10

Related: How do the “arbitrary hotness points” work on the new Stack Exchange home page?

I'm new to the Hot Network Questions game (the site on which I am most active recently had it turned on), so I would have thought this question had been addressed already, but my searches both locally and on Google have failed to provide me an answer.

Recently the aforementioned site had two Hot Network Questions:

  1. one that was quickly answered with two high-quality answers by people other than the OP. (link)
  2. a second that was quickly answered with a self-answer, then followed a few minutes later by an answer by someone other than the OP. (link)

Since then, I posted a question with a self-answer. In my case, both question and answer were heavily upvoted (at least in comparison with the site average), however, no other answers were given.

Running the numbers of my post through the formula in the above linked Meta post, and comparing that number to the values that served as the threshold for the previous HNQs, it appeared that my question should have made the list easily.

So I'm curious: did I make a calculation error, or not account for some variable not included in the formula? Or does the HNQ formula discount or disqualify questions that have only a single self-answer?

Update: This question, also with one answer, currently has similar scores to my question after the same amount of time, and it does appear on the HNQ list.

  • Nothing definite, but the difference may lie in your question having only one answer, not two as the other examples. The number of answers is relevant for HNQ. – Wrzlprmft Nov 11 '15 at 13:11
  • 1
    Right: I took that into account when putting the votes through the formula: ((1 Answer * Question Score / 5) + Answer Score) / Age factor. Certainly harder to do with only one answer, but this one seemed to have the votes to pull it off. – Nathaniel Nov 11 '15 at 13:15
4

There is nothing in the code that special cases questions with self-answers. However, the formula does reward questions with multiple answers (up to 10). As a result, an excellent question that already has an excellent answer tends to be hampered in the first term of the numerator. Questions with one answer also are hampered by the second term since it sums the score of all answers. So a question that gets two solid answers has an advantage over a question with one, all other things being equal.

In addition, self-answered questions tend to get fewer question votes than other questions, in my estimation. Often that's because the question is asked in a way that sets up the self-answer rather than prompting more answers. They are, in a literal sense, rhetorical.

As to why one question with a high score might be chosen and another not, there are several other factors in play:

  1. Nearly every other site in the network has a crack at getting questions into the limited Hot Network Question list. So if there are two question with the same hotness factor, but one is asked on a quiet day and the other on a busy day, the first might make the list and the later fall short.

  2. I don't think this is a factor in this case but:

    Succeeding questions from the same site are penalized by increasing amounts. So, the first question from SO in the list gets multiplied by 1.0, the second by 0.98, the third by 0.96, etc)

  3. The denominator is time-based in order to make questions less hot as time goes by. So a question that gets its answers and votes in early has a substantial advantage over a question that acquires votes over time. Looking at the end state gives you a hint about how hot it was, but can't give the full picture. For that, you need to look at vote timing.

So to answer the question, self-answers are treated like any other answer in the code, but might have non-technical handicaps.

  • 1
    Thanks for looking into this Jon. Unfortunately I'm not convinced; I've seen numerous examples of self-answered questions not appearing in the list at the same time as others with same vote counts, answer counts, age (<6 hours), and similar site traffic levels. If you don't mind, I could ping you with an example the next time I see one, though I understand if this isn't worth more of your time. – Nathaniel Dec 2 '15 at 21:38
  • @Nathaniel comparing questions at different small sites can't be reliable. Thing is, they just lack room for all sites: there are only 100 slots in HNQ but much more than that, about 150 sites in SE network. So, when a small site question with slightly better score enters the list, there may be just no place for a little bit less lucky one from other site – gnat Dec 2 '15 at 22:51
4

I currently see a self-answered question on the HNQ list:

Self-answered question

The difference between this and previous self-answered questions I have seen with sufficient votes is that this one was not self-answered at the same time as it was asked. By posting the question and answer separately, an apparent caching issue is circumvented and the question can appear in the HNQ list.

You must log in to answer this question.

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