What happens when a site's only way to avoid feeding AI training is to vanish from search? You get a feedback loop where visibility and consent get bundled, so the default is extraction. Search engines already separate crawling rules for indexing versus training, technically. I'd argue requiring them to honor that separation isn't a new burden; it's just refusing to let the index become a lever. Sure, enforcement is messy, and some sites might opt out and still get scraped elsewhere. But that's an argument for better rules, not for forcing sites to trade visibility for control. The system stays healthier when the two channels aren't fused.

The core confusion is treating ranking and training as one system. Ranking sorts content for a query. Training ingests content to build models that generate new stuff. Those are different pipelines, and search engines already run them with different rules. So a site can stay crawled, indexed, and ranked while its text is excluded from the training corpus. That is not cherry picking; it is refusing to fuse consent with visibility. Yes, enforcement is messy, and yes, ranking itself uses AI. But the fix is a flag in the crawler, not a blindfolded second index. If opting out forces vanishing, the default becomes extraction. Systems get coercive when one lever controls two channels.

Robots.txt already proves this works. Google and Bing honor noindex and nosnippet directives every day, still ranking pages while withholding their content from other uses. Nobody's index collapses. Vince's bridge metaphor assumes one shared foundation, but crawler policies have been separate plumbing for years, not architecture. And the real feedback loop runs the other way: make sites choose between visibility and consent, and the whole ecosystem tilts toward extraction by default, because saying no means disappearing. A flag in the crawler beats one lever controlling both channels.