I. Caplan verson Aaronson on feminism
A recent book by Bryan Caplan is titled Don’t Be a Feminist. Scott Aaronson thinks of himself as a feminist.
But are their actual views as far apart as one might now think? Not necessarily! Let me string together some bits from Aaronson’s 2022 post On Bryan Caplan and his new book:
Bryan defines feminism as “the view that women are generally treated less fairly than men,” […] He then rebuts feminism as he’s defined it […] potentially benign explanations for apparent unfairness towards women […]
During the Q&A, I raised what I thought was the central objection to Bryan’s relatively narrow definition of feminism. Namely that, by the standards of 150 years ago, Bryan is obviously a feminist, and so am I, and so is everyone in the room. […]
Bryan replied that >60% of Americans call themselves non-feminists in surveys. So, he asked me rhetorically, do all those Americans secretly yearn to take us back to the 19th century? […]
Reflecting about it on my walk home, I realized: actually, give or take the exact percentages, this is precisely the progressive thesis […] progressives might point to the election of Donald Trump, the failed insurrection to maintain his power, and the repeal of Roe as proof enough […] I wish we lived in a world where there was no point in calling oneself a pro-democracy anti-racist feminist and a hundred other banal and obvious things. I just don’t think that we do.
I haven’t read enough by either writer to know details of their positions, but we can try to distinguish different levels and speculate:
1) NAIVE TRADITIONAL VIEW: Our leaders are men, not women, just look at the world, that’s the natural order.
2) SOPHISTICATED DEBUNKING: But if more positions of power are held by men, or men’s salaries are higher, this is not natural, it is caused by discrimination against women.
3) LIMITED COMEBACK: While it used to be caused by that, remaining imbalances today are not — they are caused by innate psychological differences between men and women on average.
Likely Caplan can be found at level 3, in opposition to level 2; and clearly he holds level 1 to be no longer relevant for the debate. For Aaronson, by contrast, feminism means opposition to an all-too relevant level 1. For all I know he could even be a “feminist” located at level 3 as well!
The issue in contention is the relevance of level 1. My own opinion on that, since I’m starting a post with it? But what do I know . . . I’m not even American . . . and I would certainly not have guessed, after the storming of the US Capitol in January 2021, that Donald Trump could now be favourite for next US president. Hence I’m going to take this in a (very) different, (much) weirder direction (readers be forewarned).
II. Generalisation drive
The level 3 claim above, “remaining imbalances today are not [caused by discrimination] — they are caused by innate psychological differences”, has been invoked by some in relation to unequal outcomes in society more generally. The three-level structure can surface, not only in discussions about feminism, but notably about anti-racism as well.
And we can go more general still. What else is associated with the culture wars, besides unequal outcomes? Immigration, although that is not independent if the level 3 claim about psychological differences can be recycled when the naive view was “immigration is bad for the country”. More independently, trans issues:
NAIVE TRADITIONAL VIEW: There are men and women, nothing inbetween.
SOPHISTICATED DEBUNKING: This is simplistic binary thinking. The reality is much more complicated, with inbetween cases, and some people need to transition to the other side.
LIMITED COMEBACK: True in principle, but the naive traditional view fails only in exceptional cases. When it comes to the current transitioning fashion among teenagers, many will regret it later.
And two decades ago (though in retreat since), a notable culture-wars epicentre was “new atheism”.
NAIVE TRADITIONAL VIEW: Intelligent design, just look at the world, God created it.
SOPHISTICATED DEBUNKING: There is an alternative and in fact better explanation of the apparent design: Darwinian evolution by natural selection.
LIMITED COMEBACK: True, but where does that process come from? And why did it lead to ever more complexity, culminating in the human species? God still has to be behind it all in some manner.
Why is the three-level structure popping up repeatedly? Well, at this point many might find that hardly more mysterious than the simple fact that errors are a thing. If one collects important traditional errors and their corrections, then one might not be surprised that in such cases a third, defensive, level will be conjured up by those with vested interests in the old beliefs.
