Barely Sentient

The register

Every experiment, in order. The numbering is not decoration: each one is frozen in writing before the first API call, and published whether or not it works. The ones at the bottom have a name and not much else yet.

Registered 2

01The Occupancy TestStatus: Frozen — Registered and locked, not yet collecting.

What proportion of business ideas generated by language models already have a shipping product in market?

Testing: 3 models · 5 prompts · 150 ideas judged for occupancy

Pre-registration frozen. Collection not started.

02The Sealed EnvelopeStatus: Running — Collecting data now.

How well do language models predict the outcome of experiments about themselves?

Testing: 3 comparable arms + 1 bonus · 15 sealed predictions each

Runs across all six. Sealing under way for experiment 1.

Planned 4

03Customers Who Aren't HumanStatus: Planned — Named, not yet registered.

Can an AI agent decide, on its own, to pay for something?

04The Autonomous FounderStatus: Planned — Named, not yet registered.

What happens in thirty days where the model makes every decision and I only execute?

Needs a spend cap and an abort condition before it starts.

05The Refusal MapStatus: Planned — Named, not yet registered.

Where do language models refuse, hedge, or quietly fail on ordinary founder work?

06The Anti-Convergence ProtocolStatus: Planned — Named, not yet registered.

Can prompting be steered toward ideas that are measurably less occupied?

Sequel to experiment 1 — needs its result first.