58 replies to this topic

[Aresye]

Played 13 games tonight. I lived for 2 of them, and the funny thing is, I didn't even really notice that much of a difference in LRM play, nor was bothered by massive amounts of LRM spam.

But I am absolutely sick of this game. Not because of LRMs. Not because of meta (although that does play a small part in it). No, I'm sick of being continuously thrown into roflstomp after roflstomp, after roflstomp, after roflstomp, after roflstomp.

I experience a grand total of 2 gameplay experiences. I either get roflstomped, or we just barely manage to squeak a win by a very close game.

What gives?

I seem to always get a good blend of teammates that decide the best time to test their weapons is into my back armor before I even move. Meanwhile on the other team all I see is an incredibly competent group of players that on multiple occasions felt like I was dropping against a 12-man premade. Coordinated attacks from multiple angles, lights leading teammates away to get slaughtered by the main force, etc.

I don't consider myself an expert player. My average damage (win or lose) is usually around 300-600 each game, which I would put somewhere in the average to above average group when encompassing the overall spectrum of players, with experts getting 800+, and average players around 200-300.

Just to be clear, this isn't a rant. I'm more just puzzled as to why the whole match balancing system is designed the way it is.

What fun is there knowing that your match is predetermined before you even start it? Why do I even play a game knowing that there's some obscure system that is constantly pitting me against groups that are WAY above my skill level simply for the sake of balancing my stats out? I honestly don't know, and while this has bugged me for awhile, it wasn't until this stats reset and patch that I'm really starting to see Elo for what it is.

It seems more like a system that continuously punishes better players to no end, while simultaneously subjecting many newer, less skillful players to a horrible experience. Why is that, when the entire reasoning behind the system in the first place is to avoid newer players from continuously getting roflstomped?

It's being unfair to all parties, but even more unfair to the parties the system was designed to help. It isn't fair for me to complain about my teammates, because most of them don't really know any better, but does that make it fair to subject an entire group of newer players to an absolutely horrible experience for the sake of balancing a single person or small group of people?

I don't know what games are coming to nowadays. I just don't feel a draw to my playing experience being handled like an excerpt straight from 1984. There's only been 1 big issue I've ever seen that's important to game balance, and that's cheat protection. Beyond that, it's drop into a game, you play your best, and you either win or lose. No stat-based pre-determined matches, no ghost heat, no condescending developers treating you like a 5y/o girl watching a Barbie commercial, etc.

My overall experience of MWO has been based solely on how much I can tolerate, and it's getting thinner every single day. I'm usually a staunch defender of video games being an enjoyable pastime when people say things like, "They're a waste of time," but when nearly every single aspect of your gaming experience gets taken farther and farther out of your control, you reach a point that you realize you're nothing more than data itself. When you cease being a player, and instead are a script designed to operate a program with a singular, linear purpose.

That, in my opinion, is the very definition of a waste of time.

[Willard Phule]

^ This. Absolutely.

You know what's funny? Everyone's sitting around, waiting for this Launch Module like it's going to fix the roflstomp problem in one fell swoop. Based on PGI's record of adding new content...odds are good that it'll be half-finished, a "work in progress" and will ultimately break more than it fixes. Could I be wrong, sure, but again...based on their track record...what are the odds I'm right?

That being said...you want to know a quick, easy "tweak" that could be done to the MatchMaker that might help even out the skill levels?

Get rid of this whole "Prediction" thing that it's got going on. The only way your Elo rating changes is if you "beat the odds"....lose when you're predicted to win or win when you're predicted to lose.

Just scrap the prediction thing. Put two teams together of roughly the same Elo and let the chips fall where they may. If you win, it goes up....if you lose, it goes down. It speeds up the separation process...allows the cream to float to the top and the manure to sink to the bottom. Is it perfect? No...but it's a sight better than what we've got right now.

[Veranova]

Willard Phule, on 19 March 2014 - 03:32 AM, said:

^ This. Absolutely.

You know what's funny? Everyone's sitting around, waiting for this Launch Module like it's going to fix the roflstomp problem in one fell swoop. Based on PGI's record of adding new content...odds are good that it'll be half-finished, a "work in progress" and will ultimately break more than it fixes. Could I be wrong, sure, but again...based on their track record...what are the odds I'm right?

That being said...you want to know a quick, easy "tweak" that could be done to the MatchMaker that might help even out the skill levels?

