zero hours, then forage a 100% virgin terrain

2: Early scientists should be relatively more likely to be amateurs; later scientists, professionals

lucky amateurs

3: Early scientists should make discoveries younger (on average) than later scientists

scientists should have to spend more time reaching the frontiers of knowledge before making great discoveries

physics today, where the mean age of Nobel Prize winning achievements since 1980 is 48 years old

This is generally considered to be a function of science politics, where you need a strong career network and good connections to run your own lab, and without your own lab credit for your accomplishments will go to your mentor. I haven’t done the work you would need to distinguish between these two explanations yet, although I find it suggestive that the trend is more pronounced in theoretical physics than in biology. I’ll discuss some other ways we could test this later.

4: These trends should move more slowly for the most brilliant scientists

5: These trends should fail to apply in fields of science that were impossible for previous generations to practice.

This suggests another way to test some of the hypotheses above: machine learning should have a lower age of great discoveries. Is this true? I can’t tell. When I look at people who won the top ML prizes, they seem to be older people who had a long and distinguished career in proto-ML, eg people who pioneered the theory of reinforcement learning in the 1990s. I could try to get around this, but it would feel kind of post hoc.