259 replies to this topic

[Xtrekker]

Admittedly with just skimming over these pages I may have missed something. But, ...

If MM insists on keeping you at a 1.0 W/L ratio, when you get a win you can expect that MM is going to try to serve a loss. My assumption is that this is true regardless of one's performance, so if other members of your team carried you to victory, you are then penalized in the next match as well by potentially facing an even harder team. So finding "better matched games" is kind of moot because the system is flawed to begin with. Obviously very lopsided team games aren't fun, especially when the other team is a poor victor.

Example, just running my Atlas-D(F) this past weekend over the first 10 matches solo:

Game #	1	2	3	4	5	6	7	8	9	10
Win/Loss	W	W	L	L	L	W	W	W	W	L
Kills	3	1	2	0	0	2	2	0	0	0
Death	0	0	1	1	1	0	0	1	1	1

You can see that doing well in the beginning resulted in a facepunch by the MM, forcing losses. At game 6, MM kicks in and decides I should do better. Oh wait, you did better, so another facepunch at game 8. However you were winning, so we're going to continue to make this harder and harder for 9 and then 10.

I know there has to be more to it, but this was just the obvious result from the numbers I was recording. And it does suck to know that every victory means you go into the next match with the server trying to make you lose. In fact if I pull off a really good game, I'm a little jaded at what I can now expect. Also, if I do really well in a game, there's that thing in your head as to whether or not you really pulled off a good game or did the MM just give you some scrubs to eat.

That said...

If MM is going to be core, then yes, we need ways to customize our experience. So, I guess yes?

It's hard to understand what is happening when we don't know how many matches and players are currently active and online. It's all conspiracy and conjecture. I usually figure it out when all of a sudden I'm facing certain higher-profile competitive players multiple games at a time -- that there aren't enough people playing and I'm in an Elo bracket that isn't my normal bracket. Or this ROFLtromp is my penance for just having had a good game.

Honestly I originally thought this matchmaker system we had was just a band-aid until the real system came as part of CW, so I'm a bit disappointed to presume this is the core system for the foreseeable future.

Edited by Xtrekker, 22 September 2014 - 12:42 PM.

[wwiiogre]

Matchmaker, where to begin?

Ditch ELO all together it is nothing more than a here is your participation ribbon program. Designed to balance win/loss not the match.

ELO in multiplayer and with different mechs, equipment, modules cannot and will never be able to judge skill. ELO is a single player with no variable equipment/gear program to match skilled players such as in chess.

What MWO does with ELO is wrong, and almost always makes unbalanced matches. I have posted this in so many places so many times but here it goes again.

Ditch ELO, end of story. Instead a matchmaker can never be based on skill in a multiplayer team game.

Match tonnage class to tonnage class. Light to Light, Medium to Medium, etc.

Match IS to IS and Clan to Clan.

Match ECM to ECM.

Player skill cannot ever be judged, so instead use number of drops.

New Player: Less than 50 drops
Rookie: 50 to 250 drops
Veteran: 251-1000 drops
Elite: 1001+ drops

Now this does not measure skill, what it does measure is knowledge of the game. Any one with more than a 1000 drops who does not know the maps, the weapons, etc is the biggest farmer ever. More than likely they are a decent player with good game knowledge.

Now the hard part. Stock mech or Uber Mech. A 80ton Awesome stock mech does not equal a Dire Wolf that has been modified and this should never be the same in any matchmaker. Currently MWO says these are the same. They are not.

Modules, yet another hard part. A mech without modules does not equal a mech with modules. They are not the same, they should not be the same.

Last: Arty/Air please change these, as implemented by PGI they are the equivalent of Regimental or even Division level power. This is a company sized game. Arty should only ever hit legs, just as in table top. If arty only hit legs, you would need less powerful arty to be scary. Air is the same way, to powerful. To many players with arty or air in every single game. Its spamalicious but most players would probably toggle a choice to play without arty/air.

