42

Is there any data about which Stack Exchange sites have their questions end up in the Hot Network Questions list the most? In particular:

  1. What percentage of hot network questions comes from each site?
    1. What about all-time, this year, this month?
  2. What percentage of questions from each site end up in the list? (It seems that a high percentage of all CodeGolf.SE questions end up here.)
  3. Which site's questions stay on the list the longest?

And so on. Is there any data on these and similar questions?

6

1 Answer 1

46
+200

It's about time for an update to this; the last one was from before some major changes to the Hot Network Questions and the world has changed a lot in the past year. This time, I'll be looking at data from April 2020 up to March 2021; see the revision history for August 2018 and August – December 2018.

Stack Exchange itself only registers (since February 2019) the moment a question hits the Hot Network Question list, not when it disappears (unless a ♦ moderator removes it manually). I have some additional data (including scores and views) from August 2018 onwards, which has been collected into a PostgreSQL database which can be found here. The main information is in the snapshots table, which is just the list of 100 Hot Network Questions downloaded every 10 (later 3) minutes.

Note that some external events might cause a spike in a site's popularity; e.g. I remember that during the Falcon Heavy test flight, the number of HNQs from Space Exploration skyrocketed (pun intended).

  1. What percentage of hot questions comes from each site?

During this period, 4.76% of the Hot Network Question spots was composed of Worldbuilding, followed by Role-Playing Games with 4.06% and Academia with 3.88%. The top 3 is unchanged since the initial report period in August 2018, but the percentages have dropped significantly (presumably because of the limit of 5 questions per site which was introduced in 2019).

Rank Site % of HNQ spots # of (distinct) HNQs
1 Worldbuilding 4.76 1515
2 Role-playing Games 4.06 1820
3 Academia 3.88 934
4 Physics 3.52 1433
5 Science Fiction & Fantasy 3.50 1331
6 Code Golf 3.14 586
7 Electrical Engineering 2.79 1459
8 Politics 2.78 662
9 Puzzling 2.62 1433
10 Home Improvement 2.29 969

Note that the number of HNQs is slightly different from the one in the next list (generated by SEDE) because of two things: 1) my script missed about two days in this period. This is less than 1% of the total period so I don't think it has much influence. 2) questions which entered the list before 2020-04-01 but were still in it on 2020-04-01 don't appear in the SEDE query. Again, not much influence I hope.

  1. What percentage of questions from each site end up in the list?

You were right, almost half of the Code Golf & Coding Challenges questions become a Hot Network Question. The site is closely followed by Retrocomputing. The top 2 is the same as in 2018 but percentages are slightly higher. This can now be answered with a SEDE query:

Rank Site % of questions that hit HNQ # of HNQs
1 Code Golf 46.67 582
2 Retrocomputing 44.09 522
3 Latin Language 40.02 329
4 Role-playing Games 38.58 1834
5 Puzzling 38.56 1442
6 Chess 35.40 496
7 Bricks 34.39 195
8 Worldbuilding 30.64 1528
9 Operations Research 29.87 302
10 Bicycles 28.70 729
  1. Which site's questions stay on the list the longest?

The winner here (in this time period) is Parenting, but I suspect the results here can rather fluctuate between time periods (though they were the winner in 2018 as well). Their 71 HNQs were (on average) 'hot' for a period of well over 2 days. This is a lot less than in 2018, because there's a hard limit of 3 days now.

Rank Site Average time (hours) # of HNQs
1 Parenting 57.91 71
2 Skeptics 52.27 142
3 Code Golf 46.55 586
4 Information Security 40.57 322
5 The Great Outdoors 38.84 44
6 Movies & TV 38.38 153
7 User Experience 36.95 82
8 Politics 36.48 662
9 Freelancing 36.45 2
10 Academia 36.13 934
0

You must log in to answer this question.

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