This is a proposal to add seasons to matchmaking. Explanation Seasons in matchmaking mean that the ranking is specific to the season and not permanent. This means, when a new season starts, the ranking starts anew. A season lasts a couple of weeks to a couple of months. Rationale Once the ranking is established, matches are often against the same players. This leads to less pleasurable matches than with varying opponents. Furthermore, personal achievements like reaching a certain spot in the ladder can become possible for lower ranking players depending on the activity in each season. Additionally, the frustration of experienced players being beaten by newcomers in competitive matches is lessened, as it's not a permanent mark. This allows for more relaxation in a season where one isn't feeling in ones best form or just wants to try new strategies. Seasons are a marketing opportunity to get more attention to Zero-K and might increase player participation in competitive matches. Contra People might be proud of their historic ranking. In any case, I think historic ranking can also exist and the winners of each season also provide records. I would like to hear your opinion on this proposal.
+10 / -0
|
My guess would be that seasons would increase participation in MM. It would allow more active players more of a chance to get, for example, a top 10 spot in a season, where it would be difficult in the "forever ladder" where there are many players who don't play much but maintain a spot by playing a few games a month. It may seem less daunting to new players to the scene who are interested in the competitive/ranked side of things. New players can start strong, win 70% of their first 100 games, but not be able to see themselves on the ladder page. Personally I used to love the start of a new season on the PGT ladder in Starcraft ~15 years ago, where I had the opportunity to play vs some big famous names when everyone was at the similar ranking for the first day or two. What are peoples thought on length for such seasons? I think 3 months would be a nice starting point. It would also be a nice addition to player profile pages: Summer 2021 : 55th Autumn 2021 : 18th Winter 2022 : 7th
+7 / -0
|
This sounds like a great idea!
+1 / -0
|
I always liked to have more ladders, although I am afraid in this particular case the differences will not be as much as expected for people outside top 10 or 20. It is indeed an issue for the very competitive players, but for the rest I can see things converging to a similar position. What I would find interesting is to have any system in which you ensure you play with many different players. I wonder sometimes if the transitivity holds (meaning player A beats B, player B beats C, does that mean that A beats C?). Making seasonal ladders will not improve much the chance that A and C play together (maybe B really plays a lot).
+0 / -0
|
|
|
it resets high level player's ranks and makes them fight noobs? that just sounds like smurfing. the big problem with matchmaking is that very few play it, so you end up fighting the same people anyway and ranking doesn't really matter, since there's just no one to pick from
+3 / -0
|
|
|
There have been many discussions about this and it has always been declined for good reasons. One of them is that development time is needed for other things. Much work has gone into making the rating system more accurate. Resets would obliterate this accuracy. For players who play a lot, it would quickly converge to the current ratings anyway. For other players, it would just introduce arbitrary deviations and exclude them from the competition. quote:
Once the ranking is established, matches are often against the same players. This leads to less pleasurable matches than with varying opponents. |
Just activate wide matchmaking. Other players have complained about being paired with opponents with too high skill difference. They can deactivate wide matchmaking. League rating based matchmaking excludes this option. quote:
Additionally, the frustrating of experienced players being beaten by newcomers in competitive matches is lessened, as it's not a permanent mark. |
WHR adapts so quickly that it's not a permanent mark anyway. Only in the rating history, it's a permanent mark, just as it would be with league rating history. ~300 ZK players have participated in the following poll: "In order to be accurate, the rating system averages the results of many games. Would you like it to put more emphasis on recent games, sacrificing accuracy but allowing for faster rank changes?"
-
18% voted "I want more volatile ratings that represent my recent performance"
-
68% voted "I want more accurate ratings that represent my overall performance"
-
15% voted "I want this poll to go away"
+1 / -0
|
I saw that poll as well, and because of its wording I would have answered as well that I rather want a more accurate than a volatile rating. This is not in contradiction to seasons. Seasons have several effects, ranging of psychological like the feeling of something new starting, to publicity to the ability to test changes to balancing in a better framework as seasons could be considered units in that regard. Seasons don't necessarily mean volatile ranking based on the last couple of matches, except maybe the first couple of matches in a season.
+3 / -0
|
quote:
Resets would obliterate this accuracy |
This ladder is seasonal and has nothing to do with the normal ladder, so it wouldn't be obliterating any normal ladder accuracy. And because of that, you can play relaxed, not worrying about your ranking, and its fun (at least for me) to go up against a really high level player and get wrecked, you might learn a lot. Also if you don't like seasonal ladders than you don't have to participate in it, its just for fun.
