65

Ever since Everything That Happened with Stack Exchange, it's understandable that many users are hesitant to get re-involved on SE. It's understandable that many users are deleting their own accounts, out of protest, or out of frustration, or for many other reasons.

But I have noticed that, at the very least on Writing.SE, where I spend most of my time, questions are getting fewer views and fewer answers, fewer votes are being cast, and fewer questions themselves are being asked. I've also noticed that flags are taking a long time to be processed, probably because of the fact that Writing.SE has no active mods.

Are other sites also experiencing inactivity? Is this in direct correlation with Everything That Happened? Is it just this site, or are other sites experiencing the same patterns as well?

4
  • 12
    Outdoors.SE has become something of a ghost town and things are taking way way longer to be reviewed. Commented Oct 27, 2019 at 20:11
  • 2
    The "traffic" stats in the answers posted so far are probably not the best indicator. The events of the past month are really only of interest to the core community, and much of the traffic comes from outside sources. A better indication would be the numbers for votes and answers, both of which are provided mostly by the community.
    – Mark
    Commented Oct 27, 2019 at 20:28
  • 2
    @ab2ReinstateMonicaNow I had forgotten all about the analytics data so you're right, it wouldn't be either the reasons I suggested.
    – Script47
    Commented Oct 27, 2019 at 21:33
  • If those with access to site-analytics could base their analysis please also on the child-metas? Especially those sites with mods resigning see a huge difference there and extraordinary traffic and voting on these posts? Be it Mi-Yodeya, BibHermeneutics, Skeptics etc. (Where I do have access I see general trend-lines downwards for longer than "recent events") Commented Oct 28, 2019 at 19:29

10 Answers 10

38

Many people are focusing on questions, which is always a very low stat on Writing.SE and didn't see much change - there is not much that could possibly change there. Traffic is also popular, but didn't see a lot of change, as most people are still coming in via google.

But according to the site 5k rep analytics the number of votes dropped to around a third (low activity days) to a fifth (high activity days). I think "number of votes" would be a better metric to show how the situation is affecting the core users who are mostly providing answers and voting / editing / flagging / ...

The people that are contributing site views won't know about this. And many prolific askers aren't that active or interested in meta or other things happening on the network. The people mostly answering are the ones that are more likely to participate on meta. Of course there are exceptions to that, but in general that seems to be the trend. At least that's my experience, which is mostly based on WorldBuilding.SE and Writing.SE

I think this query that I found on Interesting queries on Data Explorer could be easier for relevant stats. It gives everyone an overview similar to the 5k / 20k rep "mod tool" site statistics, so that everyone can check every site they are interested in and see how the situation affected the sites you care about.

Here is an example for Writing.SE for the last year; you can see a big spike in march because of a contest to get past 10 questions per day - and a sharp dropoff the last weeks, especially when it comes to votes; normally the "votes / 7 (for scaling)" line is above the "answers" line, now it's below that line:

Daily site activity on Writing.SE

3
  • 1
    That's very interesting. Running that query for Christianity SE I see a noticeable drop-off in questions and especially answers for the last few days in September and first week or so in October - but an equally noticeable surge in votes for the exact same period. Commented Oct 28, 2019 at 1:42
  • 16
    This makes perfect sense to me and I suspect it is happening elsewhere but I have no easy way (or perhaps "any" way - not really sure) to check that. My gut feeling is: questions are asked by either newbies (who aren't affected by the current system crises as much) or established users who are really looking for serious assistance (who will ask anyway since they need the help!) But answers will go down more - the frequent users just aren't in as much of a mood to help and votes even more - why bother if you don't care as much any more. I know it has affected me. Commented Oct 28, 2019 at 5:39
  • 1
    @manassehkatz Some personal weaknesses aside, work-to-rule is the motto. Neither questions, nor answers, nor community moderation. As the ToS specified: SE has 90 days. If until then they don't get it together satisfactorily, then it's U+1F594. Commented Oct 29, 2019 at 21:37
26

It's near impossible to tell on your timescale and sample size.

At this point, it's completly unclear if for instance on MSE you just got used to "peak outrage activity", that is, 10x-20x as much as usual, which is now starting to die down.

For other sites, you probably need a month or two more to be able to pull actual stats, but the subjective feeling that they're becoming less active could also be influenced by your opinion or perception of recent events, in an effect not dissimilar to the placebo effect.

It's too early to tell.

2
  • 2
    +1, I agree with this assessment. But I'd like to add, the reason I'm asking is to make sure it's not just me being subjective. If other people are noticing the same patterns, I don't think it's too subjective.
    – user473022
    Commented Oct 27, 2019 at 19:24
  • 3
    @weakdna I don't think there's a way to decouple individual meta user's asessments from the situation and its effects at the moment. (nearly) everyone is in the same situation as you.
    – Magisch
    Commented Oct 27, 2019 at 19:34
13

Here's the data for Retrocomputing, re-scaled based on the traffic levels. Yeah, it's pretty obvious that something happened right around the start of October.

