12 replies to this topic

[Zyllos]

In case some individuals missed this:

Russ Bullock said:

There is no doubt that conversations about a games Match Maker are always hot topics with lots of opinions on how things are or aren't working as they should be, MechWarrior Online is no different.

Although we look forward to working closer with the community on ways we can improve the match maker's performance to provide the most competitive match's possible. I wanted to discuss the implementation of one particular change that should make an immediate impact.

One of the more common problems currently is that the match maker would be able to put together a very competitive game based on the players skill levels, however one or more of those players or groups have incompatible game modes selected. Therefore the match maker has no choice but to avoid that high quality match and continue to wait or even eventually put together a game that is significantly lower quality. This can result in a play experience that is less fun or even frustrating when lower skill players are set to play against highly skilled teams.

What I propose to do is change the functionality of the game mode selector window. Currently you select the game modes that you accept to play, and are guaranteed to avoid the game modes you do not select. The way the window will function after the change is that each box selected counts as a vote. Therefore the match maker will be able to put together the best game possible based on the players skill level then at that point will take a tally of all the votes, with the highest vote count determining what game mode you play. A tie will result in a random selection of the game modes tied.

A change like this should at the very least increase the frequency of competitive matches by a modest degree. At least we would like the opportunity to see just how much it helps with our live population numbers.

Also if the player base accepts this change and it works out well, we would consider putting this voting system in place for map selection. Players will have the ability in the front end UI to select the maps that they vote for, similar to game modes. The code will then tally votes to select the map the game will take place on.

The timing of the game mode selector change would be for the Oct 7th patch. The timing of the map vote selector is unknown as it would take additional UI work but we would push to deliver it inside of 2014.

Please let us know with your vote if you would give up the ability to guarantee your game mode for having a higher chance of getting in a competitive match.

Quote

http://mwomercs.com/...-match-quality/


There is some major problems I have with this. The ability for a group of players to basically choose what match they want, in the group queue, will lead to further decrease the modes played (less Conquest) and maps selected (Terra Therma) if extended to maps.

As per this suggestion, a group can choose to not vote for maps and modes. It's already a given that Conquest will not be played by most players. So for the odds of Conquest being selected for play will be just as small as it is now, if not even smaller.

This is because for Conquest to be selected for play, one side has to have the desire to play Conquest. But since Conquest is considered inferior for game play, the odds of Conquest tying in the map votes for it show up will be extremely low.

If this mechanic is extended to maps, you will see the same happen to Terra Therma and River City Night. These are some of the least enjoyed maps by the community.

Look at the example below:

• Group A, with 12 members, only wants to play Skirmish on Forest Colony.
  
• Group B, with 12 members, will play Conquest and Assault but on any map.
  
• Group C, with 12 members, will play Assault and Skirmish on River City Night.
  
• Group D, with 12 members, will play Conquest and Assault and on River City.
  
  • If Group A gets paired with Group B, Forest Colony is guaranteed to be picked.
    
  • If Group A gets paired with Group C, Skirmish is guaranteed to be picked.
    
  • If Group A gets paired with Group D, then the mode will be randomly selected and the maps will be randomly selected between Forest Colony and River City.

If you notice, from the above situations, the group with the narrowest selection of types will dictate the types of maps and modes that will most likely to be played. This is because if the other group places even one more vote in favor of the narrowest selection on the other side, it will automatically win.


The only time a random selection is made is if both sides show no overlapping selections, meaning that one side will have to play a map/mode that they did not vote for. All other cases, the group with the narrowest selection will always get at least the map they want or the mode they want. Because of this phenomena, the types of modes and maps being played will be along an ever decreasing selection of maps and modes.

This will be especially pronounced in a specific grouping of players, like for example, large group play. If a unit decides to want to play Conquest, but will allow other modes, the Conquest mode will be at the whelms of their peers to select Conquest. If Conquest is shunned by this grouping of players, then a single group with that grouping will rarely, if ever, have the chance to play Conquest UNLESS they decide to be disruptive to the system and select Conquest only, forcing the mode to be random.

As just a FYI, I am assuming a group that is dropping is making a single vote for the entire group on what they want to play.

The way this is presented, in a Battletech universe, is done poorly and out of context. Instead of allowing for specific modes of combat and environments to fight in be directly voted upon (never happens in a fight), instead players should have a selection of mechs available to them to pick for a specific mode and map combination that comes up.

Having a pool of mechs that is available for selection, players can tailor mechs to types of environments but can not make mechs directly to specific map and mode combinations as there will be more map/mode combinations than available mechs to a player.

