I don't have the SQL chops to do this query, so I'm hoping someone else might. I'd like to 'bin' all the questions on SO based on how long until they received their first upvoted answer. Not the time to the upvote, the time to the first answer that eventually received an upvote - in other words I don't want to include answers with no upvote in the time calculation.

The bins could be linear or logarithmic, whichever is easier to implement. If someone can even just give me a starter bit of SQL that works on stackql or another existing SO data dump site to work with I'll fiddle with it...

  • 2
    How about those that were later closed as dupes but not quick enough because all the repping users just posted dupe answers?
    – random
    Commented Feb 2, 2010 at 14:42
  • It depends on when it was posted - see the graph in this answer meta.stackexchange.com/questions/3330/…
    – ChrisF Mod
    Commented Feb 2, 2010 at 14:43
  • @random We gotta start somewhere :) Commented Feb 2, 2010 at 14:50
  • 2
    Those at 3000 rep and over should at least pretend to want to clean up the place. Those well under can at least have a shot. @mrs
    – random
    Commented Feb 2, 2010 at 14:54
  • @random: says who? Clean up tasks (such as closing duplicates) is voluntary and not mandatory.. (Like everything else here) Commented Feb 2, 2010 at 15:30

2 Answers 2


I've got a couple of graphs at the bottom of my SO stats page that attempts to graph this. With the wide variance in time to answer, I've used a log scale on the bin sizes.

These graphs don't try to do the "earlier answer that eventually received an upvote", but just "earliest answer" and "accepted answer".

The snippet of SQL the graph uses (against my database, which happens to be PostgreSQL) is shown above the graph.


The result will be different on different tag names. Questions on general topics, for instance, "PHP HTML Entities" get their answers in minutes or sometimes even in seconds. SharePoint questions like this can live for days and no answer received at all.

  • That's true - but this is for a very broad generalization. Once we've got a nice binned set of numbers we can drill down if there are obvious areas that need more exploration.
    – Pollyanna
    Commented Feb 2, 2010 at 15:20

You must log in to answer this question.

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