A chart of answers-per-visit and votes-per-visit data for the Retrocomputing Stack Exchange site.  Traffic levels are reasonably constant from mid-July 2019 until the start of October 2019, at which point voting declines to about 75% of earlier levels, while answers decline to about 30%.

8

Here is the traffic graph for Ask Different. It looks like traffic was a bit lower the week after 'the event', but subsequent weeks seem quite normal.

enter image description here

Number of posts and votes follow similar patterns. Most of the traffic is from search engines, i.e. users who are completely unaware of current events; less so for the posts and votes but even there, most users in the network don't have a clue about what's going on.

I've checked Chess Stack Exchange as well, but there it's completely unnoticeable, probably because that site gets most of its incoming traffic from Hot Network Questions.

3
8

I took a look on the data at Interpersonal Skills (IPS).

Here is what I saw on the traffic graph:

Traffic graph of IPS

I took the time to cross-reference those data with our HNQ feed and it seems that the huge drop of traffic between 20 September and 02 October is due to the fact that we were out of HNQ (aka: none of our questions were "hot" enough to go to HNQ).

Those results don't surprise me. Every HNQ events that we had proved, again and again, that our traffic is very highly related to us being visible by the rest of the network (via HNQ).

So, if "Everything That Happened" had any impact on IPS traffic, it was not significant enough to be visible here.


However (but I have no concrete data to prove it), it did seem like we had more questions related to (trans)gender/lgbt+ than we usually have (but this might just be a confirmation bias of mine).

8

Are other sites also experiencing inactivity? Is this in direct correlation with Everything That Happened? Is it just this site, or are other sites experiencing the same patterns as well?

I can't speak for all of them, but I can give a partial answer for Politics.SE. According to Area 51 stats, we're now getting 10.5 questions per day. If I understand correctly, that's some rolling average and it's up from earlier in the year.

To follow Glorfindel in posting traffic. I see the same dent, though I'm not sure if it can be attributed to the debate on Meta: enter image description here

And in keeping with Churchill's "Never let a good crisis go to waste", we even had some questions sparked by the debate on meta:

Is there a push, in the United States, to use gender-neutral language and gender pronouns (when they are given)?

Are countries other than the US concerned about preferred pronouns and how they relate to gender?

Do European politicians typically put their pronouns on their social media pages?

Also adding the number of votes as another answer did. It seems Politics hasn't really dropped by that metric either:

enter image description here

1
  • 4
    To be fair, a lot of these questions are due to Brexit and Trump. Commented Oct 29, 2019 at 23:38
4

On the two sites that I have access to Site Analytics on (Mi Yodeya and Science Fiction & Fantasy), I don't really see enough significant changes in traffic or posts that would be easily attributable to recent Meta events. There is a lot of day to day fluctuation (usually based on weekday vs weekend or different times of day) but it's always been that way.

Mi Yodeya

Posts

Screenshot of Site Analytics graph

Traffic

Screenshot of Site Analytics graph

Science Fiction & Fantasy

Posts

Screenshot of Site Analytics graph

Traffic

Screenshot of Site Analytics graph

2
  • "On the two sites that I have access to Site Analytics on (Mi Yodeya and Science Fiction & Fantasy)" - OK, I know who wrote this post :-) Commented Oct 27, 2019 at 21:01
  • @Randal'Thor Indeed, I believe I am the only person on the network with this particular combination.
    – Alex
    Commented Oct 27, 2019 at 21:02
4

A quick count of questions on Christianity SE (which lost 2 of 5 moderators for religious reasons at the end of September) reveals 89 questions between 28 September and 27 October, and 82 between 28 August and 27 September. We did however have the first LGBT+ questions I can remember in five and a half years on the site.

3

Here's my crappy data explorer query as a response to this question (with credits to one of Glorfindel's old queries for most of the code). Basically, as everyone else has been saying, it's way too early to tell. I set a 'before' sample as the month of August, and then the 'after' sample as September 27th (the date of Monica's firing) to today, October 27th.

Criticism and changes are welcome to the query - SQL is definitely not my language of choice.

4
  • 1
    Just a little comment: the query only focuses on questions, which is always a very low stat on Writing.SE and didn't see much change - there is not much that could possibly change there. Traffic also didn't see a lot of change, as a lot of people are still coming in via google. But according to the site 5k rep analytics the amount of votes dropped to around a third (low activity days) to a fifth (high activity days). I think "amount of votes" would be a better metric to show how the situation is affecting the core users who are mostly providing answers and voting / editing / flagging / ...
    – Secespitus
    Commented Oct 28, 2019 at 0:40
  • 1
    I think this query that I found on Interesting queries on Data Explorer could be easier for relevant stats.
    – Secespitus
    Commented Oct 28, 2019 at 0:49
  • @Secespitus that should be an answer; I'd gladly upvote it.
    – auden
    Commented Oct 28, 2019 at 0:50
  • I added my answer here.
    – Secespitus
    Commented Oct 28, 2019 at 1:04
0

I see no change in the posted SO analytics, fyi.

Perhaps there was a 10% drop in votes and posts (and not traffic), during the first week of October only.

You must log in to answer this question.