163 replies to this topic

[Suko]

The Amazing Atomic Spaniel, on 06 October 2016 - 09:21 AM, said:

It would balance out only if the two-man groups were of equal ability. But if one were a pair if experienced players and the other were newcomers then it wouldn't be balanced at all.

How would this be any different from a regular pug match with people of different skills? It goes without saying that the 2-mans should also be paired as closely as possible via PSR.

[SmokingPuffin]

Davers, on 06 October 2016 - 02:00 AM, said:

Where is your source that premade groups using VOIP, something many of the games you list does not have for solo players, do not have a higher win rate? Just put in the link to an official source, if you please.

I don't know how to source the data you wish. Where would I find a resource that knows how often players are using third-party VOIP software? I would be scared of any company that can provide such information.

So here's what I can tell you:

WoT and WoWS have 50% group win rate by construction. The way they do pairing ensures that every match has an equal number of equally sized groups. Also, they have significant restriction on what tier of ship you bring as a team -- your team is matchmade at the level of its highest tier ship, and your division can only be at most one tier apart.
http://wiki.wargamin...Mechanics_(WoWS)

LoL experimented with matchmaking groups and solos. It worked well for them in the middle tiers, where populations are high, but at the high end of their game groups of largely professional players dominated. The way they make it work for lower levels is to adjust the elo of the team up so they play against stronger solos, but when you are looking at professional players, there aren't any solo players of sufficiently high skill to balance out the coordination edge. As a result, LoL has reverted to a maximum of 2 player teams at the high end, and solo only for the top 200 players in the world, so they always have the ability to elo adjust the game back into expected balance.
http://euw.leagueofl...-pls-ranked-pls

In a similar fashion, Overwatch has split out full teams of 6 into their own queue, and then groups of 1-5 are paired into their solo queue. Just like LoL, they found that full teams of 6 at the high level cannot be balanced against solos, because solos that are better enough to offset the coordination advantage can't be found if the team is all professional players already. For smaller team sizes, they use elo adjustment to get a team with more groups in it less individually skilled opponents, and this is working fairly well for them.
http://us.battle.net...pic/20745504371
http://www.teamliqui...ks-in-overwatch

[SmokingPuffin]

Suko, on 06 October 2016 - 10:00 AM, said:

How would this be any different from a regular pug match with people of different skills? It goes without saying that the 2-mans should also be paired as closely as possible via PSR.

Back before they got big, Rocket League applied a technique where you matchmake a team as if all its players have the same rating as its highest rated player. This resulted in groups with very different skill levels tending to lose more than 50% of games, but groups with very different skill levels are almost by definition casual so nobody minded too much.

Now that they are a huge deal, RL has gone with a more complicated super-linear geometric skill computation function that's all fancy and probably a bit better at solving the problem, but not at all necessary.

[Mystere]

SmokingPuffin, on 06 October 2016 - 10:02 AM, said:

I don't know how to source the data you wish. Where would I find a resource that knows how often players are using third-party VOIP software? I would be scared of any company that can provide such information.

So here's what I can tell you:

WoT and WoWS have 50% group win rate by construction. The way they do pairing ensures that every match has an equal number of equally sized groups. Also, they have significant restriction on what tier of ship you bring as a team -- your team is matchmade at the level of its highest tier ship, and your division can only be at most one tier apart.
http://wiki.wargamin...Mechanics_(WoWS)

LoL experimented with matchmaking groups and solos. It worked well for them in the middle tiers, where populations are high, but at the high end of their game groups of largely professional players dominated. The way they make it work for lower levels is to adjust the elo of the team up so they play against stronger solos, but when you are looking at professional players, there aren't any solo players of sufficiently high skill to balance out the coordination edge. As a result, LoL has reverted to a maximum of 2 player teams at the high end, and solo only for the top 200 players in the world, so they always have the ability to elo adjust the game back into expected balance.
http://euw.leagueofl...-pls-ranked-pls

In a similar fashion, Overwatch has split out full teams of 6 into their own queue, and then groups of 1-5 are paired into their solo queue. Just like LoL, they found that full teams of 6 at the high level cannot be balanced against solos, because solos that are better enough to offset the coordination advantage can't be found if the team is all professional players already. For smaller team sizes, they use elo adjustment to get a team with more groups in it less individually skilled opponents, and this is working fairly well for them.
http://us.battle.net...pic/20745504371
http://www.teamliqui...ks-in-overwatch

Good.

