I guess I didn’t make it clear, but I figured that was just a random number thrown out by you. Lol!
There are basically two schools of thought on the MMR/SR duality. One group thinks that MMR/SR should be more the same number, down to not using SR at all. There are a few different ideas in this range, such as simply explaining it better, making MMR:SR a 1:1 ratio using a known formula, and just not using the SR facade at all.
As my primary concern is that SR is OVER precise in that it makes no sense to say that someone lost 1 SR, I don’t like this idea. Also (and this is assuming it works like TrueSkill, which it’s often described in the same way), MMR is 2 numbers, a mean and a range, so unless they give the uncertainty value they’re not actually giving us MMR. Though I do think there is a case that could be made to give the uncertainty value, in some format anyways.
I, and it looks like you as well, are in the opposite school of thought in that I think SR should be basically removed and the Tiers only should be given, maybe with additional breakdown of the tiers. My primary motivation is that a skill tier is about as accurate of a rating as you can actually get and it would stop this “Blizz pls give me back my 20 SR LITERALLY UNPLAYABLE” stupidity we see too often. You’re right that either LoL or DOTA does this (both?), it makes so much more sense to me, but they did go from a more appropriately precise version in Season 1 to this one, so I’m not expecting an overcorrection anytime soon.
I can’t stress enough, that we “detractors” of this thread are perfectly comfortable with the idea that the current MMR/SR system has problems, but the problem isn’t that your MMR can be correct while your SR can be undervalued. I assume that they didn’t design the system that a 7 year old would see was absolutely stupid.
Good. I thought that was rather strange. I hope you understand how I got that out of what your wrote. My apologies.
Ok. You’re saying that there is no reason to base gains or losses on chances of wins if games are balanced on SR and the wins are 50% chance anyway, right?
You’re not wrong, but we’re getting into the practical nature of the system. So in theory a game of 1000 SR against 4000 SR could be made and appropriately measured, in practice this isn’t done for two reasons. First, that game would suck for both parties. Second, you wouldn’t be learning anything about the players.
The little thing you’re missing is that balancing on SR IS taking into account who the win or loss is against. It’s part of the same feedback loop. So your quoted sentence basically says “if the games are SR balanced then there is no reason to take into account SR”, which…you know…you’ve already done when you balanced it.
I’m not sure if you read my linked post above so: https://us.forums.blizzard.com/en/overwatch/t/competitive-matchmakings-mmr-system-is-very-good/43184/28?u=ozoneooo-1681
To not take into account who you won or loss against, it would have to be completely random team assignments. That would technically work, but you’d have all the problems I explained earlier…which is basically all the problems people complain about now, but either REAL or WORSE.
I think most people would agree that if you beat someone better than you, then you should get a bit more points for it. I think some of the discontent isn’t so much THAT they do it, nor HOW they do it, but how it’s all EXPLAINED.
Which is fair as long as the basics of how the system works is actually understood. All too often it’s not.