Selection Bias in the Quarry

towards a positive view of politics

(epistemic status: really, overly vague attempt to connect a few things. doesn’t really succeed. you should read this instead.)

I

So you’ve managed to Solve The Problem.

I assume the grand theory has data and pie-charts, likely a bunch of Fraternity-Sorority mixers, maybe a prediction or two. I’m already sold, give me thirty solutions yesterday. Still, your friend reads the manifesto and spins around like the exorcist girl. Only when he starts cackling “God is dead” do you recall that he dropped Phil 101 two weeks into last semester and hasn’t been the same since. Just your luck; in the quest to formalize humanity (working title: “Melpomenometrics“) you ran into a “philosopher” who “theorizes” by calling himself a “Nietzschean” (working title: “Foucault”).

“But are people rational?”

Philosophy starts debating the nature of human consciousness. Sociology was already  Exploding Brain, but now it’s ascended. Economics hides and starts repeating the Invisibly Handed Lord’s prayer (just the word “marginal” over and over).

To the first approximation: no.

People put sunscreen on their eyeballs to guard against the eclipse. Is the claim that those same people consume products with full knowledge of their influence on a demand curve? “Yes, but those are marginal eyeballs.” What a helpful way to think about it, really opens my margins.

Sticking just with econ: Maybe people have bounded rationality and are limited by their information (obvious but trivial). Maybe everyone is naturally greedy and zero-summing their way to the truth (less obviously true). Maybe humans are just utility-calculators with digestive tracts (doubtful). Any of these could be right, presumably something is right, nihil ex nihilo and all that.

I understand why people respond this way, but it’s the wrong response. It’s responding to what the question thinks it’s asking, and the question isn’t actually asking what it thinks it’s asking.

Sorry, too convoluted, let me rephrase that: humans are rocks, which is why we have reason and free will.

II

Schopenhauer:

Spinoza (Epist. 62) says that if a stone projected through the air had consciousness, it would imagine it was flying of its own will. I add merely that the stone would be right.

The stone thinks that it causes its own path. The stone is right. What it’s wrong about is much more fundamental: the fact that it chose this path does not mean it can choose any. It elected its own path, but it did not elect whatever allows for that path. Free will may be accidentally correct. You can select your behavior but who selects the selectors?

Imagine you wanted to survey eclipse watching habits over a lifetime. The first viewing is going to be pretty bad, see above. The second will be better. People who lather their eyeballs tend to lose them, which means they won’t be part of Round 2. Of those who can watch the eclipse and choose to watch the eclipse, suddenly things start to look a lot more “rational”. A few more trials and we have some real reasonable, salt of the earth behavior.

This doesn’t have to be conscious, and it says nothing about human cognition. Let’s say that 10% of the population always wears eclipse glasses, and 90% flips a coin: heads is glasses, tails is sunscreen-on-your-retina. Hell, it could be a weighted coin – say 1:4 in favor of sunscreen. A heavier weight in favor of “blindness” will just accelerate the selection process – 80% of those tossing are gone after the first round, which leaves whoever wanted glasses and whatever lucky few got heads. Trial 1 N=100, and T2= 18 gamblers and 10 four-eyes. Everyone else got blasted. T3 10,4; T4 10,<1.”Rational” behavior wins independent of rationality. To make it truly random, have that clever 10% flip a different coin.

“Are people rational?” Maybe, probably not, depends on domain knowledge. But we can assume they’ll act rationally sometimes. Whoever doesn’t is knocked out of competition, they aren’t part of any possible pool for analysis.

III

“Are you seriously writing a paean to survivorship bias?” Yesno, I’m getting to that, the point is that this is a double-edged sword.

There are two “rationalities” here: 1) rationality as human cognition and its application. 2) rationality as a behavioral pattern. Maybe the second is “pragmatism” or [other school here], but if rationality requires acting with total awareness then we won’t be rational until someone solves science.

Survivor bias is pernicious, and it’s what happens with (1), but behavior isn’t necessarily asking for that. Selection (the thing rather than the bias) is inevitable with (1), and that’s fantastic for human society. If you wanted a single word for the rise and demise of society, selection is about as good as you get. It doesn’t explain everything, but the others track only ascent (coordination) or descent (wormwood). You’re writing a grand theory, which means graphs, which means you want the full coordinate plane. Whether the human equation is x²-1 or -x²+1 depends on your view of original sin, but only selection will get you all four quadrants. First reason: We can try to predict and talk about reasonable human behavior under the assumption it will happen. Second: selection makes it happen. Third: we learn.