Get rid of this whole "Prediction" thing that it's got going on. The only way your Elo rating changes is if you "beat the odds"....lose when you're predicted to win or win when you're predicted to lose.

Just scrap the prediction thing. Put two teams together of roughly the same Elo and let the chips fall where they may. If you win, it goes up....if you lose, it goes down. It speeds up the separation process...allows the cream to float to the top and the manure to sink to the bottom. Is it perfect? No...but it's a sight better than what we've got right now.

The "prediction" is simply based on ELO between teams.

It doesn't intentionally weight one team higher in ELO, it's just the nature of the system that there may be some small variance.
Rewarding players for beating the odds when there IS a small variance makes sense therefore.
And rewarding a player LESS when they had the advantage does too, because both of these factors help organise the ELO spread in to the bell curve.

It means you can't get in to the high ELO's easily, by accidentally rofl-stomping worse teams. But you can if you beat your equals or betters.

Edited by Veranova, 19 March 2014 - 03:41 AM.

[Svidro]

Seems like a lot of the time the matchmaker is starting off making a team for a particular ELO bracket... and then can't find anyone to fill the other team due to lack of players, and eventually just sticks anyone it can on the other team. And poof, stomp.

[SweetJackal]

Veranova, on 19 March 2014 - 03:39 AM, said:

The "prediction" is simply based on ELO between teams.

It doesn't intentionally weight one team higher in ELO, it's just the nature of the system that there may be some small variance.
Rewarding players for beating the odds when there IS a small variance makes sense therefore.

The problem is that Team Average ELO is a poor metric. 12 players at 1k ELO have the same Team ELO as 6 Players at 1500 and 6 players at 500. So, 12 competent or average players against a team of 6 above average and 6 that don't know what they are doing. Between the highest and lowest ELO in the match pool you've got a variance of 1000 even though the Team ELO is even. That's the flaw with the current system and why stomps keep happening despite "The System Is Working As Intended."

Matching based upon Team ELO creates a wide variance of skill and inconsistent results in gameplay experiences. What it should have been from the start is that each match should have a maximum variance on Player ELO to create a player pool for the match and then balance total team ELO to divide those players up evenly.

For example, if you set the maximum variance to 200 Player ELO then the average match at 1K ELO will contain 1100 ELO and 900 ELO players all evenly divided among the two teams so the Total Team ELO is within an acceptable variance. Even though Team ELO might be the same as the previous example this system creates a much more consistent gameplay experience between players and cultures a healthier game environment.

That said, there is a downside to all of that. Matching on Player ELO variance alone is far stricter than the Team ELO system that has been used since the start. This means that for it to function that it requires a larger active player pool.

The second downside, that at the more extreme ends of the ELO curves you have longer search times is easily fixed by implementing a Tier system once the ELO threshold that would create unacceptable wait times for games is past. So if that number was, lets say, 1750, then once you reach that magic number you'd get matched against everyone that reached it. But it gets far more complicated at this end.

Anyways, this is the fix that I have been preaching since MM problems were being talked about and addressed and for once it seems that PGI is actually using a smart design for a roadmap. Either they took one of my analysis posts to heart, which would be a small miracle as the last time they used one of my ideas they used my analysis of the potential pitfalls of Repair And Rearm to design the system to hit every last situation that they needed to avoid, or they are thinking on the same wavelength as I am, which would mean that they are thinking about the design of the game in the manner of the experiences it provides rather than the income such a system would generate.

Call me a cynical duck but I'm not going to question their motives on fixing the MM. There are other things to crucify PGI over.

[Veranova]

SuckyJack, on 19 March 2014 - 03:59 AM, said:

The problem is that Team Average ELO is a poor metric. 12 players at 1k ELO have the same Team ELO as 6 Players at 1500 and 6 players at 500. So, 12 competent or average players against a team of 6 above average and 6 that don't know what they are doing. Between the highest and lowest ELO in the match pool you've got a variance of 1000 even though the Team ELO is even. That's the flaw with the current system and why stomps keep happening despite "The System Is Working As Intended."

Matching based upon Team ELO creates a wide variance of skill and inconsistent results in gameplay experiences. What it should have been from the start is that each match should have a maximum variance on Player ELO to create a player pool for the match and then balance total team ELO to divide those players up evenly.

For example, if you set the maximum variance to 200 Player ELO then the average match at 1K ELO will contain 1100 ELO and 900 ELO players all evenly divided among the two teams so the Total Team ELO is within an acceptable variance. Even though Team ELO might be the same as the previous example this system creates a much more consistent gameplay experience between players and cultures a healthier game environment.

