This repo exists so people can propose bold, half-formed, or exotic ideas about the mathematics of artificial cognition without fear of being mocked for it. That protection is for people, not for claims. Keeping both things true at once is the whole point of this document.
Critique ideas as hard as you want. Never attack the person who raised them.
"This hypothesis doesn't hold up because X" is always welcome, however bluntly put. "This is a stupid idea" or any variant that targets the person rather than the argument is not, regardless of how wrong the idea is.
- Point to the specific unverified step, not just a vibe of skepticism ("this assumes the trajectory is stationary -- has that been checked?" beats "this seems sketchy").
- If you think a note overclaims, say what it would need to not overclaim (a citation, an experiment, a weaker phrasing) rather than just flagging it as wrong.
- "I don't understand why this would be true" is a legitimate, useful comment -- it doesn't require you to already have a counterargument.
- It is fine, and encouraged, to write a falsification protocol for someone else's note (see Note #008 for an example) instead of just arguing about it in the abstract.
Every research note should make clear, in its own text, which parts are:
- Established -- real math or results that hold regardless of whether this project's application of them pans out.
- Proposed -- a specific, falsifiable claim about AI systems, not yet tested.
- Open -- a genuine unknown, marked as such rather than quietly assumed.
If a note starts reading like a finished result before an experiment has actually been run, that's exactly the kind of thing this space depends on people flagging -- see Note #008 for what that looks like done well, and the audit history of the (separate, earlier) SLC/Principia work this project's maintainer has also been through for what happens when nobody does.
If you want to brainstorm with no rigor bar at all -- pure "what if," no
citations required -- that's legitimate too, but it belongs somewhere
visibly separate from research_notes/ (e.g. a speculations/ folder or a
clearly-tagged Discussion thread) so nobody mistakes a napkin idea for a
tested claim. Ask in Discussions if that folder doesn't exist yet and you
want to start one.