On Legibility vs Trust
I've been reading Planecrash lately. Here's how it begins:
This story begins in a place that would be, as seen by some other places, a high-trust society...
And the thing that stands out, once you read some of the story, is that Eliezer's Dath Ilan is very emphatically not a high-trust society. Its people are constantly questioning the motives and plans of everyone around them, "why trust what you can verify" is a universal common saying, and it's considered extremely socially taboo to cry in front of someone else (because it might impose an emotional cost they don't want to pay).
What it has, instead of trust, is extremely high legibility. Everyone thinks in basically the same way, there's a basic universally-spoken language with no ambiguities or homonyms, and microtransactions for basic favours are standard. Probably the most extreme example of replacing trust with legibility is the story about the government: A few decades ago the government decided there was a secret they needed to make sure no one ever knew. This is inherently suspicious, but instead of asking whether the government was trustworthy, the people of Dath Ilan simply made sure they couldn't use this secret to seek power by setting up an entirely new government with a completely different set of people using a global paper-based cryptographic system. It's the sort of thing you can do when you have implausibly extreme level of legibility and coordination, and need to use them to replace trust.
(As an aside, there's a whole article to be written about how Dath Ilan's a slightly-twisted representation of the circa-2010 techie/internet culture Eliezer grew up in, but that's a whole other issue).
There's a culture of people (originally based on James Scott's Seeing Like a State) that thinks all legibility is bad. I still think they're wrong, but after reading this I at least see what they're afraid of. Living in a place like Dath Ilan, that replaces all trust with legibility, sounds awful. There's three levels of thinking about this:
On the first level, Legibility and Trust are roughly orthogonal. Legibility in itself is good, and living in a high-trust societies are good, and neither of them fully substitutes for the other so we should have both.
On the second level, there is a nonzero amount of displacement happening: Legibility can reduce the need for trust, to some degree, which makes people prioritize building trust less. More indirectly, legibility lets newcomers integrate into society more easily, which means more legible societies also have more strangers, immigrants, and cultural variance, which reduces trust. So there is one point here I will concede to the illegibility crowd - you shouldn't put all your eggs in legibility, even if it seems like it could solve your problems.
On the third level - we still need legibility. Firstly, because displacement effects are second-order, so they're almost always less significant than the original benefits of legibility. Secondly, because where they don't displace, legibility and trust benefit each other - a work environment works best when you can both understand what's expected of you and trust that people will be reasonable when measuring it. And third, because legibility scales better and is easier to achieve. It's nice to imagine being part of a high-trust illegible community, but if you're not already in that community, illegibility is awful. And most people don't start out in a high-trust community they genuinely like.