Honestly speaking, I (ZZZ) don't like the whole idea (EWSCPO) at all. It's fuzzy, it's subjective and it's even slightly irritating. It can encourage massive registering of dirty accounts (btw, I hate to point it out, what the hell had happened with the guys from VN? Looks as if a half of Vietnam's population got involved in problems solving).

In short, the number of solved problems (they all are hard and nice) is quite enough for the purpose of ranking. Nothing personal, just my humble opinion.

Btw, personally I count all of you (SPOJ creators, its problemsetters, its top 10 users) for REAL GENIUSES.

OK, Adrian! Everything is OK & nice.

I (ZZZ) used to easily agree with any other reasonable opinion.

I got it, Adrian! Thank you!

Interesting, why other spoj users keep silence on this matter?

Maybe they really don't mind theirs more "granular" and "precise" rankening?

Well, one reason is that there's not so many spoj users that look around here, maybe another is that people would rather have a higher point value on harder problems.

Regarding the potential ranking scheme, I think that the idea "the fewer solutions, the more points" should not be implemented (if more points is a good thing, of course).

While there is a certain spammishness about submitting solutions en masse and milking the last miliseconds from them, I think that one of the great things about SPOJ is that allows one to experiment without any kind of reprisals rank-wise. Many test cases are not trivial to mimic, too (I can't think of a method right now to create a large, valid CMEXPR test case file, for example). What if a person had some silly compilation/runtime error? Such mistakes happen, especially if it's 2 AM and one was hacking 6 hours consecutively to get PRIME1 to work with Python . And don't forget that a correct algorithm in a relatively slow language may not run fast enough without optimizations (again, like PRIME1), and it's difficult to say exactly beforehand if a given solution will terminate in time or not. Finally, this will encourage multiple accounts from the same person (i.e., solve one problem with drafts, then submit the best one using the other account).

I'm speaking from an entirely egotistical perpective, of course, as it'll hurt my ranking relative to the C-people because I code only in Python ).

To clarify (I think I understand what you're talking about and everyone else doesn't here) - the "score" you get for solving a problem is a constant that's inversely proportional to the number of people (or accounts I guess) that have solved it already. It has nothing to do with the number of times those people solved it, the number of languages they've solved it in, the number of failed submissions you or anyone else made on the problem, or anything else but the number of people who have solved it. Correct?

(Perfectly correct. -- Adrian)

It would be good, if for each problem in a separate column the quantity of points which for it now can be received was displayed.
I think, that is not so convenient, when do not know, how many will receive the decision of a problem.

Hmm, maybe it should be done in another way? I think that number of points that one gets should depend of position of one's solution in this given problem ranks, do you agree?

In my opinion, the problem with that weighted system is in what it's based upon: Quantity of answers instead of their quality.

For example, one may solve the second most solved question in the system, FCTRL, with a very crude code that takes a lot of time and resources, but it still manages to get accepted.

Another one may do it in O(1) after an extensive mathematical analysis.

Who'd you reward the most points? And how much?

Even if speed and use of resources isn't the best metric for analyzing a well done code, perhaps the rating system can be based in how much better did you fare versus the average score of everyone in that problem.

Another second metric could be an user-rated problem system, where selected users, or top ranked and active users could vote if the problem is extremely difficult or trivial.

I think that renrutal's proposed system (and indeed many similar suggestions) defeat the purpose of having so many languages available. Think for a minute, if we ranked only according to speed, nobody would use Python or Ruby because the programs would take ages (relative to C) to run. Yet, Python, Ruby, Perl, etc. all make excellent programming languages due to ease of use, and the fact that a good program in those languages (an efficient one) will be far better than a bad algorithm in a fast language. If we went to speed competition as opposed to "difficulty based on number of successful solvers", we would see that many of the Python/Ruby/Java/etc. programmers are turned away, because they cannot compete realistically with C/Pascal/C++ programmers.

Then you could base the score per language too.

That way, the same problem can be rated trivial for Python programmers and very difficult for Prolog programmers, and if resolved, the points could be given accordingly, more to Prolog than to Python.

Two C++ users would compete for speed and memory use with other C++ users only.

Why 80,40? Why not, say, 60,30 or 100,50? What do these numbers mean?