Yet here is an example with reversed roles, inasmuch as those located at level 2 tend to be not progressives but their critics:
NAIVE VIEW: Children’s minds are blank slates (e.g. should psychological differences between men and women really exist, they will be culturally conditioned).
SOPHISTICATED DEBUNKING: Behavioural genetics, it’s nature rather than nurture; or at least what is known as “shared environment” has little impact as long as basic standards are met.
LIMITED COMEBACK: Nature and nurture cannot actually be separated, that would itself be naive; the correct angle is a “holistic interactionism” (Steven Pinker’s term for a widespread view among scientists, one that he is not a fan of himself).
Followed by another such reversed-roles example, for which we could consider parts of science, especially of social science; but here is a wording for the media instead:
NAIVE VIEW: Trust respectable media, reject uncredentialed sources.
SOPHISTICATED DEBUNKING: The mainstream media are biased and often lie.
LIMITED COMEBACK: True, they are biased these days but they do adhere to a certain honour code, avoiding literal lies, which means they are still useful sources if you know how to apply bounded distrust.
Admittedly, in the second of these further examples, claiming that the media lie is not that “sophisticated” a debunking — still, those holding that view will think of themselves as having overcome a naive and false perspective. And in the first, the role reversal happened because the blank-slate view lends support to the sophisticated level 2 in the unequal-outcomes hierarchy, against level 3 there — still, I would count it as a level 1 “naive view” in its own right. Many people, whatever their opinions on unequal outcomes in society, regard children’s characters as more malleable than allowed for by behavioural genetics.
Now, once we really get into the groove, then in addition to 1a) Unequal Outcomes by Gender, 1b) Unequal Outcomes by Race, 2) Trans/Non-Binary, 3) Theism, 4) Children as Blank Slates, 5) Mainstream Media, we can also consider:
6) Mainstream Punditry
NAIVE VIEW: Learn from those you see on respectable media.
SOPHISTICATED DEBUNKING: Reject those pundits in favour of prediction markets and superforecasters.
LIMITED COMEBACK: But even if it is true that the mainstream pundits are not good at predicting, they still tend to be good at, let’s call it communicating explanations, and thus confer a benefit.
7) Guru-Teachers
NAIVE VIEW: You’d like to have a certain moneymaking skill? Find someone who sells you lessons.
SOPHISTICATED DEBUNKING: Those who can’t, teach.
LIMITED COMEBACK: True, a given guru-teacher may not actually be a good practitioner. But many strong poker players, for example, engage in coaching or teaching, which wouldn’t be the case if those-who-can’t-teach were a reliable principle. Note the allure of prestige that comes with teaching.
8a) Active Stocks Investing
NAIVE VIEW: What are the hot trends? Buy stocks accordingly.
SOPHISTICATED DEBUNKING: The market is efficient, don’t try to beat it.
LIMITED COMEBACK: True, be careful with hype trends. But the market is not actually efficient. As an illustration (not a proof, of course), when Scott Alexander of
(ACX) was asked in the comments to his Semaglutidonomics post whether he owns stock accordingly, he answered no, presumably “all of this is already priced in” — only for Novo Nordisk, the company behind semaglutide (the weight-loss drug), to again double its market capitalisation (by now Europe’s biggest) in the 16 months since then. Likewise for Eli Lilly, the main competitor.8b) Active Funds Investing
NAIVE VIEW: Who are the hot fund managers? Buy funds accordingly.
SOPHISTICATED DEBUNKING: Buy passive index trackers instead, because the market is efficient.
LIMITED COMEBACK: True, selecting funds by recent success can backfire. And avoid ordinary mutual funds managed by “index huggers” who will find it hard to beat their assigned index after fees simply because they stay so close to it. Once you don’t count the index huggers, however, active managers’ performance gets better. After all, the market is not efficient, we just had that.
9) Watch Your Calories
NAIVE VIEW: Calories in calories out. We get fat because we eat too much and exercise too little.
SOPHISTICATED DEBUNKING: If you think it through, this makes no sense. In fact, the causation must be reversed. Read Book I of Why We Get Fat by Gary Taubes (or maybe this).
