78 replies to this topic

[Craig Steele]

So my MWO experience lately has got me wondering,

Why is that every 3/4 games matchmaking pushes the game forward with only 22 or 23 of the 24 spots filled in the roster? Is this because the game population is insufficient to support full games?

Why is it that many games seem to have lopsided outcomes, 12 - 2 / 12 - 4. So many stomps. Is the Matchmaking forcing ELO mismatches to get games to load?

It can't be all me as I am on the winning side often enough, but it does seem to be that we win easily or loose horribly.


Anyone know?

[Texas Merc]

Its a numbers game. Once your team is down (or up) by 2-3 mechs things tend to snowball.

[Reitrix]

Funny, my recent experiences have been mostly close games. Im talking winning with only 2 'mechs standing, or losing with the at 8 - 12.
Maybe I've just had a run of some damned fine luck, since I'm not running into as many Metabuild Alphapoptarts either.

Hell, i had a brilliant 1 - 5 on a Conquest where my Boars Head was the last 'mech against a DDC, ShadowHawk, Spider and 2 Hunchbacks.
Brawled the DDC to death, disarmed the Spider and ruined the SHawk before the Hunchies pulled up behind me and tore me a new one.
Been having a ton more fun with the game lately. Even if 12 v 11 games. A lot of times someone from the 12 side will disconnect to make it a fair match.

[Sephlock]

Matchmaking.... would require math to explain.

As for landslide matches, they tend to snowball.

All it takes is one decent group of players working together (or a bunch of puggers randomly acting in concert.. say... by all rushing the same spots)...

And that's not all. Due in part to the semi-random spawning, sometimes one lance will go off on it's own and get pwnt in a 12 v 4 battle, and then it's 12 v 8...

[Limidas]

I regularly drop with freinds. About 10% of the time one of us is disconected or crashing to desktop. When this happens during the load screen that persons "slots" stays blank and no new player seems to be abel to use it and then the team is one down.

[Noesis]

No idea on population numbers.

Possibly domino effect as to why some games end up with lop sided conclusions?

[Craig Steele]

Thanks guys for the responses.

It seems a little underwhelming if this appears to be the "norm". I get the snowball thing, but a lot of games I see one or 2 players doing 800+ dmg on the winning side compared to loosing side with several less than 100. Kudo's to the skilled guys pumping out the dmg, but it's not exactly encouraging for the others on both sides. My question is why is matchmaking throwing them in together?

I also see games where tonnages are heavily lopsided. Today I played against 3 Atlas's, 2 Highlanders, a Stalker and an Awesome and we had not a single Assault in the company (of 11). Theirs could not have been pre made as different lances etc. But even if it was, why would matchmaking not put one of the three groups in the other company seeing as we are all of the "same" ELO. Ofc tonnage is far from the be all and end all, but it is a pretty big factor when its 7 v 0.

Should not the Dev's would be concerned about something detrimental to long term success (ie, a regular lack lustre gaming experience).

Well, my semi "vent" is done. Back to another lottery stomp.

[Kyle Reece]

As I understand it, the matchmaker always puts together two full teams to face off to each other. Empty slots are where people have crashed, disconnected, or timed out in joining. Anecdotal evidence from matches would seem to support that.

As far as steam-rolls. There's always going to be a team that's predicted to win - it's how the MM does it. If that team also has force multipliers such as 4 man's with voice comms, cheese builds, low numbers of newbies etc it's going to be an even bigger advantage to them.

On why there's often someone scoring huge amounts of damage... Don't forget that may not be a particularly skilled player, just someone using something cheesy. For how matchmaker is described to work you can check out these Command Chair posts;
http://mwomercs.com/...79-matchmaking/
http://mwomercs.com/...-making-update/
http://mwomercs.com/...ted-april-19th/

In a nutshell, matchmaker picks an Elo number. Fills out two teams so that the average Elo of both is a close to that - trying to keep the individual tonnages and Elo scores as close as possible, but opening those restrictions out the longer it takes to get 24 people - then drops them into a "random" map.

I'd suggest that the opening out of the restrictions is where the weight imbalances and skill imbalances can creep in when available player counts are low. I've also seen it suggested that groups get the average of their group added to the team total to average out, rather than their individual Elo scores which won't be helping.

In the game one team is predicted to win, if they do their Elo goes up a little (based on the difference between the two team's Elos), if they don't it goes down a little. One team is predicted to lose, if they win their Elo goes up a lot.

[Tuann]

The last two days there are in my experience bizare weight mismatches,...
I can honestly say that if i drop in a team and our team has like 2 assaults (awesome and battlemaster) and the other side has 3DDC's, an Rs atlas, highlander and Stalker,... That that match did not go well,... And it is happening quite a lot.
Elo matchmaking is going haywire.

Quite some people in our clan experience the same. Its bizar.