+1 / -0
|
Seasonal rating is fine with me if it is done additionally to permanent rating, a bit like a tourney over multiple days in which everybody participates automatically, as long as balancing is not based on reset ratings and the permanent ladder remains fully visible. In this case, one could also use actual rating as ladder rating and remove visual rating decay. Even then, the lack of dev time remains.
+1 / -0
|
|
|
quote:
There have been many discussions about this and it has always been declined for good reasons. One of them is that development time is needed for other things. |
   Brackman doesn't speak for me. Many people who use the MM seem to want some sort of league system. The fundamental blocker is that nobody has implemented it. Development time isn't being spent on things that are more important, there is just barely anyone working on infrastructure development. If anyone wants a league then they're going to have to find someone to make it. The league will inevitably be implemented however the implementer wishes, or not at all. The only useful thing that can come from this discussion is motivating someone to actually take on the task. The primary goal of the existing rating system appears to be matching accuracy. I disagree with this goal as I think people enjoying playing the game should be the primary goal. The developer of the ladder (DeinFreund) seems to be inactive and uninterested in supporting it further or modifying it in any way that would compromise matching accuracy. It works well for converging rapidly in teamgames, but people on the ladder want more transparency than it provides. So perhaps anyone taking on the task would be better off building a system from the ground up. A previous thread with my proposal is here: http://zero-k.info/Forum/Thread/32336?postID=230904#230904 . It is a compromise between permanent and league rating. I'm not sure how many people want to dive into a league and get matched with anyone as the skill range can be quite broad. I have not delved deep into infrastructure. I think I could hack together a full rating reset, since I remember part of the code having a start date for replay processing. Changing this date every three months seems viable. But there would be nothing around it. No record of previous positions, no past ratings.
+2 / -0
|
quote:
[...] it would be difficult in the "forever ladder" where there are many players who don't play much but maintain a spot by playing a few games a month. |
I would be curious to experiment with having rating decay to address this angle and inject a bit more volatility in the ladder. It's a much simpler change to implement than full seasons, and a less drastic approach than resetting the ladder every few months (the more active the user, the less rating inaccuracy will be introduced by decay). If successful, this may encourage further work on proper seasons, and can be rolled back otherwise. (Disclaimer: I have sufficient mana to work on small scope infrastructure changes like the above, do not have the interest in a large scale redesign of the ladder, and I'm ZK adjacent and not personally interested in matchmaking)
+1 / -0
|
quote:
I would be curious to experiment with having rating decay to address this angle and inject a bit more volatility in the ladder. |
This is an interesting idea, and I'm not sure how intense the decay would be that your talking about, but there is a considerable amount of people who can't play that often, and this would turn away some of those people as they could never get a substantial ranking.
+2 / -0
|
There is already decay in the shown ladder rating compared to the actual rating in the current system, even though it is against the community poll. It depends on the rating standard deviation which is modelled as a Wiener process where it is proportional to the square root of time. For details, see this paper from Université Charles de Gaulle or DeinFreund's presentation.
+1 / -0
|
I suggested decay for the ladder point system, not any changes to WHR. Ladder points are immutable if you do not play (you will fall off the ladder after 30 days but your points remain unchanged even then).
+0 / -0
|
quote:
I think I could hack together a full rating reset, since I remember part of the code having a start date for replay processing. Changing this date every three months seems viable. But there would be nothing around it. No record of previous positions, no past ratings.
|
Plz no. Also, I'm annoyed that the last changes to the ladder system have led to the situation where @Kaki has higher real ELO/skill than me but is ranked 14 places lower. :/
+2 / -0
|
I agree with @Kaki and as well with    PRO_rANDY first post. A seasonal/quarter ladder/rating could exists next to an all-time ladder. Eg. every quarter a new ladder is created and afterwards logged. This would be like an automatically tournament, without beeing an event, someone needs to dedicate free time for. While I am not able to attend at tournaments for timing reasons, I was really happy, that the last 1v1 Tournament December 18th 2021 lead to an sudden increase of MM activity. Same for the 1v1 Tournament October 23rd 2021. I climbed the ladder mostly, because other players got unranked, not for real progress reasons. This makes the all-time ladder sort of useless. You play like 2-3 times MM as total lob and you are already top 100. As a sensitive person, this makes me cry! While I know I am a lob, it is sort of hard to figure out, where I would be in an all-time ladder without people getting unranked plus I don't know really where I am between the active people. Addtionally the current ladder could be maintained, while there could be a ladder without unranked player. Unfortunately available time is limited for everyone and I don't know how much work would be needed to set something up.
+0 / -0
|
Forgive my ignorance, what was the change  dyth68? Couldn't find anything with a quick search.
+0 / -0
|
