Ideation — large-scale fixes and expansions

Drafted 2026-07-01.

This folder is the layer above the existing planning documents: extensive ideation on deep structural fixes and larger expansions that are not already written down anywhere in this repo. It was produced from a fresh read of the source, tests, CI, and docs on 2026-07-01.

How this relates to the existing documents

• docs/roadmap.md is the shipped product roadmap (v0.5.0 spec tracking, the v1.0.0 stability gate, the out-of-scope line). Nothing here restates it.
• docs/RESEARCH-ROADMAP.md (2026-06-30) triaged the synthetic persona panel in docs/USER-RESEARCH.md into remediations R1–R9 and expansions E1–E9. Nothing here restates those either; where an idea builds on one, it cites the ID (e.g. "extends E6") and says what is new beyond it.
• Everything in this folder is net-new: it comes from reading the actual code (src/tods_validate/), the test suite, and the CI workflows, and from asking what a validator that intends to anchor a young standard needs structurally, not just feature-wise.

Contents

File	What it holds
01-deep-dive.md	Current-state assessment: architecture map with file paths, genuine strengths, structural debt actually observed, strategic position in the portfolio
02-large-scale-fixes.md	FIX-01…FIX-15 — deep structural fixes: engine design, correctness, security, accessibility, performance, operability
03-expansions.md	EXP-01…EXP-16 — expansion ideas across three horizons (deepen the core, adjacent capabilities, transformative bets), plus considered-and-rejected ideas
04-impact-and-sequencing.md	Impact×effort matrix over all FIX/EXP IDs, dependencies, a Now/Next/Later sequence beyond the existing roadmaps, and the human/real-data gates

Honest framing

These are ideas for evaluation, not commitments. None of them has been sized against real user demand — the repo's single biggest known unknown is still that no real production TODS feed has been validated (R1 in the research roadmap, and the v1.0 gate in the product roadmap). Several items below are explicitly gated on that, and 04-impact-and-sequencing.md separates them out rather than pretending they can be finished from synthetic fixtures alone. Where a claim about the code is uncertain, the text says so.