183 replies to this topic

[MischiefSC]

jay35, on 01 April 2013 - 07:54 PM, said:

No, but my fun does matter