So please have arty/air use the same as game choice and map choice. Put a toggle vote in matchmaker and watch players use it or not. I would never play in a game with arty or air in it if I had the choice at least as implemented by PGI currently. Its still over powered and under priced.

So that is my take on Matchmaker. It currently sucks, most pug solo drops are complete stomps (win by 6 or more mechs) or roflstomps (win by 8 or more mechs, 2 lances). This is unacceptable. I average more than 80% of my pug solo drops as the above. And way to many are 12-0. Because of every reason I listed above.

Chris

[Tolkien]

I like the direction this is going - part of the problem with the black box matchmaker is that it doesn't give the player a feeling of ownership of the outcome. The sentiment 'the matchmaker screwed us' is common.

Previous games such as MPBT3025 let the two sides see each other's composition in the lobby and you could adjust that until both sides were satisfied that the match up was 'balanced'.

TL:DR; It goes a long way to accepting the match outcome if you have control over the match setup. Hopefully this voting on where to go is a first step to that.

[Xtrekker]

The loss of one or two mechs (disco, AFK, suicide, or just gung-ho loners) early in the match seems to affect the match outcome of this game much more than previous versions of Mechwarrior, even after expanding out to 24-player drops. Maybe it's the influx of the twitch crowd over the past decade making it a more fast-paced game. One side gets overrun very quickly and usually can't recover from the loss of a flank lance. We seemed to be much more patient and calculating before. Of course in league play there was a consequence to random stupidity, so it was kept to a minimum. Who knows. My guess is maybe radar? Active/passive pinging was an awesome little information minigame, and you seldom got hit without knowing it was coming. I think if there was a little more situational awareness in MWO the matches would be more competitive by design.

[DerelictTomcat]

I play almost 100% pug matches in MWO. I have zero interest in playing anything but skirmish since its introduction. I sure as hell do not want to end up in a conquest match piloting a Daiashi.

[Gooner]

For multiplayer games in general, no. I believe the players should have as many options as possible to choose what game type, what maps, etc. Every other multiplayer game I've played for a long time (UT series, L4D, TF2, Killing Floor) has a much more functional server browser or lobby system than MWO so I want to see more options, not less.

However, as MWO currently exists, I don't think the change outlined in the OP will have such a negative effect - because all 3 game types are basically the same. Deathmatch, Deathmatch with a base that takes forever to capture so just focus on killing everyone, Domination that many times is won by the team that focuses on kills not points.

I vote to change it now, but would favour changing back when (if?) more varied gametypes arrive with CW.

[Mechsniper]

ELO has been a thorn in the butt of all the experienced players for some time. You can never achieve excellence in this game if you are forced to lose by matchmaker by intentionally stacking your team with terrible or new players. Choice of game mode dictates the mech you will want to take therefore making me drop for assault in a mech I wanted for capturing resources (IE a lighter assault mech made to kill lights vs a brawler atlas). This makes for completely unenjoyable games which I would just as soon alt+f4 out of, and that screws the other players out of a fair match. This is exactly what people did in the past before game mode choice, and why it should NEVER return. Please trash elo and machmake by at least a minimal total player exp earned cutoff to weed out new players from more experienced players. This is the only fair way you have that I can see to easily do it. Sorry ,but newbies should not be forced to drop with 3 year players, nor vice versa. ELO MUST GO!

[Mawai]

Xtrekker, on 22 September 2014 - 12:42 PM, said:

Admittedly with just skimming over these pages I may have missed something. But, ...

If MM insists on keeping you at a 1.0 W/L ratio, when you get a win you can expect that MM is going to try to serve a loss. My assumption is that this is true regardless of one's performance, so if other members of your team carried you to victory, you are then penalized in the next match as well by potentially facing an even harder team. So finding "better matched games" is kind of moot because the system is flawed to begin with. Obviously very lopsided team games aren't fun, especially when the other team is a poor victor.

...