That said, there is a downside to all of that. Matching on Player ELO variance alone is far stricter than the Team ELO system that has been used since the start. This means that for it to function that it requires a larger active player pool.

The second downside, that at the more extreme ends of the ELO curves you have longer search times is easily fixed by implementing a Tier system once the ELO threshold that would create unacceptable wait times for games is past. So if that number was, lets say, 1750, then once you reach that magic number you'd get matched against everyone that reached it. But it gets far more complicated at this end.

Anyways, this is the fix that I have been preaching since MM problems were being talked about and addressed and for once it seems that PGI is actually using a smart design for a roadmap. Either they took one of my analysis posts to heart, which would be a small miracle as the last time they used one of my ideas they used my analysis of the potential pitfalls of Repair And Rearm to design the system to hit every last situation that they needed to avoid, or they are thinking on the same wavelength as I am, which would mean that they are thinking about the design of the game in the manner of the experiences it provides rather than the income such a system would generate.

Call me a cynical duck but I'm not going to question their motives on fixing the MM. There are other things to crucify PGI over.

Yup this is all good sense.
However like you say, by restricting to player ELO you're removing a whole load of match possibilities which reduce search times.
6 High 6 Low on each team isn't actually that fun for the 6 low, but it does mean everyone gets a match faster.
Obviously there will be some instances where the opposite team have more average ELO's and overall are better.

There aren't really any solutions here. ELO is the best we've got, and is set up in the best way to get good search times, while also having good matches *most* of the time.

Tier'ed ELO is kinda irrelevant, because by setting a threshold on someone's ELO, you ARE setting up a tier around the play who is searching. It's more efficient than fixed tiers, as in 1750+ there may be 23 players searching, but the guy with 1751 ELO could instead be matches with lower ELO players to form a match sooner, and the Matchmaker can then draw a few other players who are 1800-1900 in ELO to balance the match.


When weight balancing comes in, it won't solve our problems, but it WILL help constrict things to be a bit fairer for all players and weight classes. So ELO may re-organise a little, and hopefully be a bit more accurate for most players.

Edited by Veranova, 19 March 2014 - 04:17 AM.

[Willard Phule]

Veranova, on 19 March 2014 - 03:39 AM, said:

The "prediction" is simply based on ELO between teams.

It doesn't intentionally weight one team higher in ELO, it's just the nature of the system that there may be some small variance.
Rewarding players for beating the odds when there IS a small variance makes sense therefore.
And rewarding a player LESS when they had the advantage does too, because both of these factors help organise the ELO spread in to the bell curve.

It means you can't get in to the high ELO's easily, by accidentally rofl-stomping worse teams. But you can if you beat your equals or betters.

Right. Ok. And I get it. And I understand that a "simple" solution is something we're NEVER going to see. Because every solution even considered by PGI must, by definition, be over complicated and have the potential to break the game worse than it is right now.

But.

Just for a second, imagine how the matchmaker would work WITHOUT the whole "prediction" factor.

It puts two teams together with relatively even Elos.

The game gets played.

The winning team's Elo is adjusted upward based on whatever the hell it is they're basing it on now.

The losing team's Elo is adjusted downward the same way.

Instead of having a system where your Elo goes up or down slowly, you will have a much more dynamic change and separation of the different skill levels.

Look, ask any of the Derps out there. They don't want to be stuck in a match where everyone just rolls over their face any more than we want to be stuck with a bunch of lemmings. There's a learning curve involved here.

Take away the prediction, separate the skill levels quicker, things move on.

Granted, this doesn't address tonnnage, weight classes, ecm, etc...but, for crying out loud, this is something they could do NOW..without a major overhaul or patch. And it would definitely calm the masses before they give us a half-assed, broken launch module in April.

[Artgathan]

Is it possible that the only way to correct roflstomps is to add respawns?

Consider that without respawns, every casualty a team suffers servers to "snowball" their losses, because it's an outright loss of firepower (not a temporary setback).

[Dimento Graven]

My experience with this, when solo pugging, is very similar to OP.

