qproj design docsOpen questions
Questions with no answer yet. Each links to where it matters.
| question | page |
|---|---|
| Parse at runtime, or only while authoring? | spike |
| What’s the latency budget for a runtime parse, on what target CPU? | arms |
| Free-text input or the thinking editor (or both)? | game |
| oAMF: what does it cover, and can it run locally? | argument mining |
| How is a debate won? CAPNA, Dung, value-based frameworks. | evaluation |
| How many gold maps before scores mean anything? | gold |
| INCEpTION or the xAIF editor as the annotation tool? All real annotation so far has been in the editor. | gold |
| Is the PTA’s 50 types too much to teach? The editor could show lever families and keep the full code internal. | game |
| How does the game stop brute-force submissions? | game |
| Shipping real texts: public-domain translations or permission, per text. | — |
| Three copies of the rules (TS, Python, Rust): a conformance suite, or one Rust core compiled to WASM? | phases |
Reported speech: do reported locutions sit in the reporter’s TA chain? (open in judgement-calls.md) | IAT |
Dialogue tool: Yarn via bevy_yarnspinner, not yet evaluated. | game |
What links here: Arms and decisions, Quell debate: plan map, Debate