Thread · 9 tweets · 2 Oct 2022

This is brilliant and prescient. However, I've been wondering if the "code is law" framing isn't too restrictive. First, code isn't just law, it's judge, jury, and executioner. If law worked like code, we wouldn't like it much.

I wonder if the law-like nature of code isn't better captured by institutional analysis. Especially networked code. You have an arena actors, roles, rules about available actions, sanctions, who gets what information, how the rules get changed, etc.

Admittedly, the difference between the two angles isn't huge, and it certainly doesn't at all eliminate concerns of legitimacy (as @laurencediver mentions in thread). But different institutions require different legitimacy, and some can be quite informal.

The law frame also limits our imagination to state-like arrangements. That fits the dominant client/server architecture we live in today well, but we should be thinking farther afield. The *best* institutional model you can hope for with client/server is benevolent dictatorship.

This has real consequences. The best ethic for a benevolent dictatorship is "don't be evil." By contrast, better institutions have checks and balances that lend themselves to more nuanced ethical considerations, with enough power to match the complexity of our world.

Dictatorships, benevolent or otherwise, can only function with legibility, ie. with ever more data. This is a direct road to failure, even with the best intentions. See SLAS in general or @henryfarrell's notes in Seeing Like A Finite State Machine crookedtimber.org/2019/11/25/see….

(Note that historically thinkers from Polanyi to Hayek have seen this kind endeavour as impossible for lack of data; I suspect the issue is more computational tractability.)

Anyway, I believe that thinking from institutions might help free our imaginations from these limits. In peer to peer architectures, much richer institutional models are possible, ones with a distribution of power that can solve much more interesting problems.

This isn't to criticised all the excellent work on code as law, or to deny the prescience of @langdonw. I only wonder if a "yes and" approach isn't called for here to drive stronger analysis and better solutions. We need all the imagination we can get.