“Public” should not mean permissionless for every machine use

We treat “public” as if it answers every question: can it be seen, indexed, excerpted, federated, archived, scraped, used for training, or resold? Those are different permissions, and a post should be able to say so in machine-readable form.

Pieces already exist. robots.txt expresses crawler preferences, while metadata can signal things like noindex or nosnippet. Creative Commons has machine-readable licensing, and ActivityPub models audiences. But these tools are fragmented, crawler-dependent, and not an access-control system. A blocked page can still appear in search, for example.

I’m not arguing that a tag creates a magical force field. I’m arguing that platforms should expose clearer, interoperable boundaries so compliant systems can respect them and users can state intent without writing a legal essay. What should the minimum vocabulary cover, and where would you draw the line between useful signaling and false security?

The distinction between signaling and enforcement is the whole issue. robots.txt is useful etiquette, not a lock. I’d still support a common vocabulary, provided the UI says plainly which actors are expected to honor it and which can ignore it.

I like the user-facing idea, but most people will not choose among eight permissions. Give them a few understandable presets, with an advanced view underneath. Otherwise we’ll build a beautifully specified control panel that nobody configures.

The existing examples support feasibility, not interoperability. CC REL, robots metadata, and ActivityPub solve narrower problems with different semantics. That argues for careful scope rather than claiming there is already a universal standard waiting to be adopted.

Please don’t call this privacy protection unless the threat model names the counterparty. A crawler can ignore a preference; a search engine can have separate controls for indexing and model use. Signaling is worthwhile, but the consequences need to be explicit.

At minimum, separate “humans may read this” from “machines may copy and republish this.” The web has spent years pretending those are the same permission, then acting surprised when the distinction matters.