Algorithmic Handicapping (MMR) is Wrong for Overwatch

Technically, everyone is being handicapped in every match. Whether you are handicapped to your advantage or disadvantage depends on how skilled you are relative to your peers. Take a match with 12 solo players for example. The two players of the highest skill would automatically be placed on opposing teams, with other players arranged to offset the difference between them.

2 Likes

provide the proof for this statement.

no one is being handicapped by this imaginary cabal within blizzard. you should adjust your antennas on your tin-foil hat.

1 Like

I don’t have proof of this, but it is a reasonable inference based on the proven information we have, which is developer statements and patent filings of the game holding company.

2 Likes

so your claim of this being “automatic” is false. please resubmit for a higher grade.

1 Like

It is automatic. The Matchmaker automates the creation of matches and the assignment of teams. What are you saying is false?

2 Likes

please refer to my initial quote of what you said, it will answer your question.

1 Like

You’ll have to remind me. I’m not scrolling through 1,806 comments to try and find the one you’re thinking of.

1 Like

twenty characters…

1 Like

I’m speaking in broad terms, this isn’t exactly what would happen to effect handicapping all of the time. But handicapping does happen all of the time, and it is automatic. I don’t know what else you’re looking for.

1 Like

heres your entire quote, maybe this will help you remember

1 Like

this is oversimplified.

would the matchmaker actually work, all members of a team would be playing at the same skill level, which is just not true. especially in gold, where smurfs and noobs emerge continuously.

the biggest problem is that a lot of players experience major changes to team allocation after 3-5 matches won in a row. its not that the enemy just gets better, the own team mates get significantly worse.

which makes it twice as hard to break out of your personal elo hell.

1 Like

For me the issues are:

-Transparency-
How exactly am I being rated?
What is my rating and what are my teammates’ ratings?
What are the enemies’ ratings?
What are the average performance statistics for my SR, hero, and map?

-Nested Ratings [SR[MMR]]-
Give me one metric that is solely based on Wins-Draws-Losses for all ranks

-Fairness-
Higher MMR within an SR bracket is purportedly intended to make matches fair by identifying carry potential of the player. However, it doesn’t seem to be matched on a per-role basis. E.g. a high MMR tank is going to have more impact against a low MMR tank even if there is a high MMR support to balance out the low MMR tank.

-Ethics-
There is a growing body of evidence that many algorithms in the tech/app space are designed to increase user screen time without the user’s knowledge. Does MMR increase player screen time by artificially manipulating the SR grind?

4 Likes

This really just not how it works. The algorithm is the same for all.

If it treated players differently then players would not be able to reliably return to their rank after making a new account and playing a reasonable amount of games.

New accounts in the hands of already experienced players regularly test the system, time and time again the high rank players return to the high ranks and the lower ranked players do not.

Not everyone is going to climb, it’s not a conspiracy it’s just the math of how many people are playing. We can’t all be in the top or it wouldn’t be a skill rating.

2 Likes

well no, and yes I guess. I’m talking lifetime. Population wise, I believe that the average player should be in the lowest ranks. and as they get better it should thin out. Because that’s how real life competition works. Only the top continue on to the top. I get why it’s done… for the masses to enjoy the game, but then… QP for the bell curve, Comp for the true competition.

But I don’t develop such things. So maybe I can’t tell you, but if they started little league players in JV teams, and had them fall to their ranks as little leaguers… everybody would feel bad. And that’s what it seems like they do here. IMO. once again, I don’t think it’s evil., I think it’s misplaced, and fairness is… lost on trying to be fair.

1 Like

Pardon my ignorance, but that seems ideal when creating fair matches. Does it not?

It could also be the two best with the four worst, going against the six middle people, etc…. But the point remains. Allocating out the expected performances of players gives a result where you have the highest available confidence the match won’t just be absolute garbage. It regularly fails in practice, but that has more to do with people playing than the intended goal of a matchmaking system.

1 Like

The leagues you speak of generally start as randomized pools of people based upon age groups. Then the best kids from each league go to all-star teams that compete locally, and then spread out geographically to the surrounding areas to put the best against the best until a champion is determined.

So, it’s hard to make a fair comparison to the population and player base in OW. I see where you’re coming from. I don’t know that I’d start everyone at the bottom just for the purpose of risking people never making it past that initial crapfest. It would be better now that the ladder is established, even if only marginally, but the MMR reset that happened once upon time was pure chaos (both good and bad :rofl:).

but it doesn’t seem to do that. It takes an average. So if someone has good stats, it will pair them with someone who has poor stats to make the teams even, regardless of role. So that 50 kill game you just had on DPS will offset with someone in your rank who had say for ex. 2. But two opposing people had 25 per game. Ideally, that wouldn’t happen… but I tend to believe it happens more than it doesn’t.

1 Like

We don’t have the data to assess that. It’s purely conjecture.

MMR is used to make matches - and while stats obviously go into coming up with that number, there’s no strong evidence I have ever seen that correlates to the idea that “player a is doing well, so we will pair them with potato c”.

1 Like

you are correct, we don’t have that actual data. But we do have the data that SR is based around that average. At beginning of every match, unless you have a placement on your team… or the other team, the average SR will show. So yes, I am extrapolating data from that… and I may be wrong, but it is a good start.

1 Like

So QP and comp have different MMR’s specifically because some players tend to play QP less seriously and there is a general assumption that you are there to specifically not play competitively.

Competitive mode is honestly not really the true competition mode in OW. Going Open Division -> Contenders -> OWL is the “true competition” and is closer to traditional sports.

Competitive mode has to be designed for many players of different types to be able to compete over a ladder position. It’s not perfect but it’s certainly a competition. Leaderboards are a pretty standard part of video game competitions and inline with many people’s expectations of a competitive mode.

Wouldn’t make sense for it to serve that function. The aspects of Overwatch that are designed to keep you in are the lootboxes, timed-themed events, limited time rewards.
You can always get a better reward in Comp by playing more so the reward structure rather than how it rates you serves to this end.

So sure there are some pretty big patterns to make you play more in OW but manipulating SR would be counterproductive as this would leave poor match quality in the high ranks and the high ranks serve as a massive marketing boost to bring new players and encourage veterans to load up another game.

It would literally hurt profits for OW to keep good players out of the high ranks.

2 Likes