We did have a fair amount of great, close, games too,... But before, mismatches were really not that obvious and certainly not so common as the last days.

Just my 2 cents..

[Monkey Lover]

Bad programming there is zero reason for a game to start without all 12 people on a team. Heck whats the complete worst is when the server knows for 100% there is not a person and says disconnected.

[Craig Steele]

Kyle, your post was very informative. Thanks.

If I understand you correctly, every time I play a match the game has pre determined the result and slanted my team mates accordingly (ie, we are all due for a win / loss and pitched against opposition to effect that)? If I do "buck the odds" one way or the other, it uses that information to determine the next match outcome.

Why am I playing?

If I am a skilled player the system loads against me until I fail. If I am a poor player the system gives me a free hit until my stats look good and then the circle starts again, it loads against me.

Whats the reward for the practice and effort I put into getting better. I can't even point to my KDA or W/L as these reflect what the game has pre determined (barring the "buck the odds" either way)

Enlightening to say the least, and certainly explains why matchmaking sets up some ridiculous battles.

I always assumed matchmaking was along the lines of, "Here's player A with a 55% W/L ratio and avg dmg 350, and here's 23 other players +/- 10% of those stats, lets throw them together in 2 sides with more or less even tonnages and let their skill and tactics determine the outcome of the battle. Oh, and if they have been waiting 60 seconds for a game, lets go to + / - 15%, 2 minutes???, + / - 20%"

But I read the matchmaking system now as "Oh you have lost a few lately, here's some food for a few games to boost your stats with some wins or, oh you won a few lately, we'll set you up as food for some other guys to bring you down a bit"

That's pretty deflating actually.....

[Charons Little Helper]

Texas Merc, on 04 January 2014 - 12:42 AM, said:

Its a numbers game. Once your team is down (or up) by 2-3 mechs things tend to snowball.

That's the main reason I prefer conquest. Instead of the first contact being one big battle - usually there are several smaller battles - each side likely winning at least one due to local superiority.

[DaZur]

Craig Steele, on 04 January 2014 - 04:27 AM, said:

Kyle, your post was very informative. Thanks.

If I understand you correctly, every time I play a match the game has pre determined the result and slanted my team mates accordingly (ie, we are all due for a win / loss and pitched against opposition to effect that)? If I do "buck the odds" one way or the other, it uses that information to determine the next match outcome.

Why am I playing?

If I am a skilled player the system loads against me until I fail. If I am a poor player the system gives me a free hit until my stats look good and then the circle starts again, it loads against me.

Whats the reward for the practice and effort I put into getting better. I can't even point to my KDA or W/L as these reflect what the game has pre determined (barring the "buck the odds" either way)

Enlightening to say the least, and certainly explains why matchmaking sets up some ridiculous battles.

I always assumed matchmaking was along the lines of, "Here's player A with a 55% W/L ratio and avg dmg 350, and here's 23 other players +/- 10% of those stats, lets throw them together in 2 sides with more or less even tonnages and let their skill and tactics determine the outcome of the battle. Oh, and if they have been waiting 60 seconds for a game, lets go to + / - 15%, 2 minutes???, + / - 20%"

But I read the matchmaking system now as "Oh you have lost a few lately, here's some food for a few games to boost your stats with some wins or, oh you won a few lately, we'll set you up as food for some other guys to bring you down a bit"

That's pretty deflating actually.....

I'm not finding any logic in your conclusion... If a match-makers formulary is not the fundamental effort to create as balanced a pairing what should it be?

How should a system which weighted to one side create balance without either adding to it's offset or removing it from the imbalanced side?

Elo ultimate goal is: 50/50 with a winner...

Edited by DaZur, 04 January 2014 - 06:51 AM.

[Joseph Mallan]

Craig Steele, on 04 January 2014 - 12:39 AM, said:

So my MWO experience lately has got me wondering,

Why is that every 3/4 games matchmaking pushes the game forward with only 22 or 23 of the 24 spots filled in the roster? Is this because the game population is insufficient to support full games?

Why is it that many games seem to have lopsided outcomes, 12 - 2 / 12 - 4. So many stomps. Is the Matchmaking forcing ELO mismatches to get games to load?

It can't be all me as I am on the winning side often enough, but it does seem to be that we win easily or loose horribly.


Anyone know?

Cause fighting isn't always fair and balanced???

[Craig Steele]

Joseph, the OP was made in Gameplay Balance. Are you saying war is not fair so the game shouldn't be?

This is a game for players entertainment though. People even pay for the entertainment offered by the game, you and I included.


I am still playing the game (had a really nice little one on one duel against a catapult in my LCT-1V) but I would like to increase my understanding of some things in the game (hence my original questions) and if appropriate, I would like to see the game improved.

Given your number of posts and tenure in the game you must have a strong understanding of the community, could you give me some advice as to where I should ask questions about game balance from other players and submit feedback on my experiences.

