Built-in M+ rating is a big mistake

As we all know, Blizzard are exposing their own rating for M+, which is a copycat from RIO with little to no changes so far, in the game. This is a big mistake and in this post I will explain why.

First, let’s check what RIO does. On paper, it does nothing bad, it just shows which instances you completed on which level to others and lets them make a decision on whether to take you or not. The problem is in that last part - that it lets other people make a decision on whether to take you or not, based on the numbers presented. Because these other people are people, their decision is subjective.

You could ask - but what about the numbers presented? The decision, while subjective, is still more informed than it would be otherwise, no? The answer is that it does not really work that way.

A simple example: let’s say there are two people in the queue with equal rating, both specs are a good fit for your forming party, but one of the specs is meta and the other is not. Who you are going to pick? You are going to pick the meta spec. It’s not a question, by the way, it’s a statement, this is true de-facto, it’s in the numbers. So, the rating was telling you that two characters were equal, but you still had a clear preference. The chances of the two characters should have been equal, but you made them unequal. You introduced your own bias.

The community has many such biases. They are introducing an error into each decision. And what’s worst, because the beneficiaries of these errors complete more instances and get more rating and get more gear and, yes, get more experience as well, these errors snowball.

The truth about RIO is that it is NOT a measure of capabilty. It is much more a measure of how conforming you are to the collective biases of the community. This does NOT mean that a non-meta spec that completed some +10s is a better fit for +15s than a meta spec that completed some +15s. No, the latter guy is a better fit for +15s currently. He has more gear and more experience. The problem is, the latter guy has more gear and more experience and the former guy less of everything because the system was reflecting community biases at every step and this was boosting the latter guy and drowning the former guy. If the former guy was being accepted at the same rate as the latter guy, he would have been on par with him wrt all of gear, experience and rating. But he is not and that’s a problem.

Ratings are a good thing. But ONLY when they are paired with automated matchmaking. Whenever they are paired with manual matchmaking, they are overrun by the collective biases of the community. Wrt WoW and M+, ratings don’t matter at low instance levels (eg, before +5, because that’s easy, risks are low) and at high instance levels (eg, after +20, because if you got up there, you have everything and picks are going to be based on setups, who knows whom, etc). But they matter in the middle, and, unlike in PVP where ratings form a pyramid and most people are in the low portion of it, in M+ ratings form a skewed bell-shaped figure and most people are in the middle. In the middle, where ratings are the main tool used to compose groups.

So, RIO is a bad idea, because all ratings paired with manual matchmaking are a bad idea. And Blizzard’s own rating is a bad idea for the exact same reason. Blizzard’s own rating is an even worse idea than an external rating from RIO, because a built-in rating is going to be used more widely and so the biases it spreads are going to snowball harder.

By introducing their own built-in rating while still staying with manual matchmaking, Blizzard are:

  • amplifying all issues with meta vs non-meta specs,
  • making it much harder to balance specs and classes, because they plain won’t see half of them in material amounts other than at low keys,
  • letting a lot of people who play unfavored specs down, forcing them to “deal with it” by switching to a “better” spec or class, making it all even more imbalanced.

An automated matchmaking system would avoid all three of these things. Alas, it seems we cannot have an automated matchmaking system, for whatever reason. But if we cannot have one, then at least Blizzard should not be taking things that are already bad and make them worse! If anything, instead of legitimizing the idea of using rating in a manual matchmaking system that de-facto amplifies community biases, they should have made it MORE difficult to check what rating you are: by taking away API means to show tooltips in an LFG window and by stopping tracking highest completed instance level altogether in the armory.

Built-in M+ rating is a big mistake, which is going to cause a lot of tears, a lot of apathy and disillusionment, and will ultimately make M+ a worse place to be and will make less people participate.

(Really sorry for the wall.)

18 Likes

how is this going to do anything but enforce more manual checks by players, though? just because you do away with raider io it doesn’t mean players will stop vetting the people they invite to their groups.

you know why? because people want to time their keys, and they’ll do whatever they can to decrease their chances of depleting. if RIO is gone, they’ll ask for achievements. if achievements are gone, they’ll stop inviting non meta specs that don’t overgear the content altogether.

4 Likes

While I’d love to see this, people will just leave keys they don’t want to do because of whatever reason.

Built-in M+ rating is better than external tools, both for usability and accessibility of the tool.

I also think Rio is a fine tool, I like it.

From the text that you quoted: … by stopping tracking highest completed instance level altogether in the armory.