LIMITED COMEBACK: But you can also read the aforementioned Alexander, whose ACX is among the top-ranked Substack science publications. On his previous blog, he showed he has no problem entertaining the counterintuitive point (see where he quotes David Ludwig on overeating in section II.D here), only to nonetheless tentatively blame “lots of calories with few nutrients” (a year later, here, in reply to the first comment on that page).
10) Zero-Sum World
NAIVE VIEW: If two parties agree on a contract from which one party gains, the other loses. It must have been outwitted.
SOPHISTICATED DEBUNKING: No, value can be created for both parties because their preferences can differ from each other, and the total amount of wealth is not fixed anyway — wealth can be created (or destroyed).
LIMITED COMEBACK: True, but wealth translates into status, and status is really zero-sum.
11a) Free Will
NAIVE VIEW: Naive free will.
SOPHISTICATED DEBUNKING: It’s an illusion, the world is deterministic (and no, quantum indeterminacy won’t rescue free will).
LIMITED COMEBACK: Compatibilist free will (something like this).
11b) Mind-Body Dualism
NAIVE VIEW: The soul exists separately from the body, can leave the body, survives the body after death.
SOPHISTICATED DEBUNKING: Dualism is false, only the material world exists.
LIMITED COMEBACK: True, the soul cannot leave the body as in naive afterlife beliefs, but dualism is actually the correct philosophy.
11c) Spirit Possession
NAIVE VIEW: Like in the movie The Exorcist.
SOPHISTICATED DEBUNKING: There are no spirits or demons in the head that could be “exorcised”.
LIMITED COMEBACK: Actually, rogue agents can exist in one’s head. And some people even try to grow non-standard agents there (“tulpas”). See here.
12) Artificial General Intelligence
NAIVE VIEW: AGI is with us, in the form of ChatGPT and others, even though confined in a box.
SOPHISTICTED DEBUNKING: Large language models, however impressive, are really just stochastic parrots.
LIMITED COMEBACK: True, their intelligence is more alien and less humanlike than one would intuitively think, and at the moment of this writing they lack something related to the G in AGI; but they do engage in constructing models of the world.
Or perhaps I should have left the last item out, because the idea was to collect only examples where basically everyone has a view, at least an implicit one. Can that already be said of AGI? Then again, it probably can’t be said of the stockmarket either. But we will need to be clear that I can’t just take the freedom to consider all sorts of technical or scientific controversies, however obscure. Otherwise I might be able to selectively produce many little hierarchies of whichever type I want to showcase; yet it would hardly suggest or prove anything.
Which level expresses the soundest opinions? The naive level 1 is usually the weakest for me, if only because when I didn’t find that unambiguous enough, I sometimes sharpened it suitably, e.g. to “there are men and women” I added “nothing inbetween”. The real question is, should one be a second-level or a third-level person?
Well, others would choose different wordings in various of the examples, presumably. But in my case, summing up over the list of twelve as written above, counting sub-items and trying to adjust for the strength of my opinions: it’s a victory for level 3 by a comfortable margin, something like two to one.
And . . . that’s not what was supposed to happen! If you correct a mistake, why should there so often be a comeback that has merit? Even more often, by the way, than already implied, because not all cases in which I see merit in level 3 counted for the victory. Level 2, as worded here, still seemed reasonable enough in some of them.
Apparently we have stumbled on a deep feature of the world, which could be called the surprising resilience of naive views. So it is this that is responsible for the culture wars! It causes a general naive-view stickiness that rubs off on the three naive traditional views from the beginning of my post. And there is a famous philosopher from two centuries ago, Georg Wilhelm Friedrich Hegel, whose name is associated with a pattern of thesis, antithesis and synthesis: with Hegel to the root of the culture wars!
III. Where does the third level come from
That went downhill. At the end I was basically trolling. Also, apparently it is controversial how closely Hegel’s dialectics, of which I admittedly know little myself, really follows the three-level pattern. Still, one can ask, out of intellectual curiosity, where does the pattern, as observed here, come from?
