Musings on the appropriate targets for standards
Consider two norms for how to critique those who violate your standards:
L: Focusing critique on those in power, who have the greatest opportunity to cause harm.
R: Focusing critique on those who it most accurately applies to, who most often cause harm.
I think both of these are to an extent needed, but also that they are to an extent conflicting with each other, because society is somewhat meritocratic and power therefore somewhat correlates with virtue.
In an extreme, they also both have failure modes. Too much of policy L, and you destroy valuable organizations for unimportant transgressions. Too much of policy R, and you leave no room for slack or diversity.
But you can’t totally drop either of them because good incentives are important. With none of L, your world ends up run by strongmen who grab all the resources for themselves. With none of R, your world ends up with destructive people running rampant, and lacks incentives for productivity.
I suggestively named these policies L and R because they seem to somewhat resemble leftism and rightism respectively.
It is probably also important to not obsess too much over the levels of these. L and R are one or two degrees of freedom. If they are set wrong, then yes, that can be a serious problem, but if they are set ~right, there’s not much more you can do with them. Meanwhile, actually acting in the real world rather than moving norms, there’s an infinitude of problems and opportunities that you can work on.
This reads a bit too abstract. A few examples could be in order.
One of the less-politicized examples I have in mind is community moderation. For instance in Well-Kept Gardens Die by Pacifism, Eliezer lays out the perspective of the R norm. Meanwhile, for moderation, L-norms would be stuff like moderation transparency, elections to select moderators, criticizing Eliezer for various things, etc..
I dispute this strongly. “Power correlates with virtue” is a very WEIRD opinion. The majority of humans, both historically and in the present time, live in places where power has little if any correlation with virtue.
I was thinking specifically in WEIRD cases.
I think it would be good to call that out specifically in the text.
So you’re asserting it isn’t a tradeoff because power has little correlation with virtue, so we can optimize for pure L policies because R doesn’t come up in practice?
I don’t “optimize for pure L” would be the takeaway from there being no correlation between power and virtue. Even if there is no correlation, you would still want good incentives at all levels, not just at the top.
I thought about this a bit more, and the more I think about it, the more I reject the L/R dichotomy. Thinking about other countries and other periods in history, it seems to me that L and R are correlated. Societies where the top leaders are powerful and unaccountable also tend to be societies where there is a lot of petty corruption and various destructive but connected people run rampant, and you can do little to criticize them. And, on the flip side, societies that corral petty corruption also tend to be societies where accountability is demanded of leaders.
In other words, it seems to me that the real failure mode is having not enough of either, not being unbalanced one way or the other.
I think the trouble with “the real failure mode is having not enough of either” is that having either requires resources such as intelligence and force invested into norms. So when you look at how it’s functioned in different periods and locations, it is potentially confounded by amount of resources available to the society.
Or another way of looking at it is that by looking at the things that have actually existed, you miss the cases where too strong norms cause collapse and thereby nonexistence, and degenerate into a system with weaker norms.
Do you have any examples of societies where, for example, anti-corruption norms were too strong (too much R) and that led to some kind of collapse? Or societies which demanded too much accountability from their leaders (too much L), and that led to collapse?
I don’t know enough history to come up with concrete examples and I’m not sure it’s even happened in large-scale societies in practice because people might err towards smaller norms or because they notice something goes wrong before applying it to the whole society.
However, surely you can see how the standard impose costs? To prevent corruption, one thing you could do less delegation of things/do more review of the decisions made by those you delegate to. But in the limit this means having the king make all the decisions, which is just not viable. Mechanistically, something would go wrong.
That’s not the only way to prevent corruption. It’s not even the way most societies prevent corruption in practice. In practice, most societies with low levels of corruption prevent it by creating a set of social norms, which people internalize, that, in turn, prevents them from acting in ways that benefit themselves, but hurt the society in which they operate. These social norms also ensure that when corruption is detected, it’s punished severely, with both social and legal sanctions.
We (and by “we”, I mean those of us living in “the West”, broadly speaking) don’t prevent corruption via intrusive checks on every single government process. Instead we inculcate values of honesty and good faith, have legal consequences for corruption when it is detected, and rely on individual citizens to report corruption when it does occur.
This is possible because we live in a society that has both high R and high L. We have high standards for everyone, and broadly expect that most people, whether they’re high-ranking politicians or low-ranking civil servants will not enrich themselves at the expense of the public purse. That is why I’m saying that there’s no tradeoff between having high R and having high L, and, in fact, the two may actually be mutually synergistic.
EDIT: Indeed, this synergy is one of the the main arguments against lowering standards for L. As the saying goes, “the fish rots from the head”. Tolerating more misbehavior among leaders leads to an organization where lower level malfeasance is tolerated too.
I feel like we have different things in mind.
I can buy that for types of rule-violations that will eventually get caught, punishing rule-violations severely and instilling values to avoid them is a good and cost-effective approach. But some types of rule-violations are complicated to detect, and if people want to prevent those, they have to proactively gather information or create broader limitations, both of which seem costly.
In the specific case of corruption, in Denmark (and most of the EU? idk) we have something called “udbud”, where basically when the government needs some firm to do a job, it has to create a description of the job and put it out on the free market. From what I’ve heard, udbud is super chaotic, costly and unreliable, and I bet it could be done more efficiently if more flexibility was allowed on the part of the decision-makers (consider: most companies do not seem to use udbud in the same rigid way). However, that would trade off against corruption in complex ways.
Within the region were L and R are both cheap, I agree that L and R are both good. But advanced forms of bad behavior often don’t seem to get legibly detected, which prevents the automatic approach that you seem to be suggesting.