21

In Delete old, unloved, users I argued that we should delete unloved users because it's a fairly good way of getting rid of those whose sole purpose is spam, as per Is spam in the "about me" field a reason to worry?.

However, this was back in prehistoric times and far more harshly than I would have done; can't complain! Shog9 wrote in his answer and a comment that the number of users that are deleted each iteration is capped:

Note that due to the extreme number of qualifying accounts on Stack Overflow, only a fraction of these actually get deleted on each pass.

... Yes, @Ben. I forget the number, but it is capped.

As there are 8,404,010 (up from 237,625 in 2012) who reach the minimum qualifying criteria 1 can this now be done better? This can also be read as I don't believe this is "completed" if there's this number of users who qualify.

By better I mean can we do one of:

  1. Prioritise those who've added a website or About Me (1,401,281), thereby getting rid of the only users who both fit the criteria and are capable of spamming via these fields.
  2. Prioritise those who've had posts deleted, and therefore have contributed detrimentally to the site.
  3. Delete more users each time the script is run; or run the script more regularly.

1. 1 reputation, not visited for 6 months and no posts - I can't test for other accounts so it's less than this but for cookie based accounts the minimum reputation is higher so this is roughly accurate.

7
  • 6
    Is this a serious problem? If a user is dormant, they're not really contributing much overhead even if they're a spammer. Their profile isn't very likely to be discovered and viewed, so I'm not seeing much harm. What am I missing? :)
    – Adam Lear StaffMod
    Sep 9, 2012 at 22:12
  • No :-)... having seen the linked question bumped a few weeks ago it was just something I was considering. It seems like it would be a very easy fix and as it's something that has obviously been implemented before why not fix it? Sep 9, 2012 at 22:15
  • 2
    Do note that the criteria for a user to be auto-deleted has changed since then (and was disabled completely for a year prior to the new rules).
    – animuson StaffMod
    Dec 8, 2021 at 15:31
  • I don't think data.SE has enough info for any of us to estimate the number of users @animuson? I suspect the changes still keep the numbers over 5m. Dec 8, 2021 at 22:57
  • 2
    Yeah, I was trying to find the actual query in the code because I was curious what the actual backlog was. But I couldn't find it and am not curious enough to write that entire query myself. It's not in the usual place for scheduled tasks so I've no idea where it is, if it still exists.
    – animuson StaffMod
    Dec 8, 2021 at 23:40
  • 2
    After doing some more digging, I'm not convinced we are actually deleting inactive users at all, and Sonic's edit to that post may be incorrect. I was able to track down a really old card where we were investigating how many users would be impacted by such a query. But I'm not able to find any queries anywhere in the code regarding inactive users being culled, there are no relevant site settings, and after scanning several months of GitHub history around that time, I've found no relevant changes being made to the code. I don't think this task ever made it into production code.
    – animuson StaffMod
    Dec 9, 2021 at 4:45
  • That'd explain why 50% of users meet the previous deletion rules then! Dec 9, 2021 at 12:48

1 Answer 1

10

I have confirmed that we do not currently run any automatic deletions on any sort of schedule and we do not have any plans to reinstate such a schedule in the future. Writing queries to detect users to qualify for deletion while maintaining the necessary exclusions is too resource-intensive to run through our scheduler, even at the previous rate limit of only 1,000 users every week.

The most recent criteria for automatic deletion were used for a one-time batch job that was successfully executed and deleted millions of users. However, we never implemented those criteria on a schedule and if we do continue deleting users based on those criteria, it will be done in manual batches at our discretion.

I'm going to decline this request because it's a bit moot - we don't have an automated process that can be improved anymore and don't plan to implement one again.

0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .