Should rule classification be applied anymore?

[HelenBurge]

After discussing this in the call, the classification of rules being manual or automated seems less relevant with AI bridging the gap.

FYI: @giacomo-petri @dan-tripp-siteimprove @zlayaAvocado @sage-putnam @shunguoy

[dan-tripp-siteimprove]

First - sorry if I'm missing something, but where is this classification currently done? I don't see any sign of it in the rule markdown file - for example, on Heading is descriptive. Nor on manual rules which are in the PR stage and which have a "manual rule" tag on the PR, such as "consistent navigation" and focus returns to trigger.

At any rate, I think the classification will still be useful, even if AI progresses very far. When we say "manual rule" what we generally mean is: a rule where precision is impossible. For example: all of the "X is descriptive" rules. They're not precise because they have no black-and-white logical "audit trail".

That's a different situation from rules like "ARIA attribute is defined in WAI-ARIA". The "audit trail" for that one is obvious.

For focus returns to trigger I think that it's inaccurate to call it a "manual rule", even without considering AI. It's true that testing for this rule will usually be done manually. But that's only because the tools that could automate it are awkward. The tools could be improved, to the point of making this as precise as an ARIA rule, without AI.

Maybe instead of "automatic" vs. "manual", we should rename the categories to "precise" vs. "imprecise". Or "black-and-white" vs. "judgment call". I can certainly see why "automated" vs. "manual" is not ideal, and will become worse as AI progresses. But we still need some kind of categorization along these lines.

[carlosapaduarte]

CG decision: remove the "manual rule" label