Now apply all of that to MWO.

[SmokingPuffin]

Mystere, on 06 October 2016 - 10:10 AM, said:

Good.

Now apply all of that to MWO.

I am quite confident that applying the WoT/WoWS system to matchmake small groups (1-3) and solos together will not result in any systemic win rate bias.

With regard to skill differential between groups, the elo adjustment techniques used in LoL and OW will be effective in MWO so long as the groups aren't of a dominant sizing. You can see that LoL and OW both have decided that things of nearly full team size need to only be matched against other nearly full teams. The only balanced match for a unit of 8 or 10 elite players is another unit of similar size, so once you are getting to larger scale groups I think you need to split the queues.

In related news, I think the current group queue is an absolute disaster. Bin packing is a hard problem, matchmaking is a hard problem, bin packing + matchmaking is a hard^2 problem. To make this queue not have the awful balance it currently has, I would rename it "lance queue" and require teams to go in as groups of 4, 8, or 12 mechs. I can then matchmake this directly, resulting in faster queues and more balanced games.

Edited by SmokingPuffin, 06 October 2016 - 10:19 AM.

[Mystere]

SmokingPuffin, on 06 October 2016 - 10:18 AM, said:

I am quite confident that applying the WoT/WoWS system to matchmake small groups (1-3) and solos together will not result in any systemic win rate bias.

With regard to skill differential between groups, the elo adjustment techniques used in LoL and OW will be effective in MWO so long as the groups aren't of a dominant sizing. You can see that LoL and OW both have decided that things of nearly full team size need to only be matched against other nearly full teams. The only balanced match for a unit of 8 or 10 elite players is another unit of similar size, so once you are getting to larger scale groups I think you need to split the queues.

This suggests to me that MWO does not need separate queues using this method.

SmokingPuffin, on 06 October 2016 - 10:18 AM, said:

In related news, I think the current group queue is an absolute disaster. Bin packing is a hard problem, matchmaking is a hard problem, bin packing + matchmaking is a hard^2 problem.

Finally, someone who gets it.

SmokingPuffin, on 06 October 2016 - 10:18 AM, said:

To make this queue not have the awful balance it currently has, I would rename it "lance queue" and require teams to go in as groups of 4, 8, or 12 mechs. I can then matchmake this directly, resulting in faster queues and more balanced games.

The problem with this is the extra 1-3 friends who now have to be dropped from the group. Such a situation was partially responsible for the mass exodus that occurred when group size was limited to only a maximum of 4.

[Zibmo]

Davers, on 04 October 2016 - 08:32 AM, said:

We have already done it and seen the results. 2 man groups win more often than solo players. The only real question is how does everyone feel about that.

QP already largely consists of 12-2 .. 12-4 stomps. Not sure what difference it makes, other than everyone I've recruited into this game has quit because Group Queue is beyond brutal for new players.

[Suko]

Essentially it boils down to what is more important:

A ) Improve new player retention (perhaps dramatically).

B ) Cater to the large groups and their whales (who seem fickle and claim to have already left the game until it "gets better").

Edited by Suko, 06 October 2016 - 03:06 PM.

[SmokingPuffin]

Mystere, on 06 October 2016 - 10:31 AM, said:

This suggests to me that MWO does not need separate queues using this method.

I think you can make a single queue that accepts both singles and small groups without giving advantage to either group

I don't think you can make a single queue that accepts singles and big groups; if you have an elite team of 8, there won't be good enough solo players you can put against them to balance the match. If you made a strictly single queue, it would be ruled by large units of skilled players.

Of course, today's group queue is generating considerably less balanced matches than a even a pure single queue setup would. It's really, really brutal matchmaking, and it heaps piles of unfair losses on the group you want to shelter most: new players trying the game with their friends.