The bias itself might be a good thing (seriously qualified statement), inasmuch as it makes us hold fast to whatever doesn’t kill us. If [action] is good for survival, any group x that does [action] is going to outcompete group y that doesn’t, but it has to be dumb enough to retain the non-obvious things. Reread that sentence, I meant what I said. If it passes the tradition for a long enough period of time then this gets tinkered with, fine tuned, etc. but note that it’s not necessarily getting more-or-less-true. Maybe the reasons [action] is thought to work are lunacy. The fine tuning only has to be behavioral, i.e. whatever works best. In other words: metis, and all the weird beliefs implied therein. If I wanted to get unbearably meta, I’d point out that awareness of survival bias itself can have pernicious effects. Some group does a local custom or two and tries to explain it. We’re all too smart to believe it because the classic example of survivor bias is the folk tradition. Not coincidentally, the folk tradition is also the classic example of local knowledge, and so we ignore it and everything falls to pieces.

This is all very obvious and I’ve said it before. Sorry, housecleaning is in order, and the more abstract principle is necessary for what comes later: if a group does this One Weird Trick really zealously, and it’s old enough to have made it through a whole lot of selection, then you should probably pay attention.

IV

This sounds like an argument for empiricism, but that’s Schopenhauer’s stone. It wants its knowledge to have selected for itself, but it can’t do that. It could be a Baconian stone that trial-and-errored its way to the best fall, I suppose. But it could also have been born with some inborn knowledge of physics, a priori understanding, whatever; Schopenhauer was a Kantian, after all. Finally, the stone may have literally no idea what it’s doing, but any stones that don’t obey the laws of physics bloop out of existence (or something, I dunno, stones don’t die). If we wanted to understand a stone’s trajectory, these would look the same, because all non-force-law stones are selected out of existence. We wouldn’t know that they existed enough to correct our bias by studying them. I’m not saying we have any these studies, nor am I saying that natural=good, nor that efficiency=good, I’m just saying that in principle we can talk about optimal behavior. We can even see it: it was already optimized for in theory.

Some Schopenhauer scholar is going to point out that I’m mangling his argument. That’s fair. I am. Schopenhauer thinks the stone’s free will aligns with a deterministic universe because the stone’s nature makes will a certain way. The same argument leads him to despise humanity: think of everything brutal, ruthless, and horrifying that we’ve done as a species, and then consider what it means to select for the people best at that. This is more or less the gigantic caveat: It’s wonderful to be able to guess at optimal behavior, and learning is great and all, but you have to know what is being selected for.

This is the other edge of that sword.

V

We do actually want to know why. We’re remarkably bad at distinguishing between “act” and “know”, which is why Foucault up there thinks it’s important whether or not [group] knows what to do in order to do it. This is also how the anthropic principle manages to be both tautology and soteriology, a conceptual oscillation pattern I didn’t even think was possible. Still: “how” people reason seems a little important.

Return to the sunscreen men. 10% are always going to wear glasses and 90% will flip a coin. People are weeded out pretty fast, but majority optimal behavior would be faster if Science went around slapping eclipse glasses down. “Hooray for science” indeed, but optimal behavior would be quicker if someone held a gun to their heads, or if you raised them thinking that the sun extracts souls from the naked eye, or if [other unreasonable thing]. This is, of course, where selection effects get so pernicious. All of your sample survived, so presumably something made them survive. But…. well, what? Everyone has a different story, everyone has a cause, without further study I may as well just assume “everything”.

We might try to backtrack from optimal behavior to rationality, and I even suggest doing that from time to time. They’ll presumably look the same. Still: what if we actually don’t know what’s being selected for?

“Without further study.” Selection bias is worst when selection effects haven’t been around long enough to whittle down your reasons or your samples. When things are new, slick, modern.

VI

I’ve written a fair amount about metis. I think it’s important, not least because it’s nice to know that people do things for a reason. I’ll even say that I largely side with it, inasmuch as I’m pretty sure that myths aren’t merely “useful” for agriculture, and it’s probably wise to be cautious about uprooting the institutions that helped us to live. Unfortunately, tenses are important and “helped” is simple past. Those traditions are gone, they committed suicide, there’s no resurrection coming.

We beat the whole “subsistence” and “bare bones survival” thing. Genetic load alone tells you this, but holy hell stop screaming about either eu- or dysgenics. You get that this is good, right? We selected ourselves right out of nature’s teeth. We mutinied against life itself and we won.

If natural selection were a film that would be the climax, but this isn’t a film. Make the martinet captain walk the plank all you want, credits don’t roll in with the tide that drags him out. You have to run a ship now. How’s that gone?