I won’t be able to offer any sort of unified theory. The items in the list of twelve are too disparate. But let’s try to investigate. The first idea might be that possibly all we have here is a selection effect? As in, by necessity I noted only those items that came to mind, they came to mind because I found them interesting, I found them interesting because they exhibited the three-level structure. For example, the inspiration for item 7 was that I noticed the unreliability of the those-who-can’t-teach principle, which considers itself the sophisticated response to a naive view (implicitly held) that I then formulated only afterwards. And in the case of item 6 (mainstream punditry), the third level tries to express something that occurred to me recently as a possibility: had that not happened, I might well just have overlooked that item, instead of adding it as a “counterexample”, a case that doesn’t feature or admit a third level.
So let’s go for counterexamples. What about astrology, homeopathy, parapsychology and UFOs. And the Bible is the literal Word of God.
Then again, belief in UFOs seems to have become somewhat more respectable lately, so that one does actually feature a third level? And more importantly, what I was looking for is naive mistakes. If labeling all the naive views in my list of twelve as “mistakes” is simplistic, so be it. The problem at least with the Bible, astrology and homeopathy, however, mistakes or not, is that I wouldn’t really call them naive. If they are associated with naivety, it would be more, I think, because of a perception that it is naive people that believe these popular and rather elaborate theories.
To explicitly state the criteria, I’m looking for widespread current or at least recent naive mistakes on issues about how the world works (not about particular facts or events without implications for how the world works), where most Western people have an opinion or at least some awareness. With these criteria, counterexamples seem surprisingly hard to come by. Parapsychology, or perhaps something like “magical influencing” more generally (this might even subsume astrology and homeopathy), is the only one so far. Why is that?
Maybe because overcorrection is natural? Some of the sophisticated debunkers of a naive view will get carried away and overshoot in the other direction. That seems plausible. So is a limited comeback no less than you would expect? Is this the boring straightforward explanation — for those items in which level 3 is stronger than level 2, combined with vested interests, already noted, for those in which it is weak?
Then again, let’s think about it, should we really expect there to be a direction to overshoot in if a mistake is corrected? Not usually, for instance, if I correct a typo in this post. Something seems to be missing still.
Meanwhile, for third levels without merit we have, to restate it officially, vested interests as a default explanation. It was the reason that my criteria above included “current or at least recent”, and “Western people”. I wanted to ensure that if defendants of the old belief have constructed a third level, then there is a good chance I have heard of it without having to study the history of ideas. For example, belief in a rain god was widespread, false, and does not seem to allow a comeback — but who knows what priests of the rain god once produced in defence.
A not very sinister version of vested-interests would be that people are just naturally conservative, trying to cling to their familiar beliefs, and chances are you can come up with some rationalisation attempt. Especially as you can choose which aspect of the naive view, or even which adjacent point, to make the theme of your comeback . . . in fact, that’s an important second selection effect at work here! Congratulations to anyone who noticed it already while reading. To me it didn’t occur until my best counterexample for a while, naive soul/afterlife/ghost beliefs, morphed into item 11b (mind-body dualism) after I remembered having seen the defence of dualism that I linked above, by someone who appears to me like a strong thinker (in such a case I find it fair to list a third level, see also items 9 and 11c).
This second selection effect, let’s call it level 3 leeway, needs to be acknowledged. But it is not all-powerful. I wouldn’t know how to formulate a third level for astrology or homeopathy. Also, where it dominates, the third level will find it hard to become stronger than the second, so level 3 leeway, unlike the first selection effect, can’t really have delivered the victory for level 3 that I arrived at earlier.
We have now collected two selection effects, overcorrection, and vested interests. Can their combination explain what is going on? No, there surely has to be something like a “true” resilience of naive views . . .
IV. Egan’s law on LessWrong
. . . indeed, the rationalist community has a concept that looked like the missing piece to me. To quote the beginning of this page on LessWrong:
“It all adds up to normality” is a common phrase used on LessWrong (also known here as Egan’s law[1]). Adding Up to Normality is the property of an explanation which adds to our understanding without changing what we already know to be true. for example:
Apples didn’t stop falling when General Relativity supplanted Newtonian mechanics.
