Labs Lose Lessons, AI Repeats Them: Inventor Vatsal Soin Files 0 to 1 Patents to Keep What Teaches in Superintelligence Race

Labs learn from failed experiments, yet many records are lost, and AI now runs far more tests than people. Two Metainventions, filed separately, share one yardstick from 0, no fit, to 1, ideal fit. The 0→1 Doctrine is designed to test each AI step before it runs. Band-Based Archive Triage (BBAT) is designed to sort lab records by their labels, each decision with a receipt. Either works alone or together

 

0→1 DOCTRINE

Tests each AI step first, against ranges people set. BBAT

Sorts stored lab records, new or old, by label. RECEIPTS

Each decision leaves a sealed, readable record.

THE QUIET LOSS

Every lab learns from what goes wrong, yet failures are the records most likely to vanish. A failed run is jotted in a notebook, saved on a drive that is later replaced, left with a researcher who moves on, or never written down. Published science also leans towards successes, so the next team cannot see what was already tried.

WHY AI RAISES THE STAKES

AI can plan and run far more experiments than people, so it produces far more failures. If those are lost, the next AI or team may repeat the same dead end at speed. Automated labs are being built to feed each result into the next design, which makes the record part of the method.

ONE LANGUAGE FOR EVERY TEST

In the 0→1 Doctrine, each measurable limit, such as temperature, purity or cost, is designed to sit on a 0-to-1 scale built from that field’s own range. Humans approve the ranges, and legal rules stay separate required checks. A step can then be tested before it runs, in chemistry, medicine or materials.

WHEN A RUN GOES WRONG

During a run, the Orchestration and Compliance Token (OCT) watches for a limit being reached and, by design, pauses the task, reroutes it or calls a human. The Actuation Compliance Receipt (ACR) then records what was tested and what followed, so even a stopped run leaves a clear note of why.

EARLY WARNINGS ARE ADVICE

The Predictive Risk Advisory Token (PRAT) is designed to warn a lab before an experiment reaches a limit, and people decide whether the limit changes. EMERGE (Emergent Meta-Environmental Response and Governance Envelope) is designed to flag wider conditions, such as a shared supply running short. In BBAT, SPAR (Systemic Pattern Analysis from Receipts) is designed to flag patterns in sealed receipts, such as a burst of purges. Each advises for review; none acts alone.

Figure 1. A failed experiment, three questions: today and with the two inventions.

THE ARCHIVE NOBODY OPENS

Behind every lab sits an archive of instrument logs, old notebooks, failed runs and files from retired equipment, new or old. Much of it is ROT data, meaning redundant, obsolete or trivial, and some sits on forgotten drives that nobody dares wipe. Keeping everything costs money and carries risk; deleting blindly loses lessons.

SORTING BY LABELS

BBAT is designed to route records by the labels they already carry, such as age, kind of result, purpose and any legal hold, without opening the files. Each record goes to keep, promote for use or purge. A failed run can carry a label that says what it is, so it is not filed away blindly.

A FAILURE WORTH KEEPING

Failed runs are not all alike. A broken instrument teaches little, while a true negative result can spare the next team repeated work. The label tells them apart. Humans set the rules for what each kind of record is worth, and BBAT applies those rules the same way each time.

THE COST OF REPEATING

A repeated experiment spends reagents, instrument time and weeks of effort, and an AI loop can repeat it faster than a person. Keeping a labelled failure may cost far less than running it again. The aim is not to keep everything, but to keep what teaches, and to show why the rest was dropped.

WHO DECIDES A PURGE

No record should vanish by habit. BBAT can leave a receipt for every purge, hold records under legal hold, and send a doubtful case to an authorised human reviewer through its Human Override Pathway. A purge then follows a rule and leaves a note. Humans set that rule, not the software.

LEARNING WITHOUT HANDING OVER FILES

Some records are worth sharing beyond the lab. BBAT can turn a group of them into one bounded band inside protected hardware. A minimum group size and band width are checked first, so a band stands for many records, not one. A partner can learn from the band while the files stay home.

ONE EXAMPLE: A CATALYST TEAM

Example, illustrative only: an AI agent tests 1,000 catalyst recipes. Figure 2 follows them from the first run to the final record.

Figure 2. One catalyst test, with illustrative figures.

EITHER INVENTION, OR BOTH

A lab running many experiments may start with the Doctrine. A lab with years of stranded archives may begin with BBAT. Each stands alone. Together, one lab can show what its AI could do and what it chose to keep, so failures stay findable for the humans who review the work.

TOWARDS SUPERINTELLIGENCE

A superintelligent age is not here, and this article does not say these inventions will bring it nearer. If AI scientists do multiply, they could repeat science’s old habit of forgetting failure, only faster. Both inventions are designed to work with labelled records, keeping or purging them by rules humans set, so lessons stay findable.

READ THIS WITH CARE

Everything here describes what the filings are designed to do, not what any deployment has shown. A kept failure helps only if its cause is recorded, and a label is only as good as the human who sets it. Neither invention replaces the scientists, the journals or the review that judge the work.

QUICK ANSWERS

Are the records opened?

BBAT is designed to sort by labels without opening the files. Anything shared leaves only as a bounded band. Does it cover results an AI generates?

Yes. BBAT is designed to work on stored data, new or old, so results an AI generates and then archives are sorted the same way.

What does a lab set up first?

People set the ranges once and the label rules once, and the inventions work from there. How is a purge checked later?

A sealed receipt can show why a purge happened without holding the record, so an authorised auditor can read it later.

CLOSING NOTE

Science has paid for every failed experiment. Check each AI run before it starts, and sort each record, new or old, before it is lost.

LIVE www.0to1doctrine.com

This can be tested, live, via API, governed against ungoverned, side by side.

THE INVENTOR

Systems theorist, inventor, and serial entrepreneur Vatsal Soin holds grants in the US, Japan, India and more; filings span six continents; latest grant, AI footwear, August 2026. Inventions: footwear, apparel fit, Sharable Size Cards, digital twins, robotics, global governed data, archive triage, quantum-resilient cryptography, AI governance. SIM–RMIT and NTU alumnus. Every prior invention was a step; 0→1 is the destination.

SELECTED REFERENCES

Granted: US Patent 12,446,652 B2 · Japan Patent 7560909 · India Patents 454081 and 599317. Filed: PCT/IN2025/051943 · US 19/489,595 · India 202511115781 · Australia AU2022450649 · India 202611113867 (23 September 2026) · PCT/IN2026/052042 (1 October 2026).

DISCLAIMER

Informational only. Not certified. No endorsement implied. Not investment advice. Examples are illustrative, not field results. Vatsal Soin · © 2026 All Rights Reserved.