Drop the metaphor: our competition is other people over different things. Consumer goods, sense of purpose, eschatological satiation, whatever. Those are how we select best behavior now. All of that is great and wonderful and it leads to a lot of naive sentiments about the glories of competition. I’ll even buy that: I’m allowed to be naive once or twice. The problem is that people are efficient, and our competition is other people, which means efficiency is not merely more-efficient-than-the-wilderness. Civilizational competition is more-efficient-than-whatever-was-more-efficient-than-the-wilderness, which means everything accelerates at an insane pace, which means new problems to overcome without the old traditions, which means there is no going back.

I’m pretty snide to [whatever we’re calling post-structuralism this week], so let me give credit where credit is due: the Critical Thinkers of the world intuitively grasped this. High Modernism claims that there’s one way to understand cognition, and that way is High Modern Rationality. Then you start poking around and realizing that everyone has a different meta-narrative. Of course, High Modernism allows that people might just be dumb and wrong (it assumes it), but a whole bunch of these seem to work. This is a serious problem for the High Modernist position.

The reason these all work is selection. The laws of physics don’t care what the stone thinks. It can think anything so long as it results in admissible behavior. It seems reasonable to pick whatever meta-narrative works best at the moment, whichever sounds coolest, whichever definitely supports your ideological presuppositions. Why not? Any of it is fine. Go even further and you find that rigorous, scientific rationality actually made everything worse, and so of course you turn on that. You may use modern science when the using is right, but so what? Paul Feyerabend lets the savage Nacerima mutilate his mouth at the yearly checkup because it’s part of the context. If anything goes it’s just another anything.

There’s a reason post-structuralism is so popular at this moment. It preys on selection bias, it obsessively points out selection bias, and we’re in the position of that guy trying to figure out which survivorship bias is the right one when there actually aren’t other trials to check against. Modernity – technology, politics, culture, whatever – means that certain things survive, and other things don’t survive, and we have no idea why. All this is new, pick and choose your favorite flavor, “you do you man, everything works.

“Selection bias is worst when selection effects haven’t been around long enough to whittle down your reasons or your samples.”

So, yeah, like right now.

VII

I’m not saying that “truth is relative” nor that it’s mere utility. I’m saying the opposite. Truth is good. It’s also flabby and weak and it needs defenders, so stop being flabby and weak yourself when it comes time to defend.

“Who selects the selectors?” Frankly, I have no idea, but apparently it’s us, and what we call “politics” is that process. It determines what pressures are in place. Competition is between humans, which means selection is about intracommunal relations. We evolved to politic, and then those different politics started selecting for behavior. Of course they would: we select for what we like and not in markets, right? How much more do we do that when it comes to behavior?

The most interesting parts of Polanyi and Scott and [others] are about this. Those old traditions had purpose. They were best principles based on given social arrangements. But arrangements have changed. Even the very worst strawman of a biodeterminist (“actually, there is no difference between a Scythian warrior and a modern accountant”) ought to take this seriously. Those bronze-age traditions were about keeping our natures in check. If it’s the same nature, it’s not obvious that those are powerful enough to hold back human nature plus all the power of modern technology. Sure, maybe nothing can. Feel free to sit down and die.

Of course, we’ve changed, too. Metis was selected for, but eventually it starts to select itself. That’s via other people, yes. Everything evolves, and we’re notably a part of everything. Culture is new, political conditions are new, technology is new. This selects for different behavior. What’s the opposite view? “Actually, behavior is acausal.” Got it. I know you don’t mean this, but doesn’t that imply that the market accepts zero input? “Demand curves are acausal, too.” Traditions are gone and/or useless, but it’s not like competition went away, so the next question you should ask is: what are we selecting for?

What [strawman pomo] says isn’t wrong, it’s just insane. If selection is done by humans to humans, then [metanarrative] exists because it’s allowed to, but that doesn’t mean it’s good. I’m not saying it’s bad either, just that metis itself was sufficiently non-obvious that most people didn’t even recognize it was best practice. Whatever we have now is even more complex, so I’m not going to judge by mere appearance. But at the risk of sounding obvious: you do get that it’s not just surface-level policy, right? If all our competitive pressures are intercommunal, then all of your political pressures are going to be changing the people under them. “Duh.” Oh, good, so we agree, because I thought we’d spent the past half-century completely ignoring this fact.

