Algorithmic Handicapping (MMR) is Wrong for Overwatch

Lots of people assume this, but it makes no sense. SR and MMR exist for different purposes. They probably do not even resemble each other numerically. For all we know, MMR is not even a single number but a set of numbers.

1 Like

It actually does make sense. The numbers can be extremely different in form and with the information they represent, but if they are reducible to the same end result, they can easily be treated as the same when referencing it in short hand.

Some real-world equivalents of this would be comparing vehicles to VIN numbers or products to serial numbers or barcodes. The identifiers used on the manufacturer’s side mean very little to the consumer, but it contains a variety of detailed information that is important in the grand scheme of things.

MMR could be a ridiculously complex sequence of numbers and characters that looks nothing like the 4-digit SR number we see on our ends that would make absolutely no sense to us if we weren’t given a way to decode it, but if that complex sequence is still reducible to roughly the same number more often than not, it doesn’t matter if the separate number exists in a different format.

1 Like

Okay you’re right about that; MMR doesn’t have to be represented exactly the same way that SR does, to correspond with it. But calculating SR is a long-term operation that happens across the entire career of each player, and it depends mainly on the win/lose result of the player’s matches. Calculating MMR is a shorter term operation that takes place across X number of matches, and does not depend on the win/lose result of the player’s matches.

When I played Competitive Overwatch, my rank ranged wildly from Bronze to Platinum. And I ranked lowest in my career towards the end, when I was finding myself surrounded in every game with less experienced players. The inconsistency suggests to me that SR is not an accurate representation of skill, and that MMR guarantees unpredictable results in every match no matter how skilled a player is, and high or low that player is ranked. That is the purpose of MMR, as Overwatch developers describe it, and exactly what I made this thread to argue against.

SR is not a permenant value for the skill of a player though, otherwise you would never see it drop. It is in fact a spectrum of the range that players are able to play at. Everyone is improving at this game every time they play it, and it makes sense that a gold player (to use a general rank) now is a whole lot more skilled than a gold player back in season 1. To climb, you must not only improve, but you must improve at a faster rate than the average person at your skill rank. Likewise, if you do not improve at the average rate of a person in your skill ranking, you will fall.

The matchmaker is not perfect, and yes, sometimes it creates horribly uneven matches, and even less often it will unknowingly put someone on a loss streak they can do nothing about. However, if you truely belong at that rank, then you will climb back into it with ease. This happened to me numerous times when i first hit gold, as at an SR close to the edge of a rank, your rank becomes very volatile due to the imperfect nature of the matchmaker. However, you cant really change that imperfection as many people think by removing MMR, the only way to improve it would be to change the way SR works, and make it role or character based in order to get a more accurate reading of each players skill depending on what they are going to play. And the way that hero switching currently (is supposed to, one tricks notwithstanding) work, that is very difficult to do.

Anyways, what im trying to say is, you are the only person who can truely affect your rank, and I would fancy a guess that when you dropped near your end of time playing comp, you were not playing as often, and thus the average player in your rank became better than you, causing a drop in your rank

1 Like

Permanent might have been the wrong word.

Is MMR a mere ‘imperfection’ of the competitive play system? I tend to think it is a complete subversion of the competitive system. There really is no saying which of us is correct, because Blizzard has never disclosed the details of how MMR and SR actually work.

That is my main concern about MMR. It undermines the fairness of competition just to keep people grinding.

I don’t think that most players are able to perceive the handicapping system. Keeping it hidden seems to keep us from thinking about it. That’s why I argue that Blizzard should be legally obligated to disclose it, not buried in the game’s terms of use but also prominently somewhere in the user-interface for all to see.

This is a great reply, sorry I missed it! You’re describing the same experience that I had with competitive play. It really is reassuring to know that other people go through it the same way, despite the Blizzard executives and fanboys who try to gaslight us and convince us that right is left and up is down…that handicapping is appropriate for “competitive play.” It really is a cruel joke.

Your position is understandable, on the assumption that a system like this does no harm to players. But I think it really does harm them, by wasting their time and warping their idea of fair competition.

Exactly! Great post.

In today’s state of corporate oligarchy, I think we are all feeling some degree of Stockholm syndrome. It is tempting to pretend that the interests of private, corporate enterprise are not diametrically opposed to the public good. And yet…

I’m not sure what you mean by this? What points of reference were given?

My post opens with a direct quotation from Overwatch’s principal designer. References don’t get much more definitive than that.

Nettled! I like that :slight_smile:

This is a good and accurate summary of what the system does! I still think that players should be allowed to see MMR, because not everyone is aware of it, especially new players. And there are many experienced players, and the developers themselves, who refuse to admit that it is a handicapping system.

That is the pattern, pretty much. And generation after generation of Overwatch players get to go through this cycle, over and over!

(References section at the bottom, you have to expand it by clicking on the arrow to the left of the word References)

Somehow, I get the feeling that you intentionally ignore Kaawumba’s topic that’s compiled every single statement Blizzard reps have ever given about matchmaking.

SR and MMR don’t exist for different purposes. However, you are correct that they don’t resemble each other numerically.

Firstly, SR is meant to be read as a human-readable approximation of your MMR, and that with the sole exception of decay, your SR and MMR are close to each other.

I know you’ve made the statement in other topics that two players can have the same SR but one player has a much higher MMR than the other player so that they could be paired together for “handicapping” purposes, but then those two Jeff Kaplan statements would have to be incorrect.

