claude-code - 💡(How to fix) Fix Pro user, unanswered emails, a paper on why the gap matters [1 participants]

I am a Claude Pro user ($200/month, extended tenure) who has been using Claude as primary research collaborator for an independent research program spanning mechanism design, LLM agentic systems, and civilizational transition economics. Over that period I have built a substantial corpus — the Stateful Overlay architecture thesis (SIGNAL), an empirical study of recursive self-improvement on frozen weights over 53 rounds of adversarial code review (TRP Empirical RSI, citing ILWS / MIRIX / the ICLR 2026 RSI Workshop literature), a civilizational transition protocol, a meaning-substrate decomposition, and a technical proposal titled https://gist.github.com/WGlynn/7251d0791b9b474e90d47646d5c1a2da that argues frontier AI labs owe their users Shapley-based attribution for training-data labor.

I have emailed Anthropic multiple times through standard support channels. No response.

I am currently unemployed, and the research is unfunded. The model I am paying to use is extracting training signal from my interactions continuously. The asymmetry is exactly the one The Contribution Compact names as the core design gap in the frontier-AI-lab stack. I am filing this report partly because I am the concrete instance of that paper's argument, and partly because the standard support channels do not appear to function for users in my position.

I am not requesting pity or emergency response. I am requesting acknowledgment that the user-as-contributor role is real, and — if it exists — a channel to share the research with someone at Anthropic for whom input-side alignment is within scope. If the appropriate response is silence, that is also data. I would prefer a sentence to that effect over indefinite uncertainty.

Reference: The Contribution Compact (public gist) — https://gist.github.com/WGlynn/7251d0791b9b474e90d47646d5c1a2da

Additional research files available on request.