Thus, I hope PGI and the community reconsiders the idea of allowing players to vote on modes and maps as those modes and maps which are considered poor for gameplay, for whatever specific reasons, will be seen less with this mechanic.

As a FYI, I voted no.

The above is incorrect! Disregard.

Edited by Zyllos, 22 September 2014 - 09:25 AM.

[Russ Bullock]

Zyllos, on 21 September 2014 - 09:30 PM, said:

In case some individuals missed this:



http://mwomercs.com/...-match-quality/

There is some major problems I have with this. The ability for a group of players to basically choose what match they want, in the group queue, will lead to further decrease the modes played (less Conquest) and maps selected (Terra Therma) if extended to maps.

As per this suggestion, a group can choose to not vote for maps and modes. It's already a given that Conquest will not be played by most players. So for the odds of Conquest being selected for play will be just as small as it is now, if not even smaller.

This is because for Conquest to be selected for play, one side has to have the desire to play Conquest. But since Conquest is considered inferior for game play, the odds of Conquest tying in the map votes for it show up will be extremely low.

If this mechanic is extended to maps, you will see the same happen to Terra Therma and River City Night. These are some of the least enjoyed maps by the community as it limits builds and is annoying to fight in.

Look at this like this:

• Group A, with 12 members, only wants to play Skirmish on Forest Colony.
  
• Group B, with 12 members, will play Conquest and Assault but on any map.
  
• Group C, with 12 members, will play Assault and Skirmish on River City Night.
  
• Group D, with 12 members, will play Conquest and Assault and on River City.
  
  • If Group A gets paired with Group B, Forest Colony is guaranteed to be picked.
    
  • If Group A gets paired with Group C, Skirmish is guaranteed to be picked.
    
  • If Group A gets paired with Group D, then the mode will be randomly selected and the maps will be randomly selected between Forest Colony and River City.

If you notice, from the above situations, the group with the narrowest selection of types will dictate the types of maps and modes that will most likely to be played. This is because if the other group places even one more vote in favor of the narrowest selection on the other side, it will automatically win.


The only time a random selection is made is if both sides show no overlapping selections, meaning that one side will have to play a map/mode that they did not vote for. All other cases, the group with the narrowest selection will always get at least the map they want or the mode they want. Because of this phenomena, the types of modes and maps being played will be along an ever decreasing selection of maps and modes.

This will be especially pronounced in a specific grouping of players, like for example, large group play. If a unit decides to want to play Conquest, but will allow other modes, the Conquest mode will be at the whelms of their peers to select Conquest. If Conquest is shunned by this grouping of players, then a single group with that grouping will rarely, if ever, have the chance to play Conquest UNLESS they decide to be disruptive to the system and select Conquest only, forcing the mode to be random.

As just a FYI, I am assuming a group that is dropping is making a single vote for the entire group on what they want to play.

The way this is presented, in a Battletech universe, is done poorly and out of context. Instead of allowing for specific modes of combat and environments to fight in be directly voted upon (never happens in a fight), instead players should have a selection of mechs available to them to pick for a specific mode and map combination that comes up.

Having a pool of mechs that is available for selection, players can tailor mechs to types of environments but can not make mechs directly to specific map and mode combinations as there will be more map/mode combinations than available mechs to a player.

Thus, I hope PGI and the community reconsiders the idea of allowing players to vote on modes and maps as those modes and maps which are considered poor for gameplay, for whatever specific reasons, will be seen less with this mechanic.

As a FYI, I voted no.

This is actually incorrect.

You can think of it like every mode and map getting at least one vote. So in this situation:

• If Group A gets paired with Group B, you would have forest colony taking about 50% of the chance and all the rest of the maps an equal share so it ends up being a 50% chance of getting Forest colony. It can not be 100% unless all 24 vote only for Forest Colony.

Same logic would apply for game modes so in your next example Skirmish would have the greatest chance perhaps 66%. I need to double confirm all the logic with the engineers ( Karl in particular ).

[Too Much Love]

I voted no.

The reason is mm quality now is so low (I give it 2 out of 10) that by simply removing the ability to choose game modes it won't become much better. In result we will lost the right to choose game mode, that's all. But mm won't improve.

All the matches I see are the stomps. MM determines the outcome of the game, you rarely can change it. So if it is how it works, howit can work better?

[Russ Bullock]

Well as stated it would work better by allowing more matches to be competitive. Please read this post and in particular my respones.

http://mwomercs.com/...-epic-dev-fail/

We can all discuss how to improve the MM and what trade offs the player base is willing to accept but we have to at least all understand all of the factors that the MM has to account for. We can't have every MM option we desire and get quality matches.

