Back in the olden days when Gmane was gmane.org, I had tons and tons of stats going… but after that meltdown, I’ve rather avoided thinking about Gmane, I guess.
So I’ve definitely not had any statistics going.
(For new readers, news.gmane.io is an NNTP server that’s both a mail-to-news gateway and an RSS-to-news gateway. It’s for those who hate using browsers to read stuff.)
But I did start wondering lately what the numbers were — is anybody using the thing? Have the AI companies started mining it and downloading petabytes of data per day? How many articles are posted per day anyway?
Then I discovered that I don’t actually have any data to work with, because I haven’t been running the programme that goes through the logs and generates stats… so I got that up and running again, and after running it for a few days (3) I have some numbers, man. (I’ve put the stats here after spending days and days on the exquisite design.)
So the number of users is around 500. I don’t know whether that’s less or more than I expected? Hm. I guess it’s less than I was expecting, even if I haven’t thought about it.
There are 100K articles posted per day (from mailing lists and the RSS feeds).
The chart of “top groups” doesn’t really tell us that much — there’s obviously something hitting the groups in an annoying way — doing a GROUP command before fetching an article. Sounds like somebody vibe-coded a downloader that doesn’t really understand NNTP (I’m shocked).
And 21M articles read per day. Now, 500 normal people aren’t reading 42K messages per day, so it does indeed look like somebody’s downloading a lot.
Heh. The server outputs 100GB per day?
So I wasn’t going to make a per-host stats chart (because of privacy), but I just got too curious. So we see that there’s only a handful of hosts responsible for 90% of the bandwidth usage. The second one is a residential IP, so it’s either an actual person or it’s somebody using one of those handy residential IP VPNs that are so popular these days.
The rest are from 161.115.141.*, which is an IP range owned by iren.com, which is indeed some kind of AI thing.
So… I’m wondering: Should I care or not? It would be easy enough to ban IPs that download excessively, but I don’t know whether I should or not. I mean, it’s an open NNTP server, and I’m not stupid enough to host at AWS, so I don’t pay for the bandwidth.
Does anybody have an opinion? Block or not? You can leave a comment here on the blog or in the thread I’m going to open on gmane.discuss in a few moments.