I have doubt about a lot of these changes. They seem targeted at players behaving badly because they overly care about rating. I believe the only thing that could stop people who care about their rating from caring about their rating is to abolish rating, at which point you could lose the players who care about rating from the site entirely. Given that I doubt this is a desired result or will ever be implemented, there will likely be a perpetual problem of players who care overmuch about their rating behaving badly.
There is also a larger issue that many individual games and communities, and even individual players, will receive systems applied generally in very different ways. For instance: the thresholds of what each 100 rating means is very different in different games. There are some games where ~500 is the pinnacle (can't stop), some where ~700 is the pinnacle (agricola), and some where 900+ is the pinnacle (azul), and what this means for the rankings underneath is quite different on a game to game basis. What an arena season's settings look like, or how much it changes seasonally, varies greatly from game to game. Even the length of a game has a massive impact on how it deals with general systems.
With the above in mind, i would like to address the individual changes:
Team play:
I have no experience with team play games on this website (which does matter), but this change sounds good. Although if I understand correctly, should this change not make the players on a team have different elo wins/losses against the opponents, they just do not win/lose against each other?
Cooperative play:
I also have no experience with cooperative play games on this website. However, my understanding of what is described fails to understand how hiding ranks to this limited a degree will stop players from being elitist and competitive about ranks. You might not be able to tell what a player's exact rank is anymore, but you will still know roughly what their rank is, with a range of variation on that. While it may no longer be possible to confidently claim you are superior to another player based on ranking, it will still be possible to unconfidently claim this, especially falsely, so this does not seem like a clear improvement.
Power leveling:
This is a clear case of the cure being worse than the disease. Firstly, I'd like to point out that multi-accounting is a problem above and beyond directly transferring rating from one account to another. I can think of three uses for multiple accounts to get a very high rating that don't rely on transferring rating from one account to another one and they have all happened on this site. (I will not explain what they are as I don't want to let people know how to cheat). This new repeated win restriction system applies everywhere, to everything. It is also deflationary - the sum total of ratings across a game can now be smaller after the game than before, as the win-streak penalty only blocks gains and doesn't prevent losses in a loss-streak. Even were this fixed, it would not be a good change.
It is not infrequent in small communities, or multiplayer games (or both), to queue the same opponent repeatedly, and any winstreak (even of 2) in this is flagged as being power levelling. For players at the top of the ratings, it is very likely that you will be winning against a lot of the general playerbase many times in a row.
On a quick count of my hit stats I have played 101 players 2 or more times in Agricola since the start of this arena season in 155 games (and another 109 players once). Not all of these games were arena or even ranked (as there is currently a new expansion in alpha requiring friendly games), but it's still a reasonable sample. I have a 100% hit rate vs 31 players with 2 games matched, 8 players with 3 games matched, 6 players with 4 games matched and 1 player with 5 games matched. (I also have 0% hit rate vs 3 players with 2 games matched in that time period, one of which is a massive upset as the player is currently ~370 rating lower than me). So that's 46/101 players that i played multiple times in the past two months where i definitely currently am on a multiple game winstreak against them, ignoring most of the players i actually play the most because i did not maintain 100% hit rates vs anyone past 5 games (and I'm also probably on a winstreak vs some of them, or they are versus me). There are currently 639 players listed across all ranked categories for the arena season - assuming all my opponents played arena i've played a third of the entire arena population, and about a sixth of it multiple times. I also have used the game filters specifically to avoid playing players in bronze (although it's not particularly effective in turn-based games) so it's in effect 382 potential opponents - meaning that I've probably played more than half of my potential opponents, more than a quarter of them multiple times, and likely have an incentive to not play against more than a ninth of my potential opponents. If I queued three games vs completely random opponents i am expected to queue vs someone I've beaten multiple times before. This is not getting into the issue that queueing games will not give you random opponents, due to timezones and other factors.
There's also an irony that the next two points are players not wanting to play with specific other players (and now there is a new rating-related reason to do this beyond abstract questions of your belief in your winrate vs specific players not matching the system's belief in your winrate vs those players - it can now reach a point where you can literally only lose rating by playing another player, so to preserve your rating you should try to not play them) and players wanting to not play to maintain their rank (when the new system is deflationary, and thus you are expected to lose rating if you play a game vs fairly ranked opponents, so you should not play to maintain your rank).
ELO Decay:
I don't entirely understand what this is for. There are two things that already do this: hiding inactive players, and arena seasons. Given that anyone who hasn't played rated games in a while isn't going to show up on that game's leaderboard and is thus already incentivized to keep playing (and if they're not it's probably because they've lost interest in the game), and that the decay is probably not frequent enough to meaningfully influence players (one game in three months is not much more than zero), this change only really affects players who start playing again after the decay happens. It should be noted that the rating system is not merely aspirational, it's also a fairly effective tool with which you can ensure a good quality, roughly evenly matched game. A game between two 700 rated players or two 300 rated players will probably be better than a game between a 700 rated player and a 300 rated player. Making large artificial changes to people's ratings degrades the value of rating as a way to judge how good a game will be. If it must be done, increase the uncertainty of a player's rating if they don't play for a long time.
The issue of decay not being frequent enough is awkward to solve - while you can increase the frequency enough that those players at the top of the rankings would have to play a noticeable amount, it will still be less than they probably played beforehand (because the most highly rated players probably tend to at some point be some of the most active players), and it will start decaying players who are playing a normal amount. Variability in games matters a lot here - if you decay once a week this affects turn based games that last a day a lot differently to games that last a month.
Red thumbs:
The red thumb system absolutely could be abused to keep players out of games for rating based reasons - I have done it myself to filter out weak player wildcards who I feel are guaranteed to kingmake randomly in multiplayer games (before I started just filtering out all weak players with the level filter). It is also used as it is the most effective form of ensuring good behaviour from players - you filter out the badly behaving players so they don't get in your games. I am not clear on what is going to happen, but this sounds like it could easily lead to a cure being worse than the disease.
I would also note that if you want to avoid a specific couple of players in arena you can do so by looking at whether they're trying to queue for a game or not (or in a game, or offline). This is far too much effort to do for many people, but if you're trying to avoid only the top couple of players it is very simple.
Overall, the changes seem to mostly not have considered the changing incentives of the new systems that are being implemented, and assume that players will choose to behave in a way that would not have required us to make changes in the first place while in some cases introducing incentives to behave badly. In addition, these changes add deflationary pressure on the rating making playing games less exciting for players who are interested in rating.
