map-based rating

Re: Fairness % of rating

Collecting the previous posts and arguments while absolutely thinking that the basic idea in general would absolutely be beneficial.
The counter-arguments to a map-based rating are absolutely correct, but maybe it's possible to find a solution that helps without destroying anything.
Practical approaches:

  1. It only matters for the map i currently want to play - if someone plays this the first (or say < 5x) time could be marked somehow - as it could be marked if a player played it 100x or 200 or even 500 (imagine a color indicator) (i absolutely know that this is not the "best" approach but i would definitely pay attention to such a badge)
  2. if one could "activate map-based rating"(like a game mode)
    2.1) maybe it is somehow possible that locks a lobby from map-change later (there is already a pre-pick! Or IF changed later, that mode is auto-turned off...) ?
    2.2) for that particular game mode the set of maps are just hand-picked (look how often the same maps are played) so that would totally beat the duplicate/half-duplicate/whatsoever map-problem.

Of course, it's not a solution already, but imho going on with thinking about it is really worth it and if we want a better rating system we cannot ignore the important map-to-rating relation and we would stop "rewarding" people to train a "i only play mid on XY-map"-behaviour (yes, spot-based rating would be even cooler!!). I would very appreciate if this overall discussion about rating would be continued with a good structure. This topic is (and should be) a long-term task imho.

That sounds cool at face value but then I remember that I already have three other ratings to keep track of, nevertheless a dozen new ones for all of the popular maps!

This was already talked to death. Basically, it is a good idea, but too much work to implement, especially when there are other bugs/features needed.

Previous discussions never led to a consensus like "would be nice to have, but other things have priority".

It would solve a problem that only exists because of another problem: that people mostly play a single map. And we should solve that instead of making it even worse.

The solution for this would probably be really complicated right? I mean there are so many maps out there in various version to calculate rating you would need a huge amount of games. Showing the times a player played a map is probably doable. I saw some site where you could display how many time a player played what map.

I'd settle for the number of times a person has played that specific map - not ideal, but presumably easier to implement? I recall someone saying there's no way of telling if the hosted map is switched before the start of the game, in which case I'd also be ok with the simplification of just doing it based on the map that the host had selected initially when creating the lobby.

I think a little counter that shows number of games on a specific map is an awesome bit of information!! Maybe possibly with win percentage but thats prolly hard to implement because of broken replays and faulty data in general

Just having a counter that tells you how often someone played a specific map (or one of it's close variants) does sound like a great idea indeed!

To minimize UI changes, it would totally be sufficient to show the "number of games played on this map and it's variants" as a tooltip when hovering over the "total number of games played" number already present in the lobby.

We talk about this one to death every two weeks. The major issue is that it's very easy to circumvent, you might think that players only play one version of Dual gap for example but there is probably over 200 different versions now. Evading a lockout, a percentage based reduction to trueskill or a individual map based rating (hiding it in order to smurf) is as easy as two clicks.

You would need to do the following for this system to work:

  • Find someone who can implement a tagging system
  • Get a group of people who can constantly monitor tags for faults/apathy
  • Apply penalties/stats to games played in maps with a certain "tag" and not on a per map basis.

The first two are not really feasable. We can't rely on having a janitor service to make sure a map doesnt not have the tag (and thus can be used to dodge penalties etc) and Dev manpower is always critically lacking too.


All fair points.

That being said, you can circumvent most of then with the "number of times someone has played the map"-approach:

  • You only need to worry about the dozen or so most commonly played maps and their derivatives, to cover >90% of games played. You really don't need to tag every map in existence, the main ones suffice.
  • If a less popular map is not in the system, or not correctly listed as the derivative of another map, the shown number will be too small/wrong, of course.
    But who cares? The number would only be used supplementary to the global rating when balancing manually, so inaccuracies on much less played maps won't matter all that much
  • Similarly, if someone wants to game the system by hosting a clone of Gap that's not tagged correctly, he doesn't gain anything. The fallback is just the current global ranking, which works well enough.

As Bennis mentioned, this would still be a nice bit of information though when manually balancing the most popular maps.

As a Seton's player I'm of course mostly thinking of Seton's and, while the map is great, it has the big problem of the highly balance sensitive air spot at lower ranks. As you all know, if you put someone with a practiced BO vs. someone who didn't practice a BO on Seton's Air, the game is essentially over before it began, regardless of their rating.

Imho, you really can't/shouldn't play opti on Seton's due to this at lower or even medium ranks (<1.2k or so). That means you must balance manually, for which such a "games played on Seton's" number would be AMAZING.

@CheeseBerry said in map-based rating:

someone wants to game the system by hosting a clone of Gap that's not tagged correctly, he doesn't gain anything.

If a player has a lower rating in global than he does "gap" maps, he gets to play the game with a lower rating than he actually should have if he chooses a different version. This is understandably a big problem.

@CheeseBerry said in map-based rating:

You only need to worry about the dozen or so most commonly played maps and their derivatives, to cover >90% of games played

This might be fine for setons but the map usage will shift away from the "most played" versions as more and more people figure out how to game the system.

Alternatively, you can upload a similar map to the main version with a different name and some minor changes and continue to play the map without the system or statistics.

I do think showing a game count (showing what percentage of your total games played are on this map) is a good universal addition seperate from any map based repeat-penalty systems. But again, people on Gap and astro will probably circumvent it.


Yeah the system can be circumvented, but what's the incentive for doing so?

The "number of games played on this map or variants"-number would have only one use case: Giving the host an easier time to manually balance games. I guess the host can use a map variant not tagged correctly but that would only result in them now not being able to see how often somebody has played the map.

Why would that be beneficial at all?

@biass if you just go based on the precise map name there's no need for tagging, monitoring, special penalties etc. All you do is show the number of times someone has placed that specific map. It's not perfect given you'll have lots of near identical maps with different names, but it's still a lot better than nothing.

Or make it easy to see the 5 (?) most played maps of any player in lobby (might require a bit of UI work). If most played map is "Gap v27" people will understand what that means if they are in "Gap v28" lobby.