Chesterton’s fence is often used as an argument in favor of various traditions: monogamy, eating meat, gender roles, and so on. It’s a heuristic that states that you should never remove or alter a system, law, or tradition until you first understand why it was originally put in place. Unfortunately, it seems to be used, without much thought, as an argument in favor of conservatism in all things.
But first, why should we expect Chesterton’s fence to work in the first place? For this, we turn to the evolutionary optimality challenge (EOC) proposed by Nick Bostrom, which is quite similar in structure.
The EOC asks: “If the proposed intervention would result in a beneficial enhancement, why have we not already evolved to be that way?” Chesterton’s fence asks the cultural equivalent: if some practice is so bad, why did it arise, and why has it persisted?
We should expect that, if some intervention really is an enhancement, evolution would already have evolved it. If it hasn’t, either the enhancement isn’t actually an enhancement, or there must be some reason evolution hasn’t produced it. Bostrom provides three broad answers: altered tradeoffs, value discordance, and evolutionary incapacity.
For altered tradeoffs, evolution might have made an organism well adapted to one type of environment, only for that organism to end up in another, where the old tradeoffs no longer make sense. Consider GLP-1s. For modern humans with obesity, taking them can result in substantial improvements in health. But if a hunter-gatherer or subsistence farmer who was struggling to get enough food took GLP-1s, the result would probably be quite maladaptive. Evolution has adapted our appetite regulation to environments where getting enough calories was difficult. Many of us now live in an environment where the problem is that calories are far too easy to obtain. GLP-1s can be an enhancement because the tradeoffs have changed since our environment of evolutionary adaptedness (EEA).
For value discordance, the problem is that evolution doesn’t optimize for what we value. Evolution “wants” inclusive genetic fitness; we want things like happiness, autonomy, and the ability to decide how we’re going to live. Contraceptives are a good example. Evolution doesn’t quite like them, since they let people have sex without producing children. But we do, because being able to choose when (and whether) to have children is quite valuable to us. This is an enhancement that evolution had no reason to select for.
The third answer, evolutionary incapacity, is about the limitations of evolution as an optimization process. Colloquially, we talk about evolution as though it’s a designer, but really it’s more of a hill-climbing process. Evolution can’t look ahead and decide to make a change that’s bad in the short term but good once several other changes have been made. This means that it can get stuck in local maxima that a designer could escape from.
Bostrom gives the appendix as a possible example. The appendix has some immunological functions, but can also become infected, resulting in a life-threatening condition (especially in a world without antibiotics or surgery). Suppose humans would, on net, have been fitter without one. Why hasn’t evolution removed it? One possible answer is that a smaller appendix increases the risk of appendicitis. Evolution might have trouble getting from a normal appendix to no appendix if the intermediate steps are worse. Humans, however, can just remove the entire organ surgically, without having to pass through those intermediate steps.
These three answers to Bostrom’s EOC apply just as well to cultural evolution. In fact, I think we should expect them to apply by default.
Sometimes the fence was solving a problem that no longer exists. Cultural evolution may have produced some practice because it was adaptive in a particular environment, but if that environment changes, the practice needn’t be adaptive anymore. Finding out what problem the fence solved can actually be an argument for removing it, if that problem has disappeared.
And I don’t think this is some obscure edge case. Almost every culture in existence is living in a world radically different from the one it inhabited even just a few generations ago. The technology has changed much faster than many of our customs have. Why should we assume we’re anywhere near a cultural equilibrium?
For example, one might invoke Chesterton’s fence in defense of traditional gender roles: surely they persisted for a reason. And they did. There’s a pretty obvious reason why men and women traditionally ended up doing different kinds of work. In a world without reliable contraception, where childbirth was much more dangerous and a lot of work depended on physical strength, there were good reasons for this division of labor. But we no longer live in that world! We have reliable contraception, childbirth is much safer, and most jobs don’t require much physical strength. Even if the old division of labor was optimal under those conditions, why should we expect it to still be optimal today? If we want to defend traditional gender roles today, we need to explain why they’re good today. Explaining why they arose in the first place isn’t enough.
Altered tradeoffs aren’t the only reason cultural evolution might produce undesirable practices. Cultural evolution is not your friend.
