13

Is it possible to find out how many of the views counted in "viewed n times" at the top left of a question are from users that could have voted?

Some approximation would do, like a rounded fraction or percentage of views of users entitled to vote on the question.

Another estimate would be the count or fraction of views of anonymous users.

I'm allways curious about that when I posted a question, and see, say, 3 upvotes, and 300 views.
That could be bad, with 1% users upvoting,
Or good, with 30% of the 10 users that where able to vote voting up.

0

2 Answers 2

16

This is not going to be feasible.

The view count is not giving you a unique users count. Instead, visits per IP address or (if available) user id are counted, once per time window. See How are the number of views in a question calculated?

My visit here now is one view, if I came back later and the time window has passed, my visit would be counted as another view. That doesn't mean I can vote both times!

If I were to log out, or my family to visit this page (their views coming from the same IP address), all visits within the same time window would be counted as just 1 view.

So all in all, 1 view per IP address or user id per time window could either be multiple visits by a voter or by non-voters, and you can also easily have multiple views per voter.

There is then, no real correlation between views and 'potential voters', and you shouldn't think of views in terms of votes.

Even if this was feasible (ignoring the performance implications for a moment), just because the viewer can vote doesn't mean they have any subject knowledge of the issue at hand. People with +100 association bonus can vote on any SE site, but that doesn't mean I understand if a question on Aviation Stack Exchange is well researched, or if an answer on Personal Finance & Money is actually helpful. So having the privilege doesn't mean the visitor is actually in a position to vote sensibly.

And instead of being helpful, any 'voters' count is going to only further anxiety among new users as to why those voters haven't actually cast a vote on their post.

6
  • Interesting... really looks like it does not work. The value not being well-defined is what breaks it, I think: It could still be that it's possible to derive a rough, noisy approximation of what I had in mind from the data you describe is available - rough as in "1 year ago" - but it's certainly not possible to represent it as a fraction of votes, and maybe not in any other meaningful way. Commented Oct 29, 2014 at 22:09
  • "My visit here now is one view, if I came back later and the time window has passed, my visit would be counted as another view." err... no.
    – Braiam
    Commented Oct 30, 2014 at 23:17
  • @Braiam: right, I forgot IP addresses are replaced by user ids. Still makes it hard enough to track actual users with the view count. Commented Oct 30, 2014 at 23:22
  • @Braiam: and my visits are still counted per time window. Commented Oct 30, 2014 at 23:36
  • 1
    So the view count doesn't show unique visitors. Isn't this actuality one of the reasons why it would be useful to have a separate count for unique users eligible to vote?
    – Fiksdal
    Commented Sep 26, 2016 at 15:18
  • @Fiksdal I'd love to be able to look back at my answer history and know which ones got a high percentage of eligible votes as opposed sorting by vote which is dominated by answers to the questions that made the "Hot Network Questions" list. Commented Sep 30, 2016 at 15:57
3

Martijn Pieters's answer is, of course, correct in saying that the view count doesn't represent unique users.

And that is actually one of the reasons why it would be very useful to have a separate count! Although I agree that the percentage thing would be very misleading, and should be left out. But displaying the number itself seems very useful to me. It is clearly feasible to show (separately) the "number of unique views by signed-in users who were eligible to upvote". You obviously wouldn't count users twice in that number.

IMO this would be quite useful information. You could even leave out from the count people who had spent all their daily votes. Of course, it might not be trivial to implement, but it seems clearly feasible. I think this would be an awesome feature, which would give us a very useful metric on how the community received questions. Trivial or not, I would personally love the feature.

Similarly, it would be possible to have a count for "signed-in users eligible to downvote".

If displaying these metrics would be an additional strain on the system, they could be an on-demand feature. (Like the votes up/down view is today.) It could even (totally optionally, I'm not necessarily advocating this) be a privilege one earns at a certain amount of reputation. This would be another way to encourage users to participate on the site and earn reputation.

In conclusion, this sounds like a very useful, cool and positive feature. Please implement it.

I've made a rough sketch of what this might look like. Just a draft, of course. Some details probably need to be changed.

enter image description here

4
  • 2
    This is simply not possible, because we don't keep a list of all past view records. The views count is denormalized - it's just a number. The user entry that prevents a user's view being counted again for the ~15 minutes gets completely deleted once that time lapses, so it's not possible to know if that user has already been counted as a potential voter or not without keeping those records forever, which is never going to happen, ever.
    – animuson StaffMod
    Commented Sep 27, 2016 at 6:04
  • @animuson Do you mean that it would be non-trivial and demand new stuff to be developed to make it happen, or do you mean that it's actually not possible?
    – Fiksdal
    Commented Sep 27, 2016 at 13:34
  • The only way to make it possible is to murder our database performance.
    – animuson StaffMod
    Commented Sep 27, 2016 at 14:38
  • @animuson Alright, thanks for clarifying that.
    – Fiksdal
    Commented Sep 27, 2016 at 14:46

You must log in to answer this question.

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