Mystere, on 06 October 2016 - 10:31 AM, said:

The problem with this is the extra 1-3 friends who now have to be dropped from the group. Such a situation was partially responsible for the mass exodus that occurred when group size was limited to only a maximum of 4.

I agree, this is unfortunate. I just don't know how to make those odd sized groups work well enough. We just don't have the player population here to serve everyone's interests.

[Mystere]

SmokingPuffin, on 06 October 2016 - 12:14 PM, said:

I think you can make a single queue that accepts both singles and small groups without giving advantage to either group

I don't think you can make a single queue that accepts singles and big groups; if you have an elite team of 8, there won't be good enough solo players you can put against them to balance the match. If you made a strictly single queue, it would be ruled by large units of skilled players.

Of course, today's group queue is generating considerably less balanced matches than a even a pure single queue setup would. It's really, really brutal matchmaking, and it heaps piles of unfair losses on the group you want to shelter most: new players trying the game with their friends.

I agree, this is unfortunate. I just don't know how to make those odd sized groups work well enough. We just don't have the player population here to serve everyone's interests.

In your example, the elite team of 8 will have to wait until a suitable equivalent team (or 2 elite 4-mans, for example) also decides to drop.

[SmokingPuffin]

Mystere, on 06 October 2016 - 12:21 PM, said:

In your example, the elite team of 8 will have to wait until a suitable equivalent team (or 2 elite 4-mans, for example) also decides to drop.

We could try this, but there are risks to it. You need to somehow explain to people why they're getting long queue times, and matchmakers are not so good at transparency for this kind of thing. The Overwatch model is pretty close to this -- they have a hidden queue split for full teams of 6, but it's all behind one button -- but they have an enormous pop relative to this game, so there is no guarantee it'll work in acceptable queue times for a smaller game.

Also, I think we'd need elo ratings or similar, at least hidden ones, in order to identify fair matches. There is a huge variance of skill among PSR 1 players. Not that I'm against elo ratings; I tend to think they work well, but of course many people disagree in the context of large group gaming.

[RussianWolf]

I'll just reiterate my first post.


If they add 2-mans into the solo queue, I stop playing. Period.

If they want to add a 2-man or 2&3-man queue, I couldn't care less but the people in groups may.

[Mystere]

SmokingPuffin, on 06 October 2016 - 12:25 PM, said:

We could try this, but there are risks to it. You need to somehow explain to people why they're getting long queue times, and matchmakers are not so good at transparency for this kind of thing. The Overwatch model is pretty close to this -- they have a hidden queue split for full teams of 6, but it's all behind one button -- but they have an enormous pop relative to this game, so there is no guarantee it'll work in acceptable queue times for a smaller game.

And that is the key. It's all hidden behind a single "I Want To Play" button with a suitable explanation understandable by the player base.

SmokingPuffin, on 06 October 2016 - 12:25 PM, said:

Also, I think we'd need elo ratings or similar, at least hidden ones, in order to identify fair matches. There is a huge variance of skill among PSR 1 players. Not that I'm against elo ratings; I tend to think they work well, but of course many people disagree in the context of large group gaming.

Well, people whined loudly and incessantly against the previous Elo system (i.e. system based on win/loss) and demanded a different one based on "skill" (i.e. damage, match score, etc.). They most certainly got what they wished for even if they don't see or stubbornly refuse to admit it. So I have no sympathy on that front.

Edited by Mystere, 06 October 2016 - 12:34 PM.

[Deathlike]

There's a point where you can't really try to have both "equal skill" MM with "equal group size" MM w/o having to make compromises with waiting times.

Even then, something has to give... and skill discrepancies happens to be the one PGI selected

People need to admit to themselves that this is still primarily a population issue first and foremost (and secondarily a NPE/skill issue of people not really getting better) and it won't fix itself as currently constituted.

[Wintersdark]

Deathlike, on 06 October 2016 - 02:08 PM, said:

There's a point where you can't really try to have both "equal skill" MM with "equal group size" MM w/o having to make compromises with waiting times.