I am sure there are lots of folks who can correct your misunderstanding.

The matchmaker does NOT insist on keeping you at a 1.0 W/L rate. NOT at all. In NO way shape or form.

What does the matchmaker do?

It TRYS (not always successfully) to put you in matches with more or less equally skilled team mates and opponents.

What does this mean? This means that your chances of winning or losing the match are about 50%. It is up to YOU (and your team) to actually win or lose. The matchmaker doesn't dictate anything. It doesn't give you a harder match on the next match after a win or an easier one after a loss. Your rating may or may not change after a match depending on what the expected outcome was and whether you fulfilled it or not. If you are expected to win or lose and you do ... your rating doesn't change. If you don't meet the expectations ... your rating changes and the maximum change in one match is 20 points ... I have no idea what the median change is but it is probably far less than this.

Furthermore, if you get better then you will tend to win more matches and your rating goes up until you are matched with players of comparable skill ... at which point you will find that you win about half.

I don't know if they have separate group and solo ratings now but it used to be that playing in a group would tend to drive up your rating due to the team work giving you wins ... then playing in the solo queue these folks would be handed a bunch of losses as their rating dropped since it was then based on individual and not group performance.

But the bottom line is that the matchmaker dictates nothing except the teams and map ... winning or losing is up to the players.

[Xtrekker]

Well, like I said, conspiracy and conjecture. Without knowing specifically what's at play in the system it's hard to get a grasp on what is actually happening. Perhaps the trends I see are just coincidence.

[Mcguire]

I'm always up for more competitive games and would gladly give up the choice of local! On a side note let me add Thanks russ for getting more involved as of late, it means the world to me

[Jetfire]

Russ Bullock, on 21 September 2014 - 10:34 PM, said:

Thanks for your question I should make it clear that in your example it isn't that Crimson would be chosen, it would be that Crimson had about a 33% chance of being chosen. Also we would likely give each map at least a vote of 1, therefore everything would have at least a very small chance of being chosen with the % increasing with the number of votes.

I like this idea a lot. Otherwise some maps would never get played and as much as I hate fighting on river city night... I want there to be maps I dislike playing on. You don't always get to pick your battles for nice sunny days in the park afterall.

Edited by Jetfire, 22 September 2014 - 02:21 PM.

[Antonio]

I chose not to vote because I think the two concepts are not mutually exclusive. The reason why people deselect modes is the currently offered modes each promote specific playstyles. I think for random non-CW matches there should be a single mode in which a variety of playstyles are viable.

[EgoSlayer]

Mawai, on 22 September 2014 - 08:50 AM, said:

One quick comment ...

Before going to the effort of implementing this ... can you verify the numbers that removing this constraint will actually make a substantive difference in the quality of matches formed overall? Run stored player data through the matchmaker removing the game mode constraint and see how many more large group vs. large group matches are formed and how the difference in Elo between teams changes and whether the difference is significant.

The basic idea sounds ok but if it can't be backed up with numbers it doesn't really mean anything. This change might or might not be worth the effort involved.

<snip>

I believe they can do this now. It was mentioned in Dev Vlog 6, maybe 7, that they can feed data into a match maker sim to see the results in the queues by changing settings.

EDIT - It's Dev vlog 5. At the ~3:18 mark
http://youtu.be/q0xBX_QEfpw?t=3m18s

Edited by EgoSlayer, 22 September 2014 - 02:40 PM.

[Archon]

I prefer being able to choose game mode.

[lazypc]

Although I seem to be the minority atm I would vote no for both.

The first because I really don't feel that conquest should have a large place in the solo queue, and I would personally choose to play more difficult matches and get better than to play conquest and feel like a lone merc in the dysfunctional family of other lone mercs.

The map voting I would say no to as well, it's already been mentioned a couple times but it becomes too easy for groups to load out fro certain maps, creating large advantages the MM could never compensate for.

