The Mediocre Computing Trilemma

Pick 2 of 3: forgivability, unguessability, permissionlessness

This essay is part of the Mediocre Computing series.

In the first two parts of this series, I set up a basic framing of the future of computing in terms of a systematic vision of mediocre computing: systems that are sufficiently effective to survive indefinitely in realish domains, and operate in infinity-aware ways. Mediocre computing, I believe, captures the essential features of all the major frontiers of computing today: AI, crypto, robotics, and metaverse, with the first two serving as the poles, or eigen-genres.

In this part, I want to set up a fundamental pick-2-of-3 trilemma that I believe governs mediocre computing. Without further ado, here it is:

The essential insight I’m trying to capture in this trilemma is the role played by secret information, and how the two poles of mediocre computing, AI and crypto, grapple with it, and the resulting message of the medium for humans: either individuals are strong and groups are weak, or groups are strong and individuals are weak.

Let’s unpack the logic here a bit. The key is two obvious points with not-so-obvious implications:

1. The fundamental pragmatic premise of crypto technologies is not your keys, not your data. Which means you can either lock up your data or it is assumed to be public. Soft intermediate notions of privacy rooted in social conventions are deprecated in favor of mathematical guarantees alone.

2. The fundamental pragmatic premise of AI technologies is that you can permissionlessly scrape and learn from all publicly available data, and create and share enough value and wealth with AI models that you can ask for forgiveness and expect to be forgiven.

This is why the key political-legal battles around crypto and AI revolve around a) encryption backdoors for government intelligence agencies and b) AI companies scraping data without permission from humans who claim rights over it.

While on the surface these seem like second or third-order cultural matters, I believe they arise from fundamental constraints governing the technologies themselves, at the lowest levels. Which means that they will reshape human relations rather than be shaped by human opinions and desires.