[ai-control] Re: [Question] Priority conflict between HTTP Header and Robots.txt in draft-ietf-aipref-attach

[Martin Thomson <[e-mail]>]

On Tue, Mar 3, 2026, at 13:13, John Mueller wrote:
> Sorry for mixing the context a bit - I was thinking back at the 
> original question where robots.txt said n, and HTTP-header said y, and 
> trying to generalize from there. 

No problem.  I just want to get to a clear shared understanding of what is and isn't agreed :)

> Perhaps the more general approach would be that individual attachment 
> mechanisms have inherent specificity assigned, and to refer to those 
> when it comes to defining order of precedence. This would indeed be 
> similar to CSS. I don't have a good sense of where that would be placed 
> though - what are your thoughts? Or does this need to be noted 
> somewhere, and picked up when the attachment draft is more fleshed out? 
> (or perhaps it already is fleshed out enough?)

Right now, the structure is that the vocabulary has a default rule, which is that all sources of preferences default to being co-equal.  It also says that specific attachments can specify a stronger precedence for that attachment relative to any other.  So this notion of precedence is permitted, but not baked in.  The case for building something like that into the structure of the grammar in the vocabulary draft is really about a balance between the added complexity that comes with it.  If we think that sort of precedence is common, that might be worth the effort.  I wasn't sure to begin with.