1

This question already has an answer here:

I'm wondering whether the reputation limit in combination with a global clock doesn't create the following effect:

"The quality of the answers is the highest around 00:00 UTC and drops until 23:59 UTC." - idea

The rationale behind it is that each day, at 00:00UTC a new "SE day" begins. So everybody has zero reputation that day. But as time passes, more and more (high reputation) users earn the maximum reputation limit. The result is that some/most of these users don't have an incentive anymore to answer questions (decently; no insult intended at all). So the number of potential answerers (as well as the answerer-quality if reputation is a good measure) drops.

Of course the claim not fully holds: people live around the global and most people won't answer questions in the middle of the night, simply to earn some reputation. Still there is probably some effect... Let's assume people work 8 hours everyday (and consult the site occasionally through the day), then the last person that become active on the site the first time that SE day does this at 16:00 UTC, so from that moment, the quality should start dropping significantly. Furthermore it depends on the concentration of IT-minded people on the globe (one expects the quality to drop faster if people in San Francisco, California are at work, since the concentration of programmers is higher in these regions).

In other words, the best time to ask a question is around 00:00 UTC (based on probability reasoning, and probably later that day based on actual data). This can create problems, since most programmers don't decide when they want to run into trouble with their work. And furthermore people living in different countries are differently aligned to the global time.

A possible solution to the problem - if any - could be that the "SE day" depends on where a user lives. There are probably some issues because users could change their location to tamper with the clock, but perhaps other solutions might work as well.

Please note that I don't claim it is the "intention" of answerers to answer less questions. It's more a subconscious process that people tend to put less effort in things where they don't receive (immediate) rewards.

marked as duplicate by random discussion Jan 7 '15 at 0:52

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • Do you have any evidence to support this theory? – psubsee2003 Jan 6 '15 at 23:59
  • @psubsee2003: no, since users probably don't have access to the votes. Perhaps moderators can somehow count the total vote balance through the day algorithmically and thus see a graph of (well at least some kind of) answer quality. Or do you know where one can have access to anonymous vote records? – Willem Van Onsem Jan 7 '15 at 0:01
  • No one has access to the voting data except developers and they can only do that but actually going to the database. But you are proposing an idea to fix a problem that you don't even know exists. – psubsee2003 Jan 7 '15 at 0:02
  • @psubsee2003: well one only needs access to the total votes per answer. And indeed I have no firm proof, is more based on probability theory, but of course that doesn't (fully) hold if there are dependencies between the variables. That's why its a discussion: do other users have the same opinion about it, and how can we solve it (if any)? – Willem Van Onsem Jan 7 '15 at 0:03
  • @psubsee2003: hmm, looks the SO dump has a score field. Will generate a graph of the average answer score in time... Give it some time please... – Willem Van Onsem Jan 7 '15 at 0:08
  • The biggest flaw in your logic is you are assuming most rep is earned the day the answer is posted, whereas most rep is residual and earned over the lifetime of the post. – psubsee2003 Jan 7 '15 at 0:08
  • @psubsee2003: the x-axis of the graph shouldn't show when the votes were casted, but when the answer was created of course. So votes on the long term, only make it a bit more robust, but there is of course an additional flaw since popular questions tend to result in more votes (regardless of the quality of the answer). – Willem Van Onsem Jan 7 '15 at 0:11
  • 1
    I cannot help but notice that you asked your question almost directly at 00:00 UTC. Unfortunately, this is not the best time to ask because most users are in the US, and it is basically the end of the day here (people are getting home to their families, or at least leaving work). – Travis J Jan 7 '15 at 0:14
  • @TravisJ: Well it was really a coincidence (well the idea was probably triggered by 00:00 UTC anyway) ;). Your claim will probably indeed hold. – Willem Van Onsem Jan 7 '15 at 0:23
  • 1
    Very interesting observation! – One Face Jan 7 '15 at 0:45
  • @random: The (name of) the question seems to suggest this is a duplicate, but nor in the question/answer the reputation seems to play a part. Perhaps the questions should be merged/extended? – Willem Van Onsem Jan 7 '15 at 0:56
  • 1
    Another carry from meta.stackexchange.com/questions/20998/… – random Jan 7 '15 at 1:06
  • @random: indeed that's probably a better duplicate candidate. – Willem Van Onsem Jan 7 '15 at 1:08
3

I believe this occurs, but not at the linear level described. The majority of activity happens during the US normal business hours, basically somewhere from 6am PST (9am EST) to 5pm PST ( 8pm EST ). I don't think the UTC time window has any significant affect on activity.

In order for reputation to be accrued there has to be new questions. Looking at the graph of new questions which accept answers within the first hour would be a good indication of the curve, and I would be more inclined to believe that the peak would be at the middle of the time window noted above.

Further, I would like to point out that the problem being raised is only even relevant with users active enough to actually hit the reputation caps on a daily basis. While these users are exceptional contributors, I do not believe having them avoid answering for the small time window remaining once they cap is an actual problem. As noted in other places, the reputation cap is to allow users like this some solace from the exchange.

  • Well the users probably don't avoid answering questions. I would compare it with blood donations in Europe, you don't get paid for them, but if you donate regularly, you get free cinema tickets,... Now I don't think people start donating for that cinema ticket they receive after two months or so, but you will see some decrease if you would abandon it: people would donate less strictly. – Willem Van Onsem Jan 7 '15 at 0:14
  • @CommuSoft - There is definitely less incentive, so they probably do not actively seek questions to answer as frequently; this is what I was referring to by avoid. – Travis J Jan 7 '15 at 0:16

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