Even then, something has to give... and skill discrepancies happens to be the one PGI selected

People need to admit to themselves that this is still primarily a population issue first and foremost (and secondarily a NPE/skill issue of people not really getting better) and it won't fix itself as currently constituted.

Exactly. Fundamentally, no MM changes are going to fix people's issues with "the matchmaker"; we simply lack the pop to have matches be more focused than roughly 30% of the population being eligible for any given match.

[Mystere]

Wintersdark, on 06 October 2016 - 02:25 PM, said:

Exactly. Fundamentally, no MM changes are going to fix people's issues with "the matchmaker"; we simply lack the pop to have matches be more focused than roughly 30% of the population being eligible for any given match.

Now imagine what PGI could have accomplished (or broken ) using all the time, effort, and other limited resources needlessly spent on dealing with MM complaints. They should have just stuck with the original (non)matchmaker.

[Suko]

Deathlike, on 06 October 2016 - 02:08 PM, said:

People need to admit to themselves that this is still primarily a population issue first and foremost (and secondarily a NPE/skill issue of people not really getting better) and it won't fix itself as currently constituted.

I completely understand it's a population issue. When Russ asked the community for advice on what to do w/ group queue a year back, I suggested reducing grouping options to 1-4, 6, and 8. Anything more than 8 should be playing FW. If I had it my way, PGI would give up on FW and give us a group and quick play (dirty casuals) queue.

Problem was that there were some very vocal large groupies who said they would rage quit if this happened. It never did happen, but most of them have since rage quit anyways. It shows what pandering to the 1% will get you.

As a small unit guy it infuriates me to see these 8+ size group guys complain about having to "break up" their group. So what? Break it up into two 4 mans. Is that really so bad? You still get to play with more buddies than I usually do and by breaking up your 10 man power block, we've improved the quality of the game for everyone.

I have to admit that I've got a major chip on my shoulder regarding the demands from these large groups. All they do is whine and complain and demand and even when things DO go their way, they leave en-mass and say the game is lacking. I don't see why PGI is even paying attention to them anymore. Let's get some fresh, non-salty blood into this game. The only way to do that is to allow 2 people to play together without getting creamed.

If you all think there's a better way to pull it off than my proposal, than that's fine with me. I don't see any serious flaws or issues in the following:
- Only ONE 2-man team on each side per match.
- The MM will not throw a 2-man onto one team until it finds another 2-man for the opposite team.
- Put tonnage min/max to prevent excessive abuse.
- Allow 2-mans to OPT-IN to the standard group queue as well.

Final point: TEST IT FOR A MONTH AND SEE THE RESULTS.
I cannot emphasize this enough. Just like 3PV, the community could be making this a mountain from a mole hill. I don't think trying it for a single month will destroy MWO. At worst, a few dozen players will "rage quit" for a month and then come back after the aftermath to see what's happening.

Edited by Suko, 06 October 2016 - 03:18 PM.

[Mystere]

Suko, on 06 October 2016 - 03:13 PM, said:

I completely understand it's a population issue. When Russ asked the community for advice on what to do w/ group queue a year back, I suggested reducing grouping options to 1-4, 6, and 8. Anything more than 8 should be playing FW. If I had it my way, PGI would give up on FW and give us a group and quick play (dirty casuals) queue.

Problem was that there were some very vocal large groupies who said they would rage quit if this happened. It never did happen, but most of them have since rage quit anyways. It shows what pandering to the 1% will get you.

As a small unit guy it infuriates me to see these 8+ size group guys complain about having to "break up" their group. So what? Break it up into two 4 mans. Is that really so bad? You still get to play with more buddies than I usually do and by breaking up your 10 man power block, we've improved the quality of the game for everyone.

You're a founder. Unless you were absent then, you should know very well what happened the last time PGI limited group sizes.

Suko, on 06 October 2016 - 03:13 PM, said:

I have to admit that I've got a major chip on my shoulder regarding the demands from these large groups. All they do is whine and complain and demand and even when things DO go their way, they leave en-mass and say the game is lacking. I don't see why PGI is even paying attention to them anymore. Let's get some fresh, non-salty blood into this game. The only way to do that is to allow 2 people to play together without getting creamed.