[Too Much Love]

Russ Bullock, on 21 September 2014 - 10:16 PM, said:

Well as stated it would work better by allowing more matches to be competitive. Please read this post and in particular my respones.

http://mwomercs.com/...-epic-dev-fail/

We can all discuss how to improve the MM and what trade offs the player base is willing to accept but we have to at least all understand all of the factors that the MM has to account for. We can't have every MM option we desire and get quality matches.

IMHO matchmaker should be improved in the following way:
1) It should not only balance ELO nubmers between 2 teams, but also should balance ELO inside one team, so there is no giant gap between players skill in one team.

2) It should not pick one team to win and the other team to lose. Itshould make close matches, not predetermined.

3) Consider make separate ELO for team drops and for solo drops. For example, if I have w/l ratio 3 to 1 in group, it is hell to play solo after that.

Edited by drunkblackstar, 21 September 2014 - 10:42 PM.

[CarnageINC]

I'm guilty of TL;DR...so this in in response to the map voting only, I don't like the idea. You will have large groups that have mechs specifically designed for certain maps and they then can control what map they will always drop on. I like the randomness of maps, it tests your builds and skills in a variety of situations.

[Zyllos]

Russ Bullock, on 21 September 2014 - 09:37 PM, said:

This is actually incorrect.

You can think of it like every mode and map getting at least one vote. So in this situation:

• If Group A gets paired with Group B, you would have forest colony taking about 50% of the chance and all the rest of the maps an equal share so it ends up being a 50% chance of getting Forest colony. It can not be 100% unless all 24 vote only for Forest Colony.

Same logic would apply for game modes so in your next example Skirmish would have the greatest chance perhaps 66%. I need to double confirm all the logic with the engineers ( Karl in particular ).

I understand. Basically, this is a straight up random weighted selection based on votes per tallied votes.

In the Group A vs Group B described above:

• All three modes will have equal chance as there is 1 vote for Skrimish, 1 vote for Assault, and 1 vote for Conquest of the 3 total votes casted (thus, 33% chance on each).
  
• Forest Colony will have 2 votes of the 15 votes casted, while all other maps have 1 vote (thus, ~13.3% for Forest Colony while all other maps have ~6.6% chance of showing up).

Again, assuming that a single vote is placed for the entire group as the group leader makes the selection before launch.

So, for the example of Group C vs Group D:

• 1 vote for Skirmish and Conquest and 2 votes for Assault, thus 50% for Assault and 25% for Skirmish and Conquest.
  
• 1 vote for River City Night and 1 vote for River City, thus 50% chance for both maps.

This is better as the widest selection of maps/modes from a group has more dictation (more votes are casted, so the narrowest selection from a group gets diluted).

Unless, as said above, that all maps/modes have an automatic 1 vote casted for it so all maps/modes have a chance of showing up. Then the above numbers change slightly.

Edited by Zyllos, 22 September 2014 - 06:11 AM.

[EgoSlayer]

What about having game mode selection being a single vote of not wanting to play that mode? Only get one vote of what not to play which has been my experience in solo queue and group queue - it's been two selected modes, and one unselected either Conquest or Skirmish.

Make it a down vote percentage for every not selected game mode, so if 12 Man A votes 'No Conquest' and 12 Man B votes 'No Skirmish' it becomes something like 67% Assault, 16.5% Skirmish, 16.5% Conquest.

And maybe give a C-Bill reward bonus if a team gets put in mode they voted against.

Same thing on maps - make them a 'Not' vote to reduce their percentage rather than raising the percentage of a desired one.

[Zyllos]

After a bit more discussion, I think I am going to change my vote as my original view of the mechanic was incorrect.

[Reno Blade]

drunkblackstar, on 21 September 2014 - 10:42 PM, said:

IMHO matchmaker should be improved in the following way:
1) It should not only balance ELO nubmers between 2 teams, but also should balance ELO inside one team, so there is no giant gap between players skill in one team.

2) It should not pick one team to win and the other team to lose. Itshould make close matches, not predetermined.

3) Consider make separate ELO for team drops and for solo drops. For example, if I have w/l ratio 3 to 1 in group, it is hell to play solo after that.

The problem Russ had described is, even in the perfect match that take all these into account, the MM has to throw away that matching because some of the players did choose to play other game mode and are not eligible for this matching.

1 is already working.

2 is already in place too. MM tries to get the closest matching (as someone described the average ELO difference is around 40).

3 is partly in also, as you get a modified ELO if you are in a group.

But as was said, the MM has to make compromises because of the game mode removing a lot of possible great matchings.

Now if you want all these improvements, the next step would be to make the game mode an internal voting system, as Russ described that gives the MM a chance to "ignore" this big stop sign and create all these very close matched groups.

