Proposals

Work SomaSoft thinks is worth doing, written down before it is built so that the reasoning can be checked against the result.

Two things to say plainly about this page. These are proposals, not products — nothing here is in development, and none of it has a customer. And this set is partial on purpose: of five founder proposals on file, three are withheld, and the reason is given below rather than hidden. A page that published all five regardless of readiness would tell you less, not more.

F-003 — Municipal infrastructure decision-support

Status: proposal only. Not a build. Needs a civic partner before it is anything.

A tool that helps under-resourced public staff see the trade-offs in infrastructure decisions — which pipe, which road, which year, and what each choice costs the people who live there.

The framing rule, which is the whole proposal:

This is capacity augmentation for under-resourced public staff. It is never headcount reduction, and never tax-base reduction.

That rule replaced the idea this proposal started as. The original framing was "reduce the tax burden", and it was rewritten because "reduce tax burden" launders a value choice as a calculation. Which services a community is willing to do without is not an optimisation result; it is a decision the community is entitled to make. A tool that presents it as arithmetic has quietly taken the decision away.

So the design commitment is that the system surfaces trade-offs and the community holds the values. It can say "deferring this resurfacing saves X and raises reported pothole complaints by Y, historically". It does not get to say which of those matters more.

What would have to be true first: a municipality willing to pilot it on data they already hold, and a written agreement that the outputs are advisory. Evanston and Chicago are the obvious candidates. Neither has been approached.

F-004 — Grounded astrophysics research assistant

Status: proposal only. Not a build. Needs a collaborating astronomer.

Of every domain considered, astrophysics is the best fit for what this system actually is — and for reasons that have nothing to do with scale.

The data is public, clean and provenanced. MAST, SDSS, ZTF, Rubin/LSST, arXiv astro-ph. No personal data, no employer or client data, no consent problem. That removes the entire class of constraints that gates the health and municipal threads before they start.

It is a field where "I don't know" is a respected answer. Astronomy is built on error bars, upper limits and non-detections. A system whose distinguishing discipline is declining to answer when it cannot ground the answer is, here, behaving like a competent colleague rather than a broken product. In most domains honest deferral reads as failure. In this one it reads as rigour.

What would have to be true first: one astronomer willing to pose real questions and say plainly when the answers are useless. Without that this is a demo, and demos have taught this project nothing it could rely on.

Withheld, and why

Three proposals are on file and are not published here.

• Clinical oncology collaboration. Clinical decision support is regulated software, and the proposal touches patient care. It does not go public, or forward, before qualified legal review.
• Quantum-computing intersection. Its own recorded status is that it "needs a real intersection before it's a proposal". That is accurate. Publishing a placeholder would pad this page at the cost of its credibility.
• Doctoral research by publication. Its own recorded status is that it is blocked on peer review and on an unresolved intellectual-property question. Both are true, and neither is resolved by publishing it.

Each of those will appear here if and when the blocking condition clears. That is the only mechanism by which anything is added to this page.

How to read any of this

Everything above is a hypothesis with a named precondition, and most preconditions are a person who does not yet exist in the plan — a municipal partner, an astronomer, an attorney. That is deliberate: this project has learned more from measuring things that did not work than from building things nobody asked for.

If you are one of those people, the proposals are the conversation, not the pitch.