Since I have several suggestions, I'm just going to combine them into this one post:
[ol]
[li]Would it be possible for the contest judging system to also have other scoring systems, like the IOI system (i.e., sum of scores out of 100 for each problem, maybe with breaking ties by time) rather than only the ACM system? This seems to have already been implemented for several contests, such as the Open Contests and the HSPL, so I hope it won't be difficult to set up. 
EDIT: The admins seem to be working on this one now. (Thanks!)
[/li]
[li]This one isn't quite a big deal
, but it would be nice to be able to stay logged into SPOJ for a longer period of time (maybe indefinitely?).
[/li]
[li]Here's another idea, but I wouldn't be surprised if it has been suggested before: users could rate aspects of each problem (difficulty, quality, etc.) on some scale (1 to 5? 1 to 10?), and from this, each problem could be given an 'overall' rating in each aspect based on the average of these ratings (or possibly a weighted average based on how many problems a user has solved).
This would probably give more useful information about the problems than the number of users solving it or the acceptance rate. For example, the number of users is often influenced by things such as the length of the description - problems with long descriptions tend to have lower numbers of users accepted.
Also, many times a low acceptance rate reflects not that the problem is difficult conceptually, but that there is a tricky corner case that people miss.
[/li]
[li]Lastly, I think acceptance rates for problems would be slightly more accurate if only a user's submissions up to his/her first AC were counted, since people frequently like to resubmit their solutions to try to optimize their programs, which 'falsely' increases the AC rate.
[/li]
[li]EDIT: Would it be possible to have the option to delete a contest (hopefully with a 'confirm' page first
)? I have several 'test' contests that would just clutter up my contest list in the future.[/li][/ol]