Inc PvP changes; q-ing needs fix first

#1 - Dec. 2, 2013, 9:48 p.m.
Blizzard Post

So with the upcoming massive changes to sPvP, inevitably, this would draw more of the PvE crowd into the sPvP world (or much less those who never ever touched sPvP at all) due to the introduction of more achievements and rewards. Indeed this is extremely exciting news for practically a lot of us players. However, some of us are already groaning at the thought of players with little to no experience randomly queuing for rated matches. As much as i hate when this happens as the match gets uneven, and a supposed MMR-based queue should have made it even (or so I’m told), this is hardly the most concerning issue I’m thinking of but rather the issue at hand of afkers. Since the dawn of the game, the fate of solo queue matches is ridiculously dependent on whether or not one of the players on a team(albeit now having separate queues). So with more players undoubtedly arriving or coming back to tournament matches, is it safe to say that these changes would encourage more competitiveness? Or do queues still needs to be fixed? On a side note if any dev is reading this, would there be an implemented fix before these changes?

#2 - Dec. 2, 2013, 9:54 p.m.
Blizzard Post

Well, now that the news is out I guess I can spoil this one tidbit regarding matchmaking. In addition to your MMR, we will also be using your ladder tier when determining who you play against, so as you travel up the ladder the odds of being matched with new players is reduced. This filtering becomes less strict as you wait in the queue, so people can still find matches during low concurrency.

I’ll be working with Josh and the others to get a blog post out and start a discussion about matchmaking in the near-ish future, and we will likely talk about the 4v5 issue at the same time.

What do you guys thing we should do in the meantime while we continue to work on the new server infrastructure, matchmaking, ladders, etc.?

#7 - Dec. 2, 2013, 11:57 p.m.
Blizzard Post

Using the ladder system as a factor of tourney queues is definitely a great idea! However the diminishing restriction over time feels rather not flexible, how about say, an option to whether or not the player is willing to risk queueing with someone with a rather lower level on the ladder and MMR.

I know several people would like an option to wait forever for a good match, but I think it would be best to design the system around the more typical use case. If there is still a need for it then we can certainly revisit the topic.

Ready buttons do seem to be the most efficient solution to afkers and I’m sure you guys have never overlooked this solution (having talked to josh before in game regarding this issue) and its definitely long overdue so i assume there are some troubles in implementing this?

I don’t think it’s really an matter of trouble but of time. We’re prioritizing our time toward the new infrastructure rather than the existing. This is also easier to implement with the new servers than it would be to add to the old.

#14 - Dec. 3, 2013, 3 p.m.
Blizzard Post

Since the new matchmaker rolled out, things have been worse for me. Before the change I was grouped with players of similar rank with the rare chance of having someone who is low rank on the team.

Using the ladder tiers in matchmaking should help with that. We’re also still discussion ways to better place new players onto the MMR curve.

#20 - Dec. 3, 2013, 10:28 p.m.
Blizzard Post

I actually think the leaderboards now are extremely inaccurate ever since the leaderboards are reset. I.E. winning 1 match almost immediately makes you climb multiple ranks up the leaderboard. Plus the MMR based queue doesn’t really work since the dawn of tourneys, being queued with ~20 ranks even before the separation of solo and team queue. I sure think it wouldn’t be simple adding a slider for this as it might be pretty tedious to adjust the formula based on the slider but as justin has said, if there is a need to it then they could get back to it.

Actually the leaderboard behavior is consistent with what we were expecting as best I can tell. The large jumps are normal for two reasons.

  1. A larger starting deviation, which is by design, that allows a player’s MMR to settle into a more accurate position much faster.
  2. The curve itself is much steeper because players haven’t finished spreading out. That means the difference between you and your neighbors on the leaderboard is likely lower than before, allowing smaller changes to have a larger impact.

Those two may seem to contradict each other, but the former applies to individuals and the latter applies to sum of every player.

#29 - Dec. 7, 2013, 4:14 p.m.
Blizzard Post

snip

I’m currently experimenting on my personal time with two things.

  1. A better way to distribute ratings adjustments so that players that are carrying feel less impact for a loss and more impact from a win. I’m uneasy with this change because it would help high MMR players stay high instead of settling them down to where they should be, but may be needed to get some players out of MMR hell.
  2. Adjusting MMR based not on win/lose, but the ratio of the final score. So that a final score of 400/500 would be seen as a minor loss instead of the same as a final score of 100/500. My concern with this change is that it would slow how quickly people’s MMR fluctuates, but that may be a good thing in the long run.

The first may be unnecessary if we use ladders as a matchmaking metric. If we do that, excluding the beginning of a season, you shouldn’t be matched as often with players at a lower skill level than your own.

Also, the reason I said players don’t have to carry is because the majority of the time (not always, I’ve seen the screenshots) the other team has players in the same situation as you. You aren’t really that unique as far as any individuals game goes, so it isn’t as one sided and unfair as many people seem to think.

Now that we’ve decided to keep ranks, I’m going to push for gating arenas, or at the very least using rank as a metric in matchmaking.

#31 - Dec. 7, 2013, 5:51 p.m.
Blizzard Post

The problem here is to see who is carrying…

Exactly right. It isn’t an easy thing to detect without creating incentives for certain ways of playing that could ruin game play as a whole.

With that you would also make a 4vs5 loose much more significant and painful. A 400 to 500 was already a good fight, and you team was almost even. I think you can get over one of those looses.
The problem is the other one, where you end up outnumbered, and you already get angry over this loss, I don’t think you should harm those poor guys further.

Right again. Leaving should always award you a full loss. I agree with the second part as well, we shouldn’t punish the remaining team just because they now have an unfair disadvantage. We can factor that into the equation though. Maybe we get the ratio after dividing the team scores by the number of players in the team.

New player playing with their first games 2-1 or 3-0 and ending up in the Top 10. Please take a look at those occurrences and adjust the system accordingly. Nobody will take your ladder or leaderboard or whatever serious when those things will happen.

One of the reasons I’m looking forward to the ladder system. I don’t think MMR was a good choice for leaderboards simply because it encourages players to distort them which ruins the quality of matchmaking as a whole.

What happens when the above behavior is not getting fixed in the new system?

I don’t think we’re ready to talk about ladder details yet, but my understanding is that it shouldn’t apply.

4vs5, people who are flaming / afking: Why are you not banning the players who are constantly doing this?

We are looking at ways to amp up the dishonor system so it is more effective. I believe we have a change coming on the 10th, but we will likely more in the future as well.

Even though the filtering should increase over time, it should also be restricted to not increase too much.

Agreed. I’m going to be writing all this so it can be tweaked via configuration files which will allow us to adjust them in response to the real world, without requiring a build.