I mean, yes, I agree, people will go check achievements and ilvl. But because such checks will be more time consuming, the level at which they won’t be done will increase (eg, instead of ratings not mattering before +5, such things won’t matter before +10 or so). The real solution, however, is automated matchmaking.

2 Likes

Biased community perception exists with and without a ranking system. Switching from RIO to a standard Blizz tool does not change this. As long as damage and healing numbers are output in some way, hardcore players will analyze them and publically discuss their findings.

Automated matchmaking would reduce the problem of bias for puggers - I agree with you there. We can hope that Blizz will consider this in later versions. But claiming that the new tool is making things worse is wrong. I would actually argue that it makes things better for newbies because it will make it clear to them that they should achieve +2, +3, +4s etc first before trying +15s, even if they have 210 ilvl.

It exacerbates the issues. Instead of having a step forward, we are having a step back. That’s my point.

Good, then they’ll be forced to make balance adjustments.

You say that as if that isn’t currently what is happening to non-meta classes.

I think you’re over estimating how much people care about the meta at low keys.
The other day I did a key with a 134 ilvl MW and a 170 ilvl Hunter.

The reality is simple, people use RIO and there’s no avoiding it. Blizzard either needs to ban it or incorporate it. They chose to incorporate it, which is the correct move. You can’t have new players seeing this third party tool talked about and have no idea what it is or what it means. It’s been integrated with the community too much at this point to undo it.

1 Like

Not really. Biases in the system are making balancing harder, not easier. As in, yeah, you see that spec X is not popular, but you don’t have much in the way of comparable numbers because so few run it. So you end up saying that spec X is perhaps fine and it’s just that the community have biases. Which is true regarding the community, but might be true or not regarding the spec. So in the end you just write it off as “it’s fine” and you have less info on how even to balance. It’s much worse for balance, not better.

And all your problems with Rio can be easily solved with 1 simple easy to use solution.
Hate rio, metas vs non metas and all that? Fine.
Create your group and invite 3 survival hunters for all I care. Invite people based on who has the prettiest name or transmog. The sky is the limit!
Nobody is being forced to use rio or any other vetting tool-method. Your key = your rules.
If you signup for somebody’s else key then accept that he may make whatever requirements he wants. No matter how silly,stupid, outrageous you may personally find it.
His key=his rules.

15 Likes

Without a rio people will just pick highest ilvl + meta. Now at least non meta’s have a chance by having an experienced rio-record.

9 Likes

ilvl acquisition is less biased than rating acquisition. You can ask “but what if that guy got ilvl from PVP?” - but that’s exactly the point! ilvl is LESS info, yes. This is GOOD. Because manual matchmaking snowballs biases and having LESS info reduces your reliance on it. It brings the plank before which you don’t even check other than very cursorily (as in, ilvl 205? fine, wherever it came from) higher, that reduces the space affected by biases.

The real solution is, as I said, automated matchmaking. But if we cannot have that, doing nothing at all would have been better than elevating RIO into a built-in rating that will be used more widely now. And restraining RIO and similar attempts at rating by taking away the data that can be used to form it would have been marginally better.

Seems you missed the whole point about meta not getting any chance anymore.

It’s the other way around, they would get more chance because the level at which everyone gets picked and the checks are very basic will go up. Ie, right now, anything before +5 is for anyone. After changes the number will go up. That’s an improvement. A marginal one, yes.

More chance? How can the non meta player stand out over the meta without something like rio?

Read my posts above.

Rio was born out of need for peps who vet their applicants. Even If Blizz did nothing. There would still be methods and tools to gauge applicants experience. So hide your char behind anonymity all you want which you seem to target anyway. No pug leader who actually wants to time a key will invite anyone based on “who asked the nicest”. They would find a way how to vet people…or the pugging scene at whole would just collapse.

3 Likes

Not seeing it. sorry.

ilvl acquisition is less biased than rating acquisition. You can ask “but what if that guy got ilvl from PVP?” - but that’s exactly the point! ilvl is LESS info, yes. This is GOOD. Because manual matchmaking snowballs biases and having LESS info reduces your reliance on it. It brings the plank before which you don’t even check other than very cursorily (as in, ilvl 205? fine, wherever it came from) higher, that reduces the space affected by biases.

Sure. But this will become feasible higher than now.

1 Like

Yeah well. We got 2 the same ilvl players applying. We got tons anyway. So we see a meta just a couple ilvl lower than the non meta. Meta gets invited.

Do you do this for +3 now? No. That plank will go higher. I said why three times. Marginal improvement.