The normal way we politic is to look at platforms, ideologies, beliefs, and analyzing those. That’s fine, I do the same thing, we all read Vox. The point is that whatever those select for is non-obvious, in exactly the same way that it’s non-obvious which metis works and which is just a fringe superstition. Actually, sorry, like-that-but-worse, because this “being new” means that we don’t have any survivors to fall back on. As in: I can more-or-less assume that there’s a reason for most local traditions, and I have zero reason to assume that now.

Which leaves us with: “Do what makes you happy.” Yes, so I’ve heard, and I get that I sound like a total bore to say: actually, there are responsibilities. But actually, there are responsibilities. Enough people do what makes them happy and it starts selecting for the people who provide that, and humans are not as obvious as their stated wants would make it seem. Take an example: The Uruk Machine outcompeted everything else. It’s efficient. It, in turn, selects for certain behaviors that make it run. This is hyper efficient, and don’t mistake efficiency for “good”. This is the mechanism behind escaping-from-nature’s-jaws, which is so nice that the Nice! gets an exclamation mark. But it’s also the mechanism behind mass movements. There’s no orthographic convention for this, so let the symbol ~ mean “likely to destroy all life on earth”, and we’ll call this Bad~

Look, I like civilization. I’m just saying that things have consequences, and you don’t get to pick one thing without the other. Yes, I know how obvious that is. I’m repeating it because everyone nods and then goes back to ignoring just how scary that is.

VIII

“We must imagine Sisyphus happy.” Technically, we must imagine the stone as happy, but same difference. That’s the problem.

If the stone gets to choose its own pressures, then it’s already going to be choosing whatever pressures allows it to thrive. But what if that stone sucks? What if it’s a monstrous stone? Or the opposite: what if there are better stone-throws possible, and we literally cannot understand what they are because we select against them. I shouldn’t have to say this, but I do: Anyone who takes that as an argument for social engineering has failed to grasp the underlying problem, which is why every social-engineering-experiment has turned out catastrophic. Turns out it’s kind of hard to just “make people better”. Unfortunately, it’s incredibly easy to make them worse. Whoops.

Even if people want me to be saying [extremely unpleasant thing], my own politics are naive as hell. “I think freedom and truth are good things,” sorry to disappoint. But if you ignore threats, how are you going to defend that? Do you have a plan, or is it just to proclaim the fruits of liberty as loudly as possible?

There’s actually hope here. Sort of. I won’t say how because I want to see if anyone else can derive it, but this is meant as groundwork for a genuinely positive politics. Hell, I haven’t even proved half the stuff here. I’ve just said it. I’m trying to start from first principles to let you know where I stand. Next time we work out the rest.

Aristotle: “Man is a rational animal.” Also Aristotle: “Man is a political animal.” This is a great first principle. Yes, I meant the singular. It’s the same statement. You don’t get one without the other, they’re interconnected; politics is a selection method for human reason and human behavior.

I’m personally interested in epistemology. So I talk about politics.

IX

I get it, you’re a realist, politics are cutthroat, you’ll just Machiavel the issues away. I agree with that, actually. I’m just saying not to cut a throat if you can’t pay the wergeld. Even Beowulf knew that much, and Beowulf is a fictional character. “Then you’re a moralist.” But I’m not making a moral argument. The entire cosmos is trying to kill us, including every other fold of your best friend’s brain. Do you think the only blowback it has to offer is a stern scolding?

Say it, “If nothing is true, then everything is permitted.” As new as everything is, that’s actually quite an old problem. Jesus surpassed the law. He claimed that metic traditions were dead, that objective morality meant nothing to sons of God, that we were in a new and rudderless world. Pretty reasonably, a whole lot of people went absolutely buck-fucking-wild. Little known fact is that early Christians were regarded as total hedonists; how justified the caricature was is a scholar’s debate, I’m just guessing “at least a little.”

This led to doctrinal problems. Theologically everything is allowed, right? We’ve been saved. But it’s also kind of a bad idea to do everything. One of the very early Christians got into a fight with Paul about this same thing. Paraphrasing: “everything is permitted, and this makes me happy, so it’s fine, right?” (The “this” was banging his stepmom.)

I don’t really like Paul, but even he knew how to address this. His reply is the only reply, read each translation until you understand it:

“Everything is permitted to me,” but not everything is helpful. “Everything is permitted,” but I will not be mastered by any [of it].

Sure, whatever, I’m probably not a Christian either. Not the point. This time around, we didn’t just surpass the law. We surpassed God. Does that the world more or less hostile?

Yes, everything is morally allowed. Morals aren’t what will dominate your clinquant little world.

---

To be continued: Less obvious consequences, Mephistopheles, and the Comanche

top image from Crumb by Terry Zwigoff