At least with biological evolution, we can point to inclusive genetic fitness. What is cultural evolution even optimizing for? Cultural practices can spread because they encourage reproduction, improve a society’s chances of winning wars, make a society more stable, or simply because people are especially inclined to imitate them. None of these things necessarily makes the people who practice them better off. Even if cultural evolution produced a culture that’s extremely good at surviving and reproducing, the individuals in that culture could still benefit from changing it, because what they value is not the same as what cultural evolution selects for.
Honor culture is an obvious example. In societies with weak states, retaliating against threats and slights (real or perceived) can be an effective way of deterring people from taking advantage of you. There’s a good reason for wanting a reputation as someone who isn’t to be messed with. But even when it serves that purpose, I don’t think many of us would like to live in a society where people regularly kill one another over insults. And, in fact, many of the concrete expressions of honor culture (e.g., dueling) have been outlawed or suppressed by stronger states.
The (non-)usage of iodized salt is a good example of evolutionary incapacity. In iodine-poor regions, people suffered from goiter and other effects of iodine deficiency for generations. So why didn’t they just add iodine to their salt? Well, how were they supposed to know? They hadn’t even discovered iodine, much less figured out that not getting enough of it was causing these health problems. Some cultures had discovered that seaweed could help with goiter, but that’s a long way from identifying the relevant chemical and figuring out how to add it to everyone’s salt. Once we had modern chemistry and industrial food production, what had been a persistent problem became relatively easy to prevent. Salt iodization became widespread only in the twentieth century.
We have access to various tools, materials, and techniques that were unavailable to earlier cultures. The Green Revolution, automation, cheap contraception, and scientific measurement all make possible practices that could not have been reliably discovered, implemented, or maintained by cultural evolution alone. Given all this, it would be very surprising if traditional cultures had already discovered most of the available cultural improvements.
Chesterton’s fence makes you think about changing your culture, [penile] subincision makes you go through with it anyway
— niplav
There’s another reason Chesterton’s fence can fail, separate from the three answers to Bostrom’s EOC: a practice can persist without ever having been especially good in the first place.
The game theory folk theorem is relevant here. Under certain conditions, when people play a game repeatedly and care enough about future payoffs, all sorts of outcomes can be sustained as equilibria through the threat of future punishment. More precisely, many feasible, individually rational payoff profiles can be supported as equilibria. Feasible means that the payoffs can be obtained through some mixture of outcomes of the original game, while individually rational means that each player gets at least their minmax payoff (roughly, what they can guarantee themselves when everyone else is trying to make things difficult for them).
What’s interesting is that nothing here says the resulting equilibrium has to be a particularly nice one. In Hell is Game Theory Folk Theorems, the example is a group of people whose temperature is determined by the average setting on their thermostats. Everyone would prefer a cool room, but everyone sets their thermostat to 99°C because anybody who turns theirs down will be punished by the others turning theirs up even higher. So they’re all stuck in an unnecessarily hot room, despite being perfectly capable of cooling it down.
You can get analogous problems with social norms. If not binding your daughter’s feet would seriously damage her marriage prospects, it can be in your interest to continue the practice, even if you’d prefer that nobody practiced foot-binding in the first place. Getting rid of the practice would require a coordinated change in expectations, not just an individual deciding that the tradition is stupid.
Obviously this doesn’t establish that any particular custom arose through a repeated-game equilibrium. But it does show why we shouldn’t infer much from the mere fact that a custom has managed to persist. There can be many stable arrangements, and some of them are terrible.
Even setting aside the substantive objections above, Chesterton’s fence has a rhetorical problem. When people use Chesterton’s fence to defend some tradition, I doubt it’s usually their true objection to getting rid of it. They already like the tradition for other reasons, and Chesterton’s fence gives them a way of defending it without having to explain what those reasons are. If you managed to explain why the tradition arose, and why that explanation no longer applied, would they actually change their minds?
To them, I suggest considering Chesterton’s meta-fence, proposed by Stuart Armstrong. In our current system of democratic market economies with large governments, the practice of dismantling old traditions and institutions is itself pretty well established and has a decent track record. By Chesterton’s own logic, shouldn’t we be reluctant to interfere with that practice until we fully understand why it exists?