Ranking v2 suggestion


#21

Against. It would be like taking StarCraft’s 1v1, teams and arcade and try to make a global rank/league out of it.

As I mentioned in my previous post, it should be separate per topic, and the profile badge should display either the best rank or let the user pick which one to use… or remove it entirely.


#22

Split them.
You already have a leaderboard per category, it’s just too hidden:
image

And you can add them as separate in main page or something:


I’m grate artist.


#23

For a general leaderboard. Obviously the exact share of each category is subjective, but it’s still interesting to have a canonical global ranking. Those who care can optimize for it, others can focus on any single category ranking they prefer. I agree with others that the single category rankings should be made easier to find and see.


#24

For a general leaderboard too.


#25

There’s no way to find a percentage for each category that satisfies everyone, so going this way would probably mean endless discussions and complaints. So I’m voting against it.


#26

We already have a general/split leaderboard with the filters. Maybe just tweak the UI a bit so that it’s less hidden ?
Also the ability to select the profile badge on his profile would be nice, as reCurse said.

In summary : You could use the new formula (BASE * min(N/500,1)) to make scoring more accurate and make a leaderboard for each (currently hard-to-find) filter.


#27

I second that reasoning :grin:

As for two choices from the original post. Neither are perfectly to my liking. The first has too high clash score I feel. The second is a bit high on contest compared to multi (yeah, that would be in my favor, but 50% seems a bit much).

The reason people prefer contest to have a high percentage because it is probably the best one to estimate a players’ botting skill (though not perfectly) and I think botting is the biggest thing we do here. But if you only care about contests, you should probably just not even look at the overall board. If you make the contest percentage too high on the overall board it will just be another contest leaderboard, but one with a large error margin (the error margin being the score from the other categories). I would prefer contest and multi to be almost the same. For example:

  • contests (40%) (base 40 000)
  • multi (35%) (base 35 000)
  • opti (10%) (base 10 000)
  • code size (10%) (base 10 000)
  • clash of code (5%) (base 5 000)

Others have made the point that everyone has something they like to do, but the time investment and skill requirements for each category are not the same. Contests and multi take a lot more out of you than the other three. That should be reflected in an overall rank. I am a bit biased because i dont play clashes of course…

I get the feeling some want the overall leaderboard to disappear entirely. That would be a shame in my opinion. I do also think that the categories should be made more visible to also encourage people focusing on the one they like. Maybe put the visibility on par with the overall ranking board.


#28

I’m with @MSmits on that one, some people seems to just want the global leaderboard to be gone. But it seems to me to be just a case of grumpy “if I’m not having fun, nobody can !”.

XP, CP, achievements, leaderboards, rankings and prices are nothing more than gamification of the platform, extra goals other than selfmotivation, to give something to aim for, extra motivation, extra gratification “You did something, get a cookie”. If you guys are looking for an objective way of telling who is the best programmer on the platform, it will never happen.

The original problem was that the player number was way too important in the formula, meaning other people not showing up would hurt your ranking and you dont have any control over it. I think that’s the thing that should be fixed. We don’t need to turn it in some gigantic overdesign.


#30

Thanks everybody for your feedback. It’s very appreciated and valuable to us.

Here below is the final proposal that we will implement this year:

  • the base formula for all games becomes: (BASE * min(N/500,1))^((N- C+ 1)/ N)
  • we don’t want to use the alternative formula (the one that uses the exponential) because it’s more complex to understand and it will increase the impact of the noise during the rank computation
  • the BASE depends on the game type:
    • Contest: 10,000
    • Multiplayer Game: 5,000
    • Clash of Code: 5,000
    • Optimization: 2,500
    • Code Size: 1,000
  • we keep the 3 best contests (no fluctuation depending on the age of the contest as the new formula fixes the problem with high vs low player count)
  • the global leaderboard is maintained: it does not reflect the “best” players in the strict sense of the term but reveal the best and most active players of CodinGame, giving everybody a constant feel of progression
  • to counter-balance this, we’ll emphasize all sub leaderboards by adding tabs on the leaderboard screen. Something similar to (design not final):
  • as we’re reworking the home page, it will be possible to view all rankings and to focus on one specifically by displaying it by default
  • the gamer card will display all rankings of the user

With this proposal, we expect to fix the initial problem without over-complexifying things and to allow everyone to focus on what is important for them.

Here is the simulation with the new formula: https://docs.google.com/spreadsheets/d/1vS-PSi3MAFfjxmIe7HBiTXMvE6l0TS3Dq7c9c0nh6Cg/edit#gid=1844624130


#31

Sorry for tuning in late, maybe too late judging by the finality of the last post.
I do not put a lot of weight on rankings and view the platform more as a way to improve my problem solving and programming skills.
Nevertheless it seems odd to me, if I understand it correctly, that CoC should have more weight (base score) than optimization games. The first is solving easy tasks, the other is contemplating for days and even weeks and trying to improve your solution over time, thus being much more challenging.


#32

Probably has something to do with reverse engineering the validators (or using known validators?).

But i agree Optimization should be valued more than CoC. But this debate can keep going on and on for ever, like a certain political event that still has no end in sight.


#33

Well, I have not used reverse engineered/known validators to up my score so far, and surely this imperfection should not be a reason to depreciate optimization games.
About the debate - it already seems a bit too late.


#34

CoC will have the same weight as 2 optimisations games. It’s not that much.

Anyway, no one in the top 5 global learderboard do CoC. So it’s not an issue :smiley:


#35

Machete did play clash :slight_smile:


#36

Traitor and heretic !


#37

Well, solving even the very hard puzzles only gives experience points, but solving tasks ranked in difficulty below the easiest puzzles gives you more ranking points than optimization games… it is still beyond me.


#38

To have an screenshot being 1st :grinning: before it could be removed.
Also I did like CoC (unpopular opinion), it’s not that bad if you play as it’s meant to be played. And I like codegolf too, you can argue that AI bots are harder and blah blah blah, in the end codegolf were fun puzzles to solve.
I don’t think it’s time to reopen discussion about what’s most important to what, it goes to a dead end. CG staff have already taken a decision after months of being discussed. I prefer some bonus on recent challenges but well, it’s fair.

I think the redesign is good. If there are people that only consider challenges as the real deal there is a tab for that.

Edit: Make forum and blog buttons more visible in the main page!


#39

Just for the record, wasn’t asking to reopen this. Was simply pointing out the discussion can carry on forever. You’ll never get everyone happy.


#40

Will the points recomputation will remove the earned points from The Great Dispatch ?

The puzzle is not accessible anymore but players still have the points. It’s not a big deal at the moment because it is not that much points. But with the new formula it will be (nearly) 2500 points for the top of The Great Dispatch. It’s pretty unfair since we can’t improve our code anymore …


#41

yes, we’ll remove the points associated to The Great Dispatch