[notjoe]

what a **** idea PGI. Why not just make a better match making system instead of this.

[jozkhan]

Surely 'giving up the ability to guarantee the game mode you play' is a developmental step backward?

[Viges]

I hate conquest so much that I chose "no". Maybe if you change it or at least make more game modes...

[Zordicron]

Russ Bullock, on 21 September 2014 - 10:34 PM, said:

Thanks for your question I should make it clear that in your example it isn't that Crimson would be chosen, it would be that Crimson had about a 33% chance of being chosen. Also we would likely give each map at least a vote of 1, therefore everything would have at least a very small chance of being chosen with the % increasing with the number of votes.

KK, here is how I would do it(map voting)

Randomly select two maps. Give players option to vote one or the other. This does two impotant things:

First, it negates the idea of "loading out for specific map meta" completely. Laser boat has a reduced chance to end up on a hot map, but what if the choice is terra therma or caustic? Still need to prepare for hot maps, BUT at least players have some choice.

Second, it gives some choice to players without creating a scenario where people like me, that like lots of variety in the map choices, end up getting tuck with whatever the majority wants over and over. Lets say whatever particular ELO bracket tends to favor one or maybe two maps, and constantly gets majority votes on them. Now, you can simply have the votes "influence" the selection algorithm, BUT if say, a mojority of players pick a map, and then dont get it, I foresee lots of "but but why?!! whaa!!!" and angry stomping about how stupid the matchmaker and map voting are blah blah etc. If we make the vote simply decide which map gets picked, period- you could have a scenario where players end up in the same maps over and over because majority rules, which might sound ok on the surface but you dont want to have negative play experiance for anyone, or at least as verty few as possible.

HENCE, if we make a random selection of two maps, one on left side of screen, one on right, players simply pick the picture of map they want and votes tally underneath, you can get some middle ground. Players have input into map choice, but not total decision on it either, leaving plenty of room for randomness and the effect that has on loadouts.

One of the Need for Speed games my boy has has this type of random/vote hybrid for it's online play for race course selection, and it works really well, players vote based on which course they like, but also based on which car they have at the time too, which ended up (as i saw in watching him play) that there was never any course "left out" because the group of players would have different tastes, and also different car stats to alter the decision process. Similar to this game, mech stats and player tastes.

EDIT: I ALMOST FORGOT THIS TOO: if you do a two map random with votes type like I mention, you can also TAG THEM WITH A GAME MODE. For instance:

Map A and B are randomly selected, lets say A is Caustic and B is Crimson. Players choose A or B to vote on which one they like to play. If we tag A and B with a game mode though, you add a new element into the mix. lets say A is Caustic conquest, and B is Crimson Skirmish. Suddenly, players arent looking just at the map, but also game mode, and now the choice isnt even just loadout, but also mech choice. This also fills a need in the game mode selection change you propose, as now players dont have a complete say in what mode it is anymore, but at least they can vote on a mode if it is more important to them then the map choice. It also keeps quite a bit of the randomness involved in the game, while still giving players that feeling of input into where/what they are doing.


As for the game mode, I stopped playing conquest because I 100% pug and pug conquest is herding cats multiplied by IT'S OVER 9000!!! However, I would be willing to play said mode again if you think it would result in better ELO matching. At least temporarily as a trial. I would leave the option to revert open, if I were you, in case the idea doesnt pan out and players just want to go back to selectable modes. EDIT: see above edit, i really think that is the best middle ground option you could do. Random maps with a choice of two to vote on, combined with random game mode asigned to each map as well.

Edited by Eldagore, 22 September 2014 - 03:37 PM.

[CG Chicken Kn]

I already leave all game modes selected, since making me spawn in different places with different objectives makes me a better player, in my opinion. So that part would not affect me.

The part about it meaning the ELO section and other balancing factors of MM would work better, would definitely affect me, and everyone else, in a good way.

So..... DO IT! Do it NAOW!........ Or whenever it's handy.