Friday, December 6, 2024

Pinboard tags again

The behavior of Pinboard tags has puzzled me. I use a Pinboard account to create public tags for Orange Crate Art posts. The tags are supposed to work for everyone, taking an OCA reader to a Pinboard page with an index — okay, a list — of all posts with that tag.

In July 2021 I found that tags were working that way only if the reader was logged into a Pinboard account. As I’m always logged in, I thought that I must have dumbly missed something obvious — that tags always only worked for someone logged into a Pinboard account. (Two e-mails about that to Maciej Cegłowski, Pinboard’s developer, went unanswered.) In September 2021 tags were once again working for everyone, Pinboard account or no. Now tags are again working only for someone logged into Pinboard. I found a workaround:

pinboard.in/search/u:M.Leddy/?query=
And now the workaround no longer works for anyone not logged into a Pinboard account.

I found — finally — an explanation from Maciej Cegłowski in a Google Group, pinboard-dev:
The intent is for all public Pinboard pages to be visible without a login. However, user+tag pages in particular are somewhat expensive to generate on the fly, so a crawler that iterates through them can quickly bog down the site.

In the past, it was fairly easy to block or throttle this kind of crawling. But in recent months, I've seen a rise in distributed crawling from China+Singapore IP addresses, with no patterns that would make the traffic easy to block. So I’m forced to either put pages behind a login, or have the site become unusably slow for everyone.

If I have to I’ll block the entire PRC address range, though I’m looking for alternate solutions that are less drastic. But I just want people to know the reason for the back-and-forth behavior on public pages.
Ugh. I hope there’s a fix. Incidentally, there’s never been an acknowledgement on the Pinboard website of the problem with public tags. The website now announces a “big code cleanup underway.”

[And as you may have noticed, I’ve removed the widget with links to the top twenty OCA tags from the OCA sidebar.]

comments: 0