Split matchmaker into mapgen and regular map queues

It has been suggested before, but then it kinda went quiet again. I suggest that we put the random map generator maps into their own matchmaker queues so we got two queues per team size. One with custom maps only and one with map gen maps only for a total of six. This way everyone can choose if they want to play only with or without mapgen or a mix of both by multiqueueing.

It is possible that some queues might see a lot less activity, but firstly I don't think that is very likely and secondly we can monitor this and then think of a way to make the experience better. As a last resort we could remove some queues again if we see that they are totally dead, but I really don't expect this to happen.

Now, before the "I want my setons/dual gap/whatever queue" dudes inevitably show up: This is not comparable. Map gen has been in literally every map pool so far, so we are basically only splitting the pool. No classic map is included in the pool all the time.

What about a mapgen 3v3 queue? That way you could both cater for anyone that wants a 3v3 game, and people who want mapgen. While I like mapgen, I worry that having 6 queues (1v1, 2v2, 4v4, each for curated pool and mapen) would be too many options)

Still waiting to get a single good map gen map in tmm before having a whole queue for it. I’d rather have a whole queue for sentons, frankly.

I like the idea of @maudlin27 , reasons being:

• (1) 6 players are less prone to startup problems (not connecting)
• (2) 6 players are less prone to connectivity problems in general (lag)
• (3) 6 players is almost guaranteed that the game can be played from start to finish with no sim slow downs (assuming 600+ units per player means we end up with with about 3600 units, which is far less then what a budget cpu is able to run soon)
• (4) 6 players is unusual meta wise, and there is likely no dedicated air slot

It would also prevent us from having six queues, even though you can enter multiple queues.

Team mapgen maps often have far too many mexes. Path-ability is also very hit and miss.

@jip said in Split matchmaker into mapgen and regular map queues:

I like the idea of @maudlin27 , reasons being:

• (1) 6 players are less prone to startup problems (not connecting)
• (2) 6 players are less prone to connectivity problems in general (lag)
• (3) 6 players is almost guaranteed that the game can be played from start to finish with no sim slow downs (assuming 600+ units per player means we end up with with about 3600 units, which is far less then what a budget cpu is able to run soon)
• (4) 6 players is unusual meta wise, and there is likely no dedicated air slot

It would also prevent us from having six queues, even though you can enter multiple queues.

This sounds more like arguments against 4v4 in general than being mapgen-specific. Why do you think 6 queues would be too much but 4 would be ok?

Having a 3v3 queue would fill a gap for me, where you end up in a game with no dedicated air player and nobody knows what to expect because the map is (always) generated.

To have six queues feel a bit much

There is currently no 3vs3 queue since that format is essentially dead. Nearly nobody plays it and there are no maps for it.
If it would be just a map problem then the mapgen 3vs3 would solve that. The question is would enough people play it?

@jip same, it would be the sweet spot between not as try hard as 2v2 but not as chill as 4v4. 3v3 mapgen sounds like a world of fun

@tagada said in Split matchmaker into mapgen and regular map queues:

There is currently no 3vs3 queue since that format is essentially dead. Nearly nobody plays it and there are no maps for it.
If it would be just a map problem then the mapgen 3vs3 would solve that. The question is would enough people play it?

As far as I can recall it was primarily dead because of not having sufficient maps. And we can't answer the last question without trying

I don't think we have the player base to support more queues (except for dedicated setons/gap queues, but that is a different discussion).

I missed this thead. I have the exact same thought. A 1v1 mapgen queue would certainly get used. Not sure about larger team mapgen queues.

For 1v1 at least what's wrong with 3 buttons: top one is mapgen + ladder pool maps (each map weighted equally), next one is just ladder pool maps and next one is just mapgen? Would be interesting to see usage after a month or two.

Same as with non-fullshare: admins believe it is bad design to have a check button that is rarely used. Even though it would not hurt to have 100 of these buttons.

Well I'm not suggesting 100s of buttons - that would be too confusing, and likely to switch off the devs straight away. Since mapgen is popular I'm not sure it would be rarely used. Can't we try it?

@Melanol

On choice in general, see also the TED talk from Barry Schwarts on choice, or a written summary of one of its examples.

And when you bing the question whether too much choice is good, you end up with:

Luckily Google is less opiniated in this case and links to an article that also mentions being deprived of choice being a bad thing.

On choice within the context of the queues: there isn't really that much choice if you want to find a game with other players. Other players need to have their options configured in such a way that they are compatible with yours. And they need to be within a certain rating limit. The already low pool of players would therefore just fragment even further, resulting in longer and longer waiting times.

You can't state that having more choice (or let alone an excess of it) is always better.

too much choice is only a problem for people that don't understand what they are choosing or the differences between available choices, so yes it is always good to have more choice, unless you don't understand what you are choosing between and need to be guided by people who know what you should choose until you learn it yourself, all the resulting "bad" stuff from having more choices comes from ignorance

for matchmaking I wouldn't know but the playerbase is so small because everyone is playing dual gap that obviously if there were more options that all need to be exactly the same for players to match with each other, as you keep increasing number of options while keeping the same playerbase size, eventually players would never be able to match with each other because of differing options between them, but this is a problem caused by small playerbase, not because "choice is bad"

@jip said in Split matchmaker into mapgen and regular map queues:

Choice

Check buttons are not radio buttons. If you click "Select all" and don't care what you play, there isn't much confusion.

But how does having a few more discrimatory queues lower the chance of getting a game when you have the option of queuing in more than one and you can see how many players are in each. They're not mutually exclusive

@tagada said in Split matchmaker into mapgen and regular map queues:

There is currently no 3vs3 queue since that format is essentially dead. Nearly nobody plays it and there are no maps for it.
If it would be just a map problem then the mapgen 3vs3 would solve that. The question is would enough people play it?

Try it?

2v2 is basically dead anyway.

Coincidentally, I wonder how much effect mapgen vs known maps has on rating. Using a dedicated queue (+ rating) might be worthwhile anyway.