Le lundi 03 novembre 2025 à 10:09 -0500, [email protected] a écrit : > On Mon, 3 Nov 2025, René Neumann wrote: > > > while it's great to have such an optimistic view on the world, I doubt that > > this is manageable in the longer run. I see other open source infrastructure > > When I briefly ran a MediaWiki installation myself, the amount of spam > traffic it attracted was incredible - and now, years after taking it down, > I still get many bots trying to access pages in that URL space. But maybe > things will be different for this one. I'm all for introducing security > and access limitations only as and when they prove to be necessary.
Yes, I prefer to introduce restrictions only if and when there actually is a problem. We'll see how this goes. For now, I'm indeed seeing that about half of all requests are from the bot that trains ChatGPT, and the other half is split between a bot from Amazon, a bot training Claude, something called "Thinkbot", yet another bot from Bing, some more bots, and under just ~3.5% of non-bot requests. But the request volume is pretty manageable still (~0.8 requests/second). And at least they're not hiding themselves in the User-Agent.
