My suggestions in case ladder rating is used for any future big tournaments:
(some kind of system to increase civ variability on unusual maps, when picking one or two specific civs is optimal but boring. Civ ban system, etc.)
Hide ladder rating a bit: group people into categories like 1500+, 1600+, but have these be inexact. A bit similar to how in some other games, people are confined to a league or something, where they play with the same group of people for a while even if they are the best in their group; people would stay in the same category of 1600+ etc. for a bit even if they technically fit into a different one, only randomly jumping into the new one.
Also add some randomness to rating changes, so knowing the history of someone and all their opponents doesn’t allow exact calculation of rating.
Benefit of doing this: not knowing exact ratings prevents people from gaming the system. No such thing as “I am 30 points higher, so I am safe”.
Maybe good for a competitive scene with a small number of players interacting with each other: moderate total point change between two players if they aren’t interacting with other players. Instead of losing to the same player five times in a row and losing 20 rating each time, it could be -20, then -16, -10, -4, -4. If both players play several games against other players, then maybe the next loss could be -20 again. Would have to ask top players how they would feel about this, though.
Matchmaking suggestion also aimed at top players: have a queue option where matchmaking is performed at fixed intervals, like every 10 minutes. When the interval is up, everyone in the queue is assigned a match, even if it’s a 2500-rated player against an 1800-rated player. *(If odd number of players, the lowest-rated player is not given an opponent unless this is their second queue attempt.)
The idea is that a top streamer can end a game, and based on the time (7 minutes after the hour, etc.) can know how long they have to finish up the game, take a quick break, etc. before they can play again. Top-rated players normally may have to endure a long queue before finding a match: this becomes a disadvantage of being a top player. If many people like and use the fixed-interval matchmaking system, then it should be better for top players. But there’s also the danger no one will use it, which means no one CAN use it due to lack of suitable opponents.
System for increasing participation and fun during a Redbull-type tournament selection through ladder rating: not everyone can be rewarded. But people could have more chance to contribute to something. In this past ladder, people only earned rating (and gained experience) for themselves, or gave rating to their opponents. Introduce two or three big teams: through some system, people are induced to join teams in a way that prevents a big imbalance. By winning on the ladder, people improve their team’s rating (but weighted in a way that discourages making accounts that deliberately perform badly). The team that performs the best is rewarded, like with an extra tournament slot.
Finally, the problem of multiplayer being hard to get into, because of new players having to lose many games before their rating stabilizes. Aspect one: average skill level changes over time. Two players who start out new and play 500 games against each other will increase in skill a lot, but their average rating will stay the same. Some statistical analysis of a big dataset could figure out how rating should change to accurately reflect skill in this restricted case of two players, but also how the overall playerbase’s skill tends to change over time. In practice, this means making the gain in rating for the winner of a match sometimes be more than the loss in rating for the loser.
Aspect two: allowing players to judge and state their own skill to influence rating changes. Not everyone cares about rating points, or about who they win or lose against and their skill level; but for people who do care about points, they don’t want to lose a lot of rating to a “smurf”, and might not want to lose a lot of points to a highly-skilled player who has a low rating simply because the ladder is new with everyone starting at 1k.
So we allow players to suggest their own optimal rating. We have,
Real Rating (RR)
Claimed Rating (CR)
Tentative Matchmaking Rating (TMR)
TMR starts out the same as RR. If a player does not set a CR, TMR stays the same as RR or maybe fluctuates somewhat more quickly than RR. The queue always tries to find an opponent with a similar TMR.
So, if a new account always starts with RR of 1k, then TMR will also be 1k. If the previous suggestion is followed, then a RR of 1k now will reflect the same skill level as an RR of 1k three years ago; but regardless of whether that’s the case, the new account will initially be matched against other players with TMR of 1k, even if they set a Claimed Rating of 300 or 2000.
If a player loses and their Real Rating is above their Claimed Rating, they lose more rating points (RR, or maybe just TMR, decreases more) than normal, but their opponent’s RR increases by the normal amount. If they win, RR changes for both players by the normal amount.
If a player with high Claimed Rating wins, they gain standard RR, but their opponent loses less RR than normal. (Subject to the suggestion about limiting total rating changes based on interactions between two players, so two players don’t set high CR and trade wins.) But TMR for the winning player increases more than normal. If the player with high CR loses, RR changes normally for both players.
Those are the basic components. Some players want to quickly reach a lower rating because they want to win games, even if it’s because they’re against bad opponents. Some players want to gain points, which means winning games, but don’t want to make the players they beat feel bad. TMR is about trying to more quickly estimate the skill level of a player in order to make matches with a 50% chance winning for all players, as well as get closer to a 50% win rate for recent games if there has been a winning or losing streak.