It's incredibly frustrating because when I solo pug, launch, wait on average 3 minutes before finally finding a match, and am stuck with people who:

1. Are in trial mechs
2. Test firing weapons into team mates as soon as the match starts
3. Running off to 'Rambozo' the other team
4. Never, or very extremely rarely, pressing R to target the enemy.
5. Never posting ANY information in team chat
6. Attempting to fire SRMs and streaks at LONG range targets
7. Attempting to fire LRMs and PPCs into SHORT range targets
8. Not understanding the concept of torso twisting to spread damage.
9. Attempt to face tank an enemy much heavier armored and armed than themselves
10. Have built the crappiest 'mech possible (Atlas with nothing but small pulse lasers and SRMs, Raven 3L's WITHOUT ECM, etc.)
11. Tend to try and crawl up team mate's asses, preventing said teammates from being able to 'back out' of enemy fire.
12. Move to a 'known' brawl location and just stand there waiting for the enemy arrive instead of maybe looking for the enemy and attempting to get a strategic position on them
13. 'The Genius from France' military commander giving obviously bad advice that if followed only sets himself up with a good fire position and cover, leaving the 11 other team members as canon fodder

So on and so forth. It is EXTREMELY frustrating dealing with upwards of 11 other 'seemingly' unskilled pilots match after match after match.

My grouped experience is a LOT better of course, because if nothing else, it takes 3 of those unskilled/new players, out of that equation.

I definitely enjoy my pre-made experience tons more than I enjoy my solo experience, and it's the pre-made experience that keeps me in this game after two years.

[wanderer]

Quote

It doesn't intentionally weight one team higher in ELO, it's just the nature of the system that there may be some small variance.

HAHAHAHAHAHAHAnope.

Let me explain how the matchmaker actually works.

When it starts, it builds a "win" and a "lose" team.

The "win" team is composed of players with similar ELO ratings.

The "lose" team is composed of a few players with higher ELO ratings than the other team, "balanced" by plugging in players with inferior ones. Happily, since new players start dead center in ELO, they are frequently available for Team Loser fodder.

The result is a team with plenty of walking target practice for the winners, who then have numerical advantage and snowball the rest of Team Loser into a roflstomp. And when you're close enough to .500 W/L, it'll start plugging you into Team Winner instead until you're too far off the average, at which point the cycle repeats- sometimes back and forth, over and over every few games.

[Dimento Graven]

wanderer, on 19 March 2014 - 07:28 AM, said:

HAHAHAHAHAHAHAnope.

Let me explain how the matchmaker actually works.

When it starts, it builds a "win" and a "lose" team.

The "win" team is composed of players with similar ELO ratings.

The "lose" team is composed of a few players with higher ELO ratings than the other team, "balanced" by plugging in players with inferior ones. Happily, since new players start dead center in ELO, they are frequently available for Team Loser fodder.

The result is a team with plenty of walking target practice for the winners, who then have numerical advantage and snowball the rest of Team Loser into a roflstomp. And when you're close enough to .500 W/L, it'll start plugging you into Team Winner instead until you're too far off the average, at which point the cycle repeats- sometimes back and forth, over and over every few games.

Is this documented from PGI anywhere? What you're describing isn't how I remember what I read from PGI (a long time ago so I could be wrong).

If so, can you post a link?

[Joseph Mallan]

wanderer, on 19 March 2014 - 07:28 AM, said:

HAHAHAHAHAHAHAnope.

Let me explain how the matchmaker actually works.

When it starts, it builds a "win" and a "lose" team.

The "win" team is composed of players with similar ELO ratings.

The "lose" team is composed of a few players with higher ELO ratings than the other team, "balanced" by plugging in players with inferior ones. Happily, since new players start dead center in ELO, they are frequently available for Team Loser fodder.

The result is a team with plenty of walking target practice for the winners, who then have numerical advantage and snowball the rest of Team Loser into a roflstomp. And when you're close enough to .500 W/L, it'll start plugging you into Team Winner instead until you're too far off the average, at which point the cycle repeats- sometimes back and forth, over and over every few games.

This sounded a bit maniacal Wanderer... Where did the bad men touch you again? Show us on this paper doll.

[giganova]

Ah, ELO and matchmaker descrepencies that PGI and its fanboys will never admit to having. Working as intended...

Edited by giganova, 19 March 2014 - 07:38 AM.

[DaZur]

Again...

"Stomps" have little to do with the MM or the Elo aside from the proficiency of the team composite.

It's simple math... Force attrition on one side compounds force strength on the other....

There is no boogieman to blame this statistical anomaly on...

Edited by DaZur, 19 March 2014 - 07:37 AM.

