But that neglects the length of time the problem has been available: a problem that has been solved by 100 people in a year is probably harder than one which was solved by 100 people in the first day.
So some measure of the difficulty of a problem should be how many people have solved it per day since its release. The smaller this number, the harder the problem.
Now, before you all jump on me pointing out that this calculation makes all sorts of unjustified assumptions that are probably incorrect - I know. It's just a bit of fun, so don't take it too seriously.
So here are the - maybe - top ten hardest problems:
* 1: Problem 143 (View Problem) (02 Mar 2007; 514 days ago): solved by 204: 0.40/day * 2: Problem 106 (View Problem) (07 Oct 2005; 1025 days ago): solved by 413: 0.40/day * 3: Problem 88 (View Problem) (04 Feb 2005; 1270 days ago): solved by 512: 0.40/day * 4: Problem 153 (View Problem) (05 May 2007; 450 days ago): solved by 191: 0.42/day * 5: Problem 126 (View Problem) (18 Aug 2006; 710 days ago): solved by 313: 0.44/day * 6: Problem 152 (View Problem) (27 Apr 2007; 458 days ago): solved by 202: 0.44/day * 7: Problem 154 (View Problem) (12 May 2007; 443 days ago): solved by 197: 0.44/day * 8: Problem 105 (View Problem) (23 Sep 2005; 1039 days ago): solved by 485: 0.47/day * 9: Problem 103 (View Problem) (26 Aug 2005; 1067 days ago): solved by 499: 0.47/day * 10: Problem 156 (View Problem) (25 May 2007; 430 days ago): solved by 202: 0.47/day