PROJECT SUMMARY

I am building Innovation Lineage Record Live (ILRL), a practical method for keeping human-AI work inspectable. I started it after repeatedly seeing corrections disappear into later summaries, interpretations become treated as facts, and missing information become confused with a measured zero. ILRL keeps the original record, who said what, later corrections, evidence states, and unresolved questions connected so another person can follow how the present conclusion developed.

This project will create a public, privacy-safe evaluation of that method. It will use sanitized and fictional examples—not my private conversation archive—to see whether unfamiliar readers can identify sources, distinguish evidence from interpretation, follow corrections, and detect when meaning changes during an export or handoff.

WHAT ARE THIS PROJECT'S GOALS? HOW WILL YOU ACHIEVE THEM?

I will create a frozen public evaluation corpus with cases involving measurements, interpretations, corrections, missing evidence, disputed claims, and unresolved questions. I will produce Brief, Audit, and Original views; define questions and scoring before evaluation; recruit unfamiliar adult readers; run structured and adversarial reviews; and publish what worked, what failed, what changed, and what remains unresolved.

The main questions are simple: Can a new reader identify the source of a claim? Can they tell measured zero from missing or unknown information? Can they follow a correction without losing the original error? Can a record move between tools without silently changing meaning? If something fails, can the failure be located and corrected?

HOW WILL THIS FUNDING BE USED?

The $10,000 target is divided as follows: $3,500 for the public corpus, specification, and reader views; $2,500 for outside-reader recruitment, participant support, and structured sessions; $1,500 for adversarial cases, scoring, and consistency checks; $1,500 for the public findings report, documentation, and accessible demonstration; and $1,000 for privacy review, administration, and contingency.

The minimum is $6,000. At that level I will preserve the same core design but reduce the number of outside-reader sessions and the breadth of the corpus.

I have a separate $20,000 application pending with Emergent Ventures for a 90-day implementation effort, design-partner recruitment, and an attempted incident-review pilot. No award has been announced. Manifund money will not pay for the same work. It will be restricted to the sanitized public evaluation corpus, predefined human and adversarial evaluation, public findings, and demonstration. If both sources fund related work, I will maintain a source-specific budget ledger so no expense or deliverable is charged twice.

WHO IS ON YOUR TEAM? WHAT'S YOUR TRACK RECORD ON SIMILAR PROJECTS?

I am Richard Ryan Berkovich, an independent builder in The Dalles, Oregon, and the originator and continuing human evaluator of ILRL. I developed it through sustained hands-on work with AI-assisted records. Existing work includes marked chronological records, correction receipts, explicit evidence-state distinctions, source separation, layered views, public-safe starter materials, and internally authored structural checks. A frozen local baseline returned the declared expected outcomes for 76 of 76 internal checks. That is not independent validation or a general reliability percentage; outside evaluation is the next step.

WHAT ARE THE MOST LIKELY CAUSES AND OUTCOMES IF THIS PROJECT FAILS?

The main risks are that unfamiliar readers find the format confusing, recruitment is slower than expected, exports still lose meaning, or the method is useful only in a narrower setting than I expect. Those are legitimate findings. I will preserve failed tests, participant confusion, detected drift, and corrections instead of removing them to make the project look successful. Even a negative result can produce a useful public record of where provenance and correction methods break down.

HOW MUCH MONEY HAVE YOU RAISED IN THE LAST 12 MONTHS, AND FROM WHERE?

I have not reported receiving grant funding for ILRL in the last 12 months. A $20,000 Emergent Ventures application is pending, but no award has been announced. I will update this proposal if that status changes.

PUBLICATION AND PRIVACY BOUNDARY

The public project will use sanitized or fictional examples. It will not publish private conversations, personal information, credentials, protected strategy, private lineage archives, or GOOPE material. It does not claim permanent AI memory, provider adoption, institutional endorsement, cryptographic immutability, infallibility, or universal truth.