The Economics of Recursive Self-Improvement

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

A back-of-the-envelope calculation suggests that feedback loops are not currently strong enough to generate a self-sustaining acceleration, though they appear to be strengthening. We conclude by assessing the plausibility and implications of such an acceleration.

elasticity.institute →

Connections

Related to

SIA: Self Improving AI Framework SIA (Self Improving AI Framework) is a direct practical instantiation of the recursive self-improvement dynamics the paper attempts to model economically - the paper provides theoretical context for evaluating such systems
The Shape of Things to Come, Part 1: The Continuous Thunderdome 'The Shape of Things to Come' analyzing competitive AI dynamics directly intersects with the economic modeling of recursive improvement and whether acceleration becomes self-sustaining

Challenges

Intelligence is not the main bottleneck The paper's finding that feedback loops aren't yet strong enough for self-sustaining acceleration directly challenges the premise that intelligence alone is insufficient - it suggests the bottleneck may be in the feedback dynamics themselves, not just non-intelligence factors
The Dynamo and the Computer: An Historical Perspective on the Modern Productivity Paradox The productivity paradox paper (Dynamo and Computer) contextualizes why recursive AI improvements may not yet translate to measurable economic acceleration - the lag between capability and economic feedback loops

Supports

The Economic Implications of Learning by Doing - Kenneth Arrow (1962) Arrow's 1962 'Learning by Doing' economics paper provides foundational theory for the feedback loops and returns-to-experience that the recursive self-improvement analysis explicitly builds upon
Ethan Mollick: Meaningfully Better AI Releases Are Accelerating Mollick's empirical observation that meaningful AI releases are accelerating provides real-world data that supports the paper's tentative conclusion that feedback loops 'appear to be strengthening'

Type:	link
First published:	Tuesday, 11 August 2026 at 23:01 UTC
Location:	https://elasticity.institute/rsi-paper.pdf
URL:	https://mattwood.fyi/i/554b2536