I saw a really bad answer to a question on StackOverflow the other day with many downvotes. I've just been wondering if there is a defined limit to how low a question's score can go.


The lowest score a post can get is equal to the number of users who have at least 100 reputation, minus 1 if the question was asked by one such user. That is, if everyone who could went and downvoted the item.

Note that the actual lowest score on the site can be less than this amount. Downvotes aren't retracted if a user ends up below the threshold, so the actual possible lowest score is equal to the largest number of users to simultaneously exist (not consistent with the number of users in total, due to account merging and deletion). The above number would be the more reasonable and practical one, this would be your "theoretical lowest".

It's unfeasible for such a question to survive long enough without actually getting deleted (it would have to be a monster to get THAT low), but that's not automatic. If it does survive for that without intervention, it's rather unreasonable to arbitrarily limit how many people can cast an opinion on a post. The same would qualify for how high a post can go.

  • 2
    I just remembered that migration actually allows double-vote. So the theoretical lowest score on S[OFU] is the same across all sites and is equal to the number of simultaneously existent users across all 3 sites, while the lowest score on Meta is equal to that number plus the users of Meta (since you can't migrate out of Meta). – Grace Note Jul 7 '10 at 14:40
  • 1
    @Grace: Interesting. What if a question is migrated repeatedly between 2 sites? – kennytm Jul 7 '10 at 14:44
  • 2
    @Grace: no, remember that now migration between sites removes the votes if it has negative score. – Ether Jul 7 '10 at 14:49
  • @Ether Thanks, I missed that announcement, I was just going by my memory of things before my time. So, to that point, the theoretical lowest score is what I left in my post body, then. – Grace Note Jul 7 '10 at 14:51
  • 2
    So, the answer is: "In theory, yes. In practice, no." – Vivian River Jul 7 '10 at 14:55
  • @RisingStar I could not have phrased it better myself, even with ludicrous thought exercises (which, by the way, thanks for inspiring them). – Grace Note Jul 7 '10 at 14:56
  • There are almost 7 billion people in the world. Suppose that they all registered for SO accounts at once. How many "openIDs" could be in the system at once before the server melts down? What about the internet infrastructure to accommodate everyone registering at once? The theoretical limit could possibly be well below what you describe when Armageddon comes :-) – Vivian River Jul 9 '10 at 13:32

I guess the absolute minimum is -2,147,483,648 (i.e. no limit).

As of today there are 37,086 users having rep ≥100 on SO so the minimum would be -37,086 today.

The lowest scored question still not deleted is Create A Compiler For Windows standing at -76.
The lowest scored answer still not deleted should be this answer at -48.
(see also What are the most upvoted/downvoted questions and answers on the sites?.)

I don't think any non-community-wiki questions/answers will survive below -100.

  • 1
    -2,147,483,648 is not the same as no limit, nor is it safe to assume that that is the minimum allowable value by the architecture and data storage mechanism in use. Many databases use lower limits, and many languages do not cram its integers into 4 bytes. – Ether Jul 7 '10 at 14:51
  • @Ether: The minimum of an int of SQL server is -2^32. msdn.microsoft.com/en-us/library/ms187745.aspx. The SO database is running on SQL server. By "no limit" I really mean "no artificial limits". – kennytm Jul 7 '10 at 15:07
  • the intent of my pedanticism was to coax you to qualify your answer. My work here is done. :) :D – Ether Jul 7 '10 at 20:45
  • How did this Windows Compiler question get back to score 0?! – Felix Dombek Jun 15 '11 at 10:18

You must log in to answer this question.

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