A protocol for the systematic rediscovery of scientific ideas already present in the world archive — lost to paradigm, timing, and institutional inertia for decades.
The scientific system is optimised for producing new knowledge — not for preserving existing knowledge. When a paradigm wins, alternative paths close. AI systems trained on this corpus amplify the mainstream. Exceptions disappear into archives no algorithm is scanning.
This is not a historical curiosity. It is a systemic failure reproducing itself right now.
The paper that described CRISPR before CRISPR existed. Published 1987. Ignored 25 years. SDA would have found it in month 1.
m3mbrane is, to our knowledge, the first systematic operational framework for algorithmically detecting, verifying, and financing ideas already present in the scientific archive. No existing DeSci project occupies this niche. 90% of the required infrastructure already exists — m3mbrane builds only the discovery algorithm on top.
Every search tool finds what you already know to look for. Reverse Discovery finds answers to questions you haven't asked yet. The client describes an R&D problem — SDA builds its embedding — link prediction finds structurally related work in the citation graph.
An open system without filters inevitably attracts noise. Funding is structurally unavailable until all four barriers are passed — in order.
Select a known premature discovery. See how SDA scores it across four dimensions — CAD, CDE, TMG, TRS.
A clean embedded view of the product surface. No page transitions, no mockup detour — the interface lives directly inside the site as a product window investors and partners can inspect.
This block shows the operating surface of m3mbrane as software: a discovery queue, scored candidates, citation history, validation flows, bounty routing, assets, and governance views.
The goal is simple: eliminate the gap between white paper and product imagination. A visitor should understand that m3mbrane is not only a thesis — it is a usable interface.
Google Scholar and Semantic Scholar find what you already know to search for. m3mbrane finds what nobody is searching for — because nobody knows it exists. That requires GNN on citation subgraphs, link prediction, and learning-to-rank on premature discovery cases. Nothing in the market does this.
Two-token architecture where governance weight is determined by contribution, not capital. This directly addresses the core failure of most DAOs — plutocracy dressed as decentralisation.
The first measurable indicator of how well science uses its own past. Nature Index measures what science produces today. Discovery Index measures what it already knew — but didn't realise. A language readable by Nature, the Financial Times, and corporate R&D directors simultaneously.
The algorithm finds anomalies in the citation graph. The scientist evaluates scientific merit. But between the archive and the algorithm there is a layer the machine cannot fully close: first-pass navigation through volume.
Inspired by Galaxy Zoo (150,000 volunteers classified galaxies more accurately than the automated algorithms of the era) and Foldit (players without biology degrees solved a protein folding problem in ten days that structural biologists had worked on for fifteen years) — m3mbrane opens the archive to anyone.
m3mbrane builds governance on Elinor Ostrom's principle from "Governing the Commons" (1990): commons governance fails not when there are many participants — but when those with the most capital gain disproportionate control over the rules. That is why $MBG cannot be purchased. Governance weight is determined solely by verified contribution — whether you are an ORCID scientist, a community member with high RP, or a laboratory with confirmed replications.
It may have been sitting in an archive since 1987. We are looking for scientists, investors, laboratories, and curious people who want to be part of this. The White Paper is available now.
Scientists, investors, laboratories. Updates on launch, validator onboarding, and bounty opportunities.