Reply To: Anyone else seeing AI snippets ignore real pages?

[Mason]

Yeah, but I think a lot of people are still underestimating how messy this is. It’s not just “semantic signals” in some clean little sense. Half the time it feels like the system is just latching onto whatever looks easiest to reuse, and that can be some junk forum post or scraped garbage if it happens to line up better on the surface. I’ve seen decent pages get ignored too many times to pretend it’s random. And honestly, the “best page wins” crowd is still stuck in old Google brain. That doesn’t seem like what’s happening anymore. Sometimes the page is right there, indexed, linked, all the usual stuff, and it still loses to some thin crap because the wording or entity mapping is a better fit. Annoying as hell, but that’s what it looks like. Nathan’s been saying basically this, but I wouldn’t go as far as treating it like some magic semantic overhaul either. Feels more like retrieval is getting sloppier in practice, not smarter. In my opinion,