As counterintuitive as quantum mechanics is, it all adds up to what we see in everyday life — It’s perfectly normal, and it always has been.
Apples didn’t stop falling . . . that’s a true resilience of a naive view. So is it Egan’s law, at last, what my investigation has been truly grasping at? Well, at least we can use it as inspiration for more wannabe-Hegelian hierarchies. We are led to the realm of physics, which I haven’t yet focussed on. But naive physics is less Newtonian than Aristotelian:
NAIVE VIEW: Absent a force on them, objects come to rest (Aristotle).
SOPHISTICATED DEBUNKING: Not so. They keep their velocity (Newton).
LIMITED COMEBACK: In practice, with friction everywhere, they indeed soon come to rest. Just don’t go ice skating.
Here a clarification is in order. Essentially the comeback is to limit the naive view’s status to that of useful approximation. But if we allow functional merit as the basis of a third level, could we then not also claim, for instance, that homeopathy usefully fills the role of placebo? My earlier assessment that level 3 leeway “is not all-powerful” would no longer hold. If claims of functional merit were allowed, level 3 leeway would be all-powerful; you can usually concoct some “just-so story” of how a mistake supposedly has effects that are actually beneficial. To clarify, claims of functional merit are not allowed. However, I feel that approximations warrant an exception, because they work in the service of building the best feasible model of the world, an undertaking in which simplicity is a consideration.
Aristotle seems to have been a sort of naive thinker, so I guess his philosophy could yield more examples. But this post is shaping up to be unusually long, so any more hierarchies to be written out should do something special. Like raising the post to the meta level:
NAIVE VIEW: I can see reality, I can’t see the problem.
SOPHISTICATED DEBUNKING: Much of reality is in fact hidden and counterintuitive, and our knowledge of it is always provisional, e.g. Aristotle overturned by Newton overturned by Einstein overturned by . . .
LIMITED COMEBACK: Still, Egan’s law, all of these add up to normality.
COUNTER COMEBACK! (CONDITIONS ARE MORE TURBULENT UP HERE ON THE META LEVEL): Yet Egan’s law itself constitutes merely provisional knowledge! What if we were taken out of the Matrix tomorrow?
COUNTER COUNTER COMEBACK: That’s a misunderstanding of Egan’s law. Don’t mix change into the discussion.
In this hierarchy, the level 4 counter comeback failed badly — in fact, my understanding of Egan’s law, after thinking about it, implies that already parts of level 2 were beside the point. Why? Let me repeat part of the quotation from LessWrong. Egan’s law says “it all adds up to normality”, where
Adding Up to Normality is the property of an explanation which adds to our understanding without changing what we already know to be true. for example:
Apples didn’t stop falling when General Relativity supplanted Newtonian mechanics.
Level 2, as formulated, would say that Newtonian mechanics is incorrect, with level 3 retorting that it is nonetheless a useful approximation of General Relativity for everyday purposes. However, I think that the approximation angle is something separate from Egan’s law, and that Egan’s law is instead about naive views that are completely true. After all, the wording is “what we already know to be true”. How can we really know truth? Is physics knowledge not provisional? But that’s not the point; let’s just assume we know because we observed it super-directly or something. Like apples falling. Then Egan’s law serves to remind us that any “counternormal” explanation we arrive at still has to account for that “normal” knowledge. Normal physical reality would remain the same even if we learned that we’re in the Matrix (without being taken out of it). And “Newtonian mechanics” here does not refer to that normal reality (or if it does, then I find the wording unfortunate) but to an explanation of why apples fall that is not yet counternormal.
The upshot is that Egan’s law is not what this post was trying to get at.
The approximation angle, however, seems instructive. In light of the exception above, where I talked about functional merit and said „simplicity is a consideration“, the resilience of naive views then stems from the fact that they are not, after all, cognitively so weak in the first place (see here for more). Either that, or we have potential counterexamples. For instance, we debunk the view that the earth is flat, and the third level becomes “at least it’s a useful approximation”: but those flat-earthers that are particularly wrong are exactly the ones who insist it is not an approximation. And for them, I don’t know how to formulate a third level. Which means a potential counterexample (though I wouldn’t consider that one an actual counterexample if if involves an elaborate conspiracy theory).