Barring any changes to the queues, use private matches for training purposes. Others have done so.

[Deathlike]

Suko, on 06 October 2016 - 03:13 PM, said:

I completely understand it's a population issue. When Russ asked the community for advice on what to do w/ group queue a year back, I suggested reducing grouping options to 1-4, 6, and 8. Anything more than 8 should be playing FW. If I had it my way, PGI would give up on FW and give us a group and quick play (dirty casuals) queue.

Problem was that there were some very vocal large groupies who said they would rage quit if this happened. It never did happen, but most of them have since rage quit anyways. It shows what pandering to the 1% will get you.

Well, PGI isn't great in the decisionmaking business, regardless of what you think.

Quote

As a small unit guy it infuriates me to see these 8+ size group guys complain about having to "break up" their group. So what? Break it up into two 4 mans. Is that really so bad? You still get to play with more buddies than I usually do and by breaking up your 10 man power block, we've improved the quality of the game for everyone.

I've been in a small unit before. It was totally atrocious to me. How do practice with playing with others? Private matches? I'm not committing premium time for that either (and I have hordes of it, but not worth spending for that at all). At that time, I NEEDED C-bills to function. Nowadays I don't care/need any of that.

Quote

I have to admit that I've got a major chip on my shoulder regarding the demands from these large groups. All they do is whine and complain and demand and even when things DO go their way, they leave en-mass and say the game is lacking. I don't see why PGI is even paying attention to them anymore. Let's get some fresh, non-salty blood into this game. The only way to do that is to allow 2 people to play together without getting creamed.

If you all think there's a better way to pull it off than my proposal, than that's fine with me. I don't see any serious flaws or issues in the following:
- Only ONE 2-man team on each side per match.
- The MM will not throw a 2-man onto one team until it finds another 2-man for the opposite team.
- Put tonnage min/max to prevent excessive abuse.
- Allow 2-mans to OPT-IN to the standard group queue as well.

Final point: TEST IT FOR A MONTH AND SEE THE RESULTS.
I cannot emphasize this enough. Just like 3PV, the community could be making this a mountain from a mole hill. I don't think trying it for a single month will destroy MWO. At worst, a few dozen players will "rage quit" for a month and then come back after the aftermath to see what's happening.

What bothers me more is people not wanting to work together regardless of the reason. That is why we have a god forsaken solo queue where people don't bother to learn even basic teamwork (helping the guy next to you at the current moment).

Group sizes are not really the problem... again, the core problem is people not wanting to work together regardless of the reasons. These are the people that source of any complaints with regards to gameplay in MWO.

Edited by Deathlike, 06 October 2016 - 03:24 PM.

[Tarl Cabot]

Davers, on 06 October 2016 - 07:42 AM, said:

I don't understand why everyone keeps saying that having a 2 man on each side will "balance out" when the direct evidence provided by PGI clearly states that it doesn't, and 2 man groups win more often than solo players.

Then that guy I quoted brings up LoL as an example of how integrating group and solo que is totally fine- completely ignoring how Riot recently apologized for "shattering the player's trust" by doing it, and all the threads about how its pointless dropping against all the units on VOIP.

I would say it did not work out the first time is due to PGI attempting to be "fair" by calculating the average of the group.

I would like to see PGI to run with it for at least a week but with exceptions: 2-group is not averaged, it is seeded by the higher tiered player, so a Tier 1 + Tier 4 will drop with Tier 1 settings. Set a weight limit, min 80 to 155 max. Prevents 2*80tons. Max one group per side. Also, that 2-man group is permanently set for group queue (no opt out), with an option to opt into solo queue.

There is VOIP in-game now, unlike before and the drops are now 12vs12 instead of 8vs8.

Edit.. On second thought, lets not do it. There are too many ways it can really be messed up /shudders.....

Edited by Tarl Cabot, 06 October 2016 - 03:32 PM.