Historical programming-language groups disappearing from Google
([Development] Jul 28, 2020 15:04 UTC (Tue) (corbet))
- Reference: 0000827233
- News link: https://lwn.net/Articles/827233/
- Source link:
As Alex McDonald notes in [1]this support request , Google has recently banned the old Usenet groups comp.lang.forth and comp.lang.lisp from the Google Groups system. " Of specific concern is the archive. These are some of the oldest groups on Usenet, and the depth & breadth of the historical material that has just disappeared from the internet, on two seminal programming languages, is huge and highly damaging. These are the history and collective memories of two communities that are being expunged, and it's not great, since there is no other comprehensive archive after Google's purchase of Dejanews around 20 years ago. " Perhaps Google can be convinced to restore the content, but it also seems that some of this material could benefit from a more stable archive.
[1] https://support.google.com/accounts/thread/61391913?hl=en
[1] https://support.google.com/accounts/thread/61391913?hl=en
Historical programming-language groups disappearing from Google
I used to read comp.lang.lisp for a couple years around 2004 and it was an amazing experience. The complete disappearence of such archives would be a great loss.
Historical programming-language groups disappearing from Google
I used to read comp.lang.lisp for a couple years around 2004 and it was an amazing experience. The complete disappearence of such archives would be a great loss.
Historical programming-language groups disappearing from Google
This looks like ridiculous overreaction from some braindead automation. I thought Google was supposed to have the best AI in the world, but whenever they let that AI take any decisions that affect real people it seems to make disastrous mistakes, usually with no obvious human checking before disaster is inflicted and no recourse other than signal-boosting via the press, since Google always does this with no humans in the loop and no appeals procedure (not that it's clear who could be consulted in the case of historical archives!). This is... not a sensible way to work.
Deleting historical archives because of spam is even less sensible: it's not like more spam is going to materialize in the past history of comp.lang.lisp.
Historical programming-language groups disappearing from Google
I suspect that it's happening because Google Groups is two products merged together:
The old "dejanews" Usenet archives, which I'll call "Usenet" throughout this comment.
Google's in-house forum system, which I'll call "forum" throughout this comment.
It makes sense to remove entire forum groups which have always had a spam problem, and where the forum group owner isn't willing to use Google's tools to moderate it and keep it spam-free; after all, if you've created a forum group for the purpose of spamming, or if you simply gave up the moment spammers found you, there's probably not much non-spam in the group. This is doubly true since the tools have been there since the forum group was created, and advertised to you as the forum group creator; AIUI, Google has reached out to their owner of record for such forum groups and asked them to clean up, so anything left is something that nobody still cares about.
However, that analysis ignores Usenet. Usenet predates Google's spam handling tools (after all, it predates Google), and has never had good tools for dealing with spam problems. Further, because there's no creator or owner on Google's systems for any given Usenet group, there's no-one to reach out to, so there's no-one who can (e.g.) close the group to new posts and clean up history, like there is for forums. Thus, unlike with forum groups, Google has no way to contact someone and say "hey, this group is spammy, please fix".
All it takes is someone designing an AI setup to clean out forum groups that are zero signal, and then running it on both Usenet and forum groups to get into this situation; chances are high that nobody involved in this decision has even realised that the two things are different, because they've been merged together a long time ago.
Historical programming-language groups disappearing from Google
I suspect that it's happening because Google Groups is two products merged together:
The old "dejanews" Usenet archives, which I'll call "Usenet" throughout this comment.
Google's in-house forum system, which I'll call "forum" throughout this comment.
It makes sense to remove entire forum groups which have always had a spam problem, and where the forum group owner isn't willing to use Google's tools to moderate it and keep it spam-free; after all, if you've created a forum group for the purpose of spamming, or if you simply gave up the moment spammers found you, there's probably not much non-spam in the group. This is doubly true since the tools have been there since the forum group was created, and advertised to you as the forum group creator; AIUI, Google has reached out to their owner of record for such forum groups and asked them to clean up, so anything left is something that nobody still cares about.
However, that analysis ignores Usenet. Usenet predates Google's spam handling tools (after all, it predates Google), and has never had good tools for dealing with spam problems. Further, because there's no creator or owner on Google's systems for any given Usenet group, there's no-one to reach out to, so there's no-one who can (e.g.) close the group to new posts and clean up history, like there is for forums. Thus, unlike with forum groups, Google has no way to contact someone and say "hey, this group is spammy, please fix".
All it takes is someone designing an AI setup to clean out forum groups that are zero signal, and then running it on both Usenet and forum groups to get into this situation; chances are high that nobody involved in this decision has even realised that the two things are different, because they've been merged together a long time ago.
Historical programming-language groups disappearing from Google
This looks like it spans other groups as well. comp.sys.sinclair also has been "banned".