[Satan n stuff]

EgoSlayer, on 22 September 2014 - 07:47 AM, said:

What about having game mode selection being a single vote of not wanting to play that mode? Only get one vote of what not to play which has been my experience in solo queue and group queue - it's been two selected modes, and one unselected either Conquest or Skirmish.

Make it a down vote percentage for every not selected game mode, so if 12 Man A votes 'No Conquest' and 12 Man B votes 'No Skirmish' it becomes something like 67% Assault, 16.5% Skirmish, 16.5% Conquest.

And maybe give a C-Bill reward bonus if a team gets put in mode they voted against.

Same thing on maps - make them a 'Not' vote to reduce their percentage rather than raising the percentage of a desired one.

I'd go so far as to having players choose between voting against one or more maps, or voting against a game mode, as the large number of players that hate specific maps, most notably Terra Therma, might cause those maps to be removed from map rotation altogether even though there are also a lot of players that like those maps.

Some kind of point system that lets you vote against a specific percentage of the options might be useful because as the number of options increases the number of undesirable options will also increase.
For example with a point system you could vote against one game mode or several maps. The point value of each option would depend on the number of options available, more options would mean fewer points spent per vote.
Say a player doesn't like large maps, they can choose to vote against all the largest maps, but voting against any map would make that player unable to vote against a game mode, or any other game setting that might be put to a vote.
You don't like [insert map here]? Well you'll just have to put up with [insert game mode here] if you want to vote against it.

[Too Much Love]

Reno Blade, on 28 September 2014 - 01:31 AM, said:

1 is already working.

Proof link plz

[Reno Blade]

drunkblackstar, on 28 September 2014 - 08:09 AM, said:

Proof link plz[/size]

This is what I found:
http://mwomercs.com/...69#entry3696769

Karl Berg, on 06 September 2014 - 06:08 PM, said:

Quote

That's what I was trying to get at. So it does effectively alternate adds to each team for new population. In balancing Elo does it shoot for an initial 'range' target or does it grab a 'bundle' of high/low to match target for the match?

So if I get this right it grabs a couple of the oldest waiting players in the queue and they effectively set the Elo target for the match as well as pre-seed themselves in the 4x3 matchup. Then it looks through the whole queue for fits based on criteria of exact tonnage match > weight class and matching Elo > high/low to target?

For all the talk of nobody taking Lights I'll get matches with 4 lights on each side. Does it prioritize weight class over Elo or vice versa?

Yes, it effectively alternates adding to each team, as long as the size of the groups being added are roughly equivalent. In the case where an 8 man is added to team 1, and a 5 to team 2, then the next team to receive a group would be team 2 since it is still the smaller.

And now this is exactly where it gets complicated.. Yes, the average Elo of the seed group is saved off. This become the baseline skill level around which the entire game is built up on. There are also obviously multiple competing constraints that the matchmaker is attempting to simultaneously satisfy, including group size, Elo, weight class, game modes, region, and factions when we enable that feature. Most of those constraints require some form of tolerance decay, or release valve as I describe them to design. The decay functions are exponentials, with constant start and stop inputs, that vary with respect to time (age). That evaluation of a given decay function is quantized onto an output range, and these are fed into a search algorithm that locates potential matches and returns the most desirable. All of these parameters are tunable on the fly from our matchmaking command center.

This somewhat analog approach to balancing simultaneous constraints seems to work quite well in practice. We can easily make small tweaks to inputs to subtly change our desired biases in match outputs. There are a few constraints that always take priority. These are what we refer to as hard-constraints, as opposed to the softer constraints like Elo, or the number of heavy mechs in a match. Those hard constraints are currently team size, balanced weight class matching (although this switches off after 4 minutes searching in group queue), region, and game mode.

Quote

Is it pattern generated, as in when I finish a Terra Therma match another TT match starts or just scrolling down a menu? As in if I finish that TT match and immediately drop another am I likely to get caught in that same new TT match or is this just random chance? I find that if I drop again immediately after the last I'm more likely to catch the same map or is that just anomalous?

It's a random dice roll, using the c standard random function, seeded with system time, every time a server boots. There should be no magical stride that sees you getting placed on the same server on the same map repeatedly. There have been times where I could swear there was some selection bias in the system myself though. One of the engineers went so far as to write up a quick simulation of distributed RNG's like we're using just to see if there was a loss of entropy when compared to a single random number generator. We found that the distribution using a distributed system was less random. Not terribly surprising I guess. Anyways, all of this has reminded me to go back and change this so that it uses a centralized RNG; so thanks!