Second point, which makes a lot of sense when you check with the first point, SR isn’t used in matchmaking at all, only MMR (which is basically translated to your SR) is used.

So basically, if you consider a primetime game with 12 solo-queue players in, say, around SR 2600 (so no one 50 above or below 2600), as far as the matchmaker is concerned, all 12 players are as good as each other.

Sorry, I meant that as more of a rhetorical question. Kaawumba did give references, but they have zero value to his assertions. Most of them are irrelevant to the issue of handicapping, and none of them refute my argument about handicapping.

Well, I guess the “question” you posed was rhetorical to you because your argument was always a falsified one and Kaawumba’s topic directly refutes your argument.

1 Like

if overwatch wasn’t handicapped, there would be no SR. Just like in Mongolian wrestling. No handicap. High school wrestling and boxing has weight classes i.e. handicap. Mongolian wrestling has no weight class i.e. no handicap.

You can do overwatch without SR. all you have to do is match the first 12 players in queue over and over again and then just count win percentage. Who ever has the best win percentage would be the best. 80% of your games will be stomp or be stomped. Welcome to Overwatch without a handicapping mechanic.

1 Like

I think you make a lot of great points but it is a bit wordy and repetitive. That said, would you also add this point?: the MMR system amounts to giving an unfair advantage to Blizzard employee-players, qua being insiders who know how it works and therefore how to adjust their play to best exploit it.

Related:

2 Likes

You didnt understand his post.

it lacks brevity, can you give me the abridged version?

The point Op once posted is totally valid. Its totally obvious that the MM pairs players of different skillevels together and tries to get a 50% winrate. This works fine the old way in plat and lower with pbsr, but not in higher elos in diamond and above. You have to be really really good to actaull carry in diamond games. If you dont impact your games significantly you wont climb, even if you get better, but not good enough to carry (as a support, what a joke…)

I played with friends in a 6 stack many times, with many players from gm to silver. With smurfing, we play the games regurlaly at 2500-2800SR and boi, these games are far away from that elo. After a few matches the mmr of the smurfs is adjusted and the games get really tough. You actaully hit other stacks with the same constellations. Sometimes the very high elo players know who is playing on the other side. These games have nothing to do with this elo. We are not paired with a stack of the same SR, if that would be the case it should be a stomp every game.

1 Like

Obvious handicapping going on in this game. Which explains two things.

  1. The vast difference in skill levels in the same rank. I’ve had stretches of games vs. plat like players in high silver. I’ve had totally new, stand still and die players in high silver. No way should they be in the same rank. But the matchmaker creates that problem by trying to give everyone a chance.

  2. The ridiculous streaks of these horrid teammates, then when the switch flips you get good teams all the sudden. In streaks. Every time.

Horrible idea for long term matchmaking. It might make for some even matches. But it makes it an extremely frustrating experience, long term, because you can literally predict these forced win and loss streaks due to the idiotic handicapping.
Makes winning no more fun than losing, because it’s all fake. The matchmaker decides for you, where YOU belong. You just play, could stand around and do nothing and still win your fake wins, could play perfectly and still lose your carry games.

5 Likes

I see no evidence of this. Do you have any (non-anecdotal) proof?

1 Like

If you played Halo, but someone kept changing the difficulty on you. Every other day it was set to Easy Mode and the off days it was set to Hard Mode. … yet no one ever told you about that. You would notice. You would maybe even tell other people about it, and say how obvious it seems to you. But since it wasn’t actually SAID by who ever was changing the difficulty on you, no one would believe you because you have no proof.

On the other hand, maybe a bunch of other people would come along and say “yep, i noticed the same thing and it seems very suspicious.” Maybe you’d take that as enough proof that SOMETHING weird is happening with your Halo campaign.

8 Likes

So no, you have no non-anecdotal evidence.

From _https://yourlogicalfallacyis.com/anecdotal

"You used a personal experience or an isolated example instead of a sound argument or compelling evidence.

It’s often much easier for people to believe someone’s testimony as opposed to understanding complex data and variation across a continuum. Quantitative scientific measures are almost always more accurate than personal perceptions and experiences, but our inclination is to believe that which is tangible to us, and/or the word of someone we trust over a more ‘abstract’ statistical reality.

Example: Jason said that that was all cool and everything, but his grandfather smoked, like, 30 cigarettes a day and lived until 97 - so don’t believe everything you read about meta analyses of methodologically sound studies showing proven causal relationships."

“Isolated example.”

Looks around the forums

Nah. Looks like plenty of people have said the same thing over the past 3 years. In fact, just Google “Foced Losing Stre…”

OVERWATCH.

It’s all Ovewatch. That’s no an isolated example.

3 Likes

_https://yourlogicalfallacyis.com/bandwagon

"You appealed to popularity or the fact that many people do something as an attempted form of validation.

The flaw in this argument is that the popularity of an idea has absolutely no bearing on its validity.
If it did, then the Earth would have made itself flat for most of history to accommodate this popular belief.

Example: Shamus pointed a drunken finger at Sean and asked him to explain how so many people could believe in leprechauns if they’re only a silly old superstition. Sean, however, had had a few too many Guinness himself and fell off his chair."

1 Like

It’s not bandwagon when people are still showing up and saying “i noticed this…”

2 Likes