Igor
You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
Written with AI assistance, like the rest of this project: the project is about making AI-generated work checkable, so I say so up front. Every fact below is mine to answer for.
AI agents increasingly run research on their own and report "discoveries". Overseers cannot easily tell real findings from self-deception or overclaiming. In 2025, GPT-5's "solved" Erdős problems turned out to be already solved in the literature. SDCP is a working prototype of oversight for autonomous AI research: pre-registration before computation, hash-sealed predictions, null control worlds that measure how often the agent claims structure that is not there, and a generator/evaluator split where a second, independent AI model reviews the first, can stop it, and alone opens sealed results. Project page: https://claude.ai/artifact/UYFp2CXdF78dLEPTU4H6VU
1. Blinded real-data test: a structure is deliberately removed from the engine's vocabulary; false-discovery controls and success criteria are frozen and hashed before the run; the independent evaluator opens the result, and it is published either way.
2. Public verification protocol: a short, reusable specification.
3. One worked audit of a public "AI discovered X" claim using the protocol.
4. Open-source release of the tooling: data-custody kit, one-shot sealed executor, fail-closed evaluator with exact CRPS scoring.
Funding goal $13,200 for 6 months: $9,600 stipend for me (includes income tax; this would be my primary income), $2,400 AI model usage for the generator and evaluator models, $1,200 (10%) buffer. Minimum $2,400 covers compute only; I would then keep working part time.
Just me, Igor Trofimenko, an independent, self-funded researcher, working with two AI models from different companies: one generates, the other independently audits and can stop the work; I act as the authorizer. Since August 2026 SDCP has produced calibrated false-discovery rates on null worlds, blinded runs opened only by the evaluator, an append-only record of my own withdrawn novelty claims and rejected engine versions, and a first pre-registered prediction on public NHANES data confirmed in an independent later cohort (preliminary; the effect is probably already known). No new discovery yet.
Most likely failure: the blinded test comes back negative. That is still a useful published result, because it measures the limits of an AI discovery engine under honest controls. Second: the protocol gets little uptake outside the project; the tooling and protocol stay public either way.
None received so far. Pending: Emergent Ventures (applied 2026-10-08, $12,000) and EA Funds Transformative AI Fund (applied 2026-10-08, $13,200). If more than one comes through, I will adjust amounts to avoid double funding.
There are no bids on this project.