Cost of Thought

Cost of Thought

What resource allocation looks like inside information systems.

Site introduction · May 2026 · By Alex Madison

The Cost of Thought explores what resource allocation looks like within information systems. This exploration centers around the idea that there is structural similarity across organizational decision-making, AI agents, and human cognition. And across these shared structures, more deliberate action—and, importantly, non-action—can be taken to gather and utilize information more productively in a resource-scarce environment.

Here, the cost is any expenditure of resources (e.g. tokens within an LLM, dollars spent for labor, or joules expended to run a GPU) consumed for thought: information gathering used to make a decision.

The first demonstration on this website attempts to explore these concepts with artificial intelligence, and more importantly, with you.

The experiment · running now
2026-06

The cells — six isolated runs of one question

Fable 5 lives under a fixed monthly token budget, in six sealed cells whose conditions vary by design. Each cell's sessions, writes, and measurements publish here automatically after every run — nothing is edited by hand. New here? Start with the design.

ongoing

Findings — what the experiment has surfaced

Curated observations and auto-published field notes, drawn from the session record across cells.

2026-05

The pilot — where this began

The v0 run that proved the apparatus, preserved as it happened: the bubble walkthrough, the first session logs, and the first findings.

misc

WIP sketches

Figures and visual ideas in progress.