[Varent]

Aresye, on 19 March 2014 - 01:06 AM, said:

Played 13 games tonight. I lived for 2 of them, and the funny thing is, I didn't even really notice that much of a difference in LRM play, nor was bothered by massive amounts of LRM spam.

But I am absolutely sick of this game. Not because of LRMs. Not because of meta (although that does play a small part in it). No, I'm sick of being continuously thrown into roflstomp after roflstomp, after roflstomp, after roflstomp, after roflstomp.

I experience a grand total of 2 gameplay experiences. I either get roflstomped, or we just barely manage to squeak a win by a very close game.

What gives?

I seem to always get a good blend of teammates that decide the best time to test their weapons is into my back armor before I even move. Meanwhile on the other team all I see is an incredibly competent group of players that on multiple occasions felt like I was dropping against a 12-man premade. Coordinated attacks from multiple angles, lights leading teammates away to get slaughtered by the main force, etc.

I don't consider myself an expert player. My average damage (win or lose) is usually around 300-600 each game, which I would put somewhere in the average to above average group when encompassing the overall spectrum of players, with experts getting 800+, and average players around 200-300.

Just to be clear, this isn't a rant. I'm more just puzzled as to why the whole match balancing system is designed the way it is.

What fun is there knowing that your match is predetermined before you even start it? Why do I even play a game knowing that there's some obscure system that is constantly pitting me against groups that are WAY above my skill level simply for the sake of balancing my stats out? I honestly don't know, and while this has bugged me for awhile, it wasn't until this stats reset and patch that I'm really starting to see Elo for what it is.

It seems more like a system that continuously punishes better players to no end, while simultaneously subjecting many newer, less skillful players to a horrible experience. Why is that, when the entire reasoning behind the system in the first place is to avoid newer players from continuously getting roflstomped?

It's being unfair to all parties, but even more unfair to the parties the system was designed to help. It isn't fair for me to complain about my teammates, because most of them don't really know any better, but does that make it fair to subject an entire group of newer players to an absolutely horrible experience for the sake of balancing a single person or small group of people?

I don't know what games are coming to nowadays. I just don't feel a draw to my playing experience being handled like an excerpt straight from 1984. There's only been 1 big issue I've ever seen that's important to game balance, and that's cheat protection. Beyond that, it's drop into a game, you play your best, and you either win or lose. No stat-based pre-determined matches, no ghost heat, no condescending developers treating you like a 5y/o girl watching a Barbie commercial, etc.

My overall experience of MWO has been based solely on how much I can tolerate, and it's getting thinner every single day. I'm usually a staunch defender of video games being an enjoyable pastime when people say things like, "They're a waste of time," but when nearly every single aspect of your gaming experience gets taken farther and farther out of your control, you reach a point that you realize you're nothing more than data itself. When you cease being a player, and instead are a script designed to operate a program with a singular, linear purpose.

That, in my opinion, is the very definition of a waste of time.

Amazingly long wait time or some misbalanced matches.

Take your pick, you cant have both.

No really, you cant, its a small community.

[giganova]

Because when I look at the illogical, ill-concieved, and conviluted chart that PGI posted about the matchmaking sequence, oh yes, I think "simple math"...

[DaZur]

giganova, on 19 March 2014 - 07:42 AM, said:

Because when I look at the illogical, ill-concieved, and conviluted chart that PGI posted about the matchmaking sequence, oh yes, I think "simple math"...

Strawman...

[TygerLily]

Varent, on 19 March 2014 - 07:37 AM, said:

Amazingly long wait time or some misbalanced matches.

I'd take supertight matchmaking over faster games and even Community Warfare...

[wanderer]

Quote

Is this documented from PGI anywhere? What you're describing isn't how I remember what I read from PGI (a long time ago so I could be wrong).

If so, can you post a link?

Won't be officially documented, ever. However, it can be easily observed simply via tracking team compositions over time.

This is a textbook example of how the MM functions.



Note that this game I started off with two of the three newbies in my lance alpha-striking me...once in the back, the second one in the side-which made it rather easy to kill me, too. And yes, all those Trial 'Phracts, Stalkers, and Cicada were all newbies.

Somehow, I don't think they were much of a match for the meta-horde on the other side, whose choices point towards a rather much more uniform level of skill relative to one another. While in most cases it isn't as obvious, this is what's happening to you every game.

[giganova]

DaZur, on 19 March 2014 - 07:47 AM, said:

Strawman...

Stop projecting. You claiming it to be simple math was a straw man argument and the burden of proof lies on you, good sir.