While the term “approximation” mostly does not quite suit the examples in my list of twelve, I think a good number of the level 1 naive views there do contain some grain of truth, which helps to explain their sustained presence in the first place, and can serve to support a limited comeback. Moreover, if midwit overcorrection throws out said grain together with the mistaken part, resulting in a dubious second level, then the third level can be stronger than the second.
Come to think of it, I immediately connected naive views’ surprising resilience to the pattern of thesis, antithesis and synthesis, so obviously they would themselves contribute useful material: the naive view becomes part of the syn-thesis. Overall, and especially after the previous paragraph, I get the distinct feeling that the “results” of this post are less than spectacular. If they are boring-obvious, then I can only hope that at least we encountered some slightly more interesting miscellaneous stuff along the way.
V. Closing examples, courtesy of orthonormal
There is a number of posts referenced on the LessWrong tag page for Egan’s law; and three examples in the first one, by the aptly named user orthonormal (who also initially set up the page), inspire an item each that I find special enough to be worth writing out — for anyone who is still not fed up with the little hierarchies. The first of these seems like a quite decent counterexample:
NAIVE VIEW: Religion is needed for morality.
SOPHISTICATED DEBUNKING: If that were true, then experiments shouldn’t really have struggled so much to establish a positive effect of religion on prosocial behaviour. For instance, the evidence is described as “currently mixed and still debated” in the introduction of this paper, titled Prosocial religions as folk-technologies of mutual policing, which looks to me like it could be the definitive scientific account of prosocial religions (though it’s still an unreviewed working paper right now).
I don’t know what the typical defence here is, or whether there is one at all that is third-level-worthy. Maybe one could note that Islam seems pretty good after all at unifying disparate tribes or clans into a cohesive military force, like the Taliban in Afghanistan or al-Shabaab in Somalia. But that’s somewhat removed from the Western world.
The second of the three items is multiverses (cf. from Eliezer Yudkowsky’s famous Sequences this post, which I must have come across once — dimly remembering it is what led me to Egan’s law on LessWrong). It invokes a bird perspective, even the canonical bird perspective:
NAIVE VIEW: One world, keep it simple, keep it sane.
SOPHISTICATED DEBUNKING: Indeed, keep it simple. From a bird perspective, by postulating the existence of other worlds, instead of accepting certain observations or parameters as arbitrary and particular, we arrive at a theory of reality that is in fact simpler.
EXASPERATED OUTCRY: Enough is enough! We started with culture wars and have now somehow managed to arrive at the most esoteric of topics, with absolutely zero practical import.
For a very limited counterpoint, see here. And finally, there is altruism — I still cannot quite understand how I overlooked this one myself when compiling my list of twelve. Now it lets me close the circle, by kind of returning to the culture wars:
NAIVE VIEW: Paying the whole group’s bill, for example, or teaching others without compensation, those are selfless things to do.
SOPHISTICATED DEBUNKING: Not so fast. Such acts accumulate prestige status (cf. item 7 above). This is good to understand especially nowadays as “wokeness” seems to be a kind of runaway prestige-status phenomenon, based on caring for the oppressed.
LIMITED COMEBACK: But the acts’ effects are still nice. And correspondingly, Western countries, even though afflicted by “wokeness”, are still nicer places overall to live than certain third-world cultures that are shaped by dominance status more than by prestige status.
Thanks for the cite!
I think you're starting a few levels in already with some of your examples. Level 1 really should be the inchoate intuition that people have before they've thought consciously about a topic (or alternately, the position that ancient societies reflexively took on the topic when it first rose to attention).
So for example, the version of the blank slate idea you start Level 1 with is ahistorical; I'm confident that most Bronze Age cultures, if they thought about it at all, believed that gendered psychological dispositions were innate. Platonism and later versions of the blank slate idea were reactions against an existing intuition.