(we lost the LCT-1V game btw, 12 - 2 stomp, I just couldn't kill another 10 of them after I got the Pult)

Thanks in advance.

[Joseph Mallan]

Craig Steele, on 04 January 2014 - 07:09 AM, said:

Joseph, the OP was made in Gameplay Balance. Are you saying war is not fair so the game shouldn't be?

This is a game for players entertainment though. People even pay for the entertainment offered by the game, you and I included.


I am still playing the game (had a really nice little one on one duel against a catapult in my LCT-1V) but I would like to increase my understanding of some things in the game (hence my original questions) and if appropriate, I would like to see the game improved.

Given your number of posts and tenure in the game you must have a strong understanding of the community, could you give me some advice as to where I should ask questions about game balance from other players and submit feedback on my experiences.

(we lost the LCT-1V game btw, 12 - 2 stomp, I just couldn't kill another 10 of them after I got the Pult)

Thanks in advance.

You mean I cannot be entertained watching my side get mowed down by a Suprior Opponent? You saying you cannot be satisfied that you got 2 kills in an otherwise lopsided battle? I was overjoyed that I was able to get 4 kills vs a team that killed every other member of my team before I fell. If you can fulfill your 8% towards a victory or better you should be able to leave a loss with some dignity and pride.

I keep reading how competetive some players think they are, but forget that part of competing is losing with dignity, and continuing to try to improve even when the Mountain keeps growing!

Edited by Joseph Mallan, 04 January 2014 - 07:44 AM.

[Felbombling]

Neglect and lack of forethought may be rearing their ugly heads at this stage. The recent matchmaker screen shots and discussion have me a little worried, and I'm wondering if the situation hasn't reached a critical stage now. A major imbalance happens and the devs ignore it for months, such as the PPC fiasco. No tutorials for new players to learn the ropes. Pitting organized groups on VoIP against PuG players. Lack of information in the Mechlab for things like ghost heat or LRM minimum range limitations. No game lobby at all. The introduction of ECM. Ugh.

I begin to wonder if PGI/IGP imagine some endless pool of players who will eventually come back or try out the game. Things that seem to be no brainers are constantly pushed back behind the magical moving goal posts of UI 2.0 or community warfare... especially if a new Mech package is available for sale. I'm not sure how much slack you need to give a small development team for being small, but the goodwill has to end at some point, doesn't it?

[Allen Ward]

I doubt that there is a lack of players at any time given to fill up games. Disconnects happen (client crashes, memory allocation errors) so sometimes people drop out of a game being set up. But: a match takes about 2-5 minutes to build up - and PGI is not able to make it with 24 players at about 30% of the time? Or find players/mechs that roughly compete?

If I have a high kill rate and low death rate but have lost most matches (due to lack of skill of the rest of the team or similar) and I choose a Trial Mechs, where do I end up? In a game full of rookies or on a game with experienced players running mastered mechs? Or random? The matchmaker just can't handle this correctly.

Every 3rd or 4th game one side is heavily overpowered (by tonnage/weight class) - the lighter team still has a chance of winning, but it is really hard to win against superior firepower (although a heavy mech doesn't mean lots of firepower automatically). So actually mechs should be judged by a Battle Value and Tonnage while playes should be judged by ELO ranking (not only based on kill/death or win/loss ratio, but also on overall hit accuracy, ingame endurance and probably more stuff). These three together should be used by the matchmaker.

If the matchmaking sequence would be more transparent and flexible, it could still take several minutes to fill in all slots but the matchmaker could make sure that a game starts with 24 mechs and not less. All other MMO games/FPS can do it, only MWO fails here. The problems with disconnects after game start would be enough to deal with.

[3endless8oogie]

Funny to see that some guys still dont get the difference between reality and games

Edited by 3endless8oogie, 04 January 2014 - 09:41 AM.

[Kyle Reece]

Craig Steele, on 04 January 2014 - 04:27 AM, said:

If I understand you correctly, every time I play a match the game has pre determined the result and slanted my team mates accordingly (ie, we are all due for a win / loss and pitched against opposition to effect that)? If I do "buck the odds" one way or the other, it uses that information to determine the next match outcome.

Not quite. The MM doesn't determine the outcome, it only tries to predict it based on team average.

From the descriptions given, it also doesn't care about your win/loss rate. Evening out to a 50/50 win/loss rate is a result of the MM doing it's job, but it doesn't take that into consideration for picking which people you'll be playing with and as long as you win the right games your Elo can be high even if your W/L is terrible.

Basically, the term "Matchmaker lottery" is exactly the right name for it. Every time you click launch you could be grouped with any combination of people, in any combination of 'mechs, and either be on the high or low Elo side which may or may not make a difference to if you're fated to win or lose that match. The MM tries to take some of the random out, and avoid some of the unbalanced possibilities, but - at least from my own experience - still struggles for whatever reason.