15

Lately a few folks have mentioned that they were unable to suggest edits because the queue itself was full. Apparently, the Suggested Edits queue is only allowed to contain 40 items.

That seems low for Ask Ubuntu which is busier than all but two of the other Stack Exchange sites (Stack Overflow's queue has a capacity of 500).

People frequently hitting the queue capacity limit before the per-user limit (a user is only allowed to have five pending edit suggestions) seems bad to me. While we could definitely use more active reviewers of several of the queues including Suggested Edits, I believe suggestions would still get cleared promptly if the capacity were increased. According to my stats about 1320 reviews have been done in the last 30 days (which is 44 per day). I think it would be safe to increase the capacity of the Suggested Edits queue to at least 50.

What do y'all think?

5
  • How about increasing the limit to 70?
    – Random Person
    Commented May 30, 2021 at 9:22
  • 2
    @RandomPerson Not sure what the optimum number would be, but I think increasing it to 70 might slow things down too much at the current rate of review. Maybe 55 or 60 would be ok
    – Zanna Mod
    Commented May 30, 2021 at 10:35
  • 1
    Why would any increase slow things down? Surely increasing the number of items pending review can only speed things up or, at worst, make no difference. How could it ever cause a slowdown?
    – terdon
    Commented Jun 2, 2021 at 12:59
  • @terdon it could mean that the time taken for any individual suggested edit to be approved or rejected would increase, as that particular suggestion would have a smaller probability of being the next one to be reviewed due to the larger number of pending items
    – Zanna Mod
    Commented Jun 2, 2021 at 13:03
  • 2
    Stack Overflow's queue has a capacity of 500, but it also has a reviewer limit of 40 suggested edits per day instead of 20 which helps to make the larger suggested edit queue more manageable. It should be noted that in Stack Overflow the reviewer is often tasked with reviewing changes in code which takes longer per suggested edit reviewing task.
    – karel
    Commented Jun 3, 2021 at 7:22

2 Answers 2

Reset to default
8

I think the immediate need for this has passed but let's look at some data since we're here and I've been looking at data.

This is that graph Zanna mentioned in her answer. This is suggested edits per week from July 2019 until today:

Graph of suggested edits per week over time starting in July 2019 until today. There are three lines being charted, registered user edits, anonymous user edits, and total of the two. The average number of suggested edits per week is around 100-250 from the July 2019 until April 2020 then there's a huge increase where there were suggested edits in the 350-400 range some weeks down to 200 other weeks. from April 2020 until October 2020. At that point it goes back to an average between 100 and 200 per week. There's also a big spike in anonymous edits in July 2020

I'm guessing that I'm looking at a slightly wider window that Zanna did because, while in the last year, there's been a decrease, the average per week is actually closer now to what it was in 2019. I'm not quite sure what was happening midyear in 2020 to cause such a huge increase in suggested edits, particularly from anonymous users at that time but... there you have it.

I do agree with the assessment is that fewer people are reviewing - or at least, that fewer reviews are happening. Based on another query, I can see that y'all did have a period where you were hitting 40 on some days, and that's when this post was written. The immediate impact of this post, though, is that y'all haven't hit 40 since.

27540 - The number of reviews in the queue per week, maximum, minimum, and average. January 2020 until today. Prior to March, the high number of reviews at any given time was around 25-30. Starting in March, the high count of reviews within a week was closer to or at 40.

The chart above shows the maximum, minimum and average number of reviews in the suggested edit queue over a week. You can see that there were two recent weeks that reached 40 and two others that got very close. When we zoom in, as in the next chart, you can see that the last time y'all hit 40 was on May 24th, but, over the last week, y'all have stayed at almost half the cap, even at the max, with the recent high on June 6th of 25.

27540 - The same graph as before but in 1-day intervals from May 10th until June 8th. Four days (May 20, 21, 22, and 24) hit 40 reviews at one time but the max has been lower since.

I'd also point out that y'all are clearing the queue entirely or nearly at times, which wasn't as common since the big increase in pending reviews in March. So, good job to you all for answering the call, and thanks to Zanna for bringing this to the community's attention.

So, since it seems like y'all have it under control, I think I'm going to leave the setting where it is for now. Let me know if you have any questions. :)

2
  • 1
    Thanks for the info! It would be interested to also see the evolution of the number of reviewers for the Suggested Edits queue over time, if such metric exists.
    – BeastOfCaerbannog
    Commented Jun 9, 2021 at 11:05
  • 2
    Here's what we were looking at
    – Zanna Mod
    Commented Jun 10, 2021 at 4:20
8

jokerdino pointed out some more data to me, and we noted that the number of suggested edits being made per day/week/month has actually been gradually declining at least since 2015 and is currently at a particularly low level. The fact that people are running into the queue size limit in these circumstances indicates that reviewing of the queue has declined even more than suggestions.

In the ancient past, there was a number in the top bar indicating pending suggested edits, to encourage people to review them promptly. During my time, thousands of years ago, I feel like that number rarely reached double figures, although there were more suggested edits per day back then. The redesign of the top bar both removed that indicator, and also made the review queues less discoverable by changing the labelled button (it said "review") to a mysterious icon. This may partly explain the decline in reviewing, in addition to many people having recently reduced or redirected their activity (including me - these days I mostly handle flags and stuff).

Therefore, in addition to increasing the queue's capacity (or even if we leave that as it is) we need more reviewers!

6
  • Can the "data" be made public?
    – Random Person
    Commented May 30, 2021 at 10:36
  • 2
    Maybe I can find a public source... the page we used has got "do not share the details publically" written on it
    – Zanna Mod
    Commented May 30, 2021 at 10:38
  • Is the data a part of mod tools?
    – Random Person
    Commented May 30, 2021 at 10:41
  • 2
    Yes it's buried in our secret broom cupboard
    – Zanna Mod
    Commented May 30, 2021 at 10:59
  • One reason for declining numbers of reviewers might be those pesky audits (especially the vandalizing ones with the extra loading time) - they break my workflow and remind me how much time I have already spent with this task. I do know that you have no say in that case but once in a while I have to get that off my chest :-(
    – guntbert
    Commented Jun 7, 2021 at 21:13
  • As a note, the indicator isn't gone... it's actually quite visible still - and the suggested edits queue (if I remember correctly) is the queue with the smallest number of items needed to turn the indicator on (and red). :) The numbers in the dot are gone, though - because they were nearly always incorrect, which people constantly complained about. :P
    – Catija Staff
    Commented Jun 8, 2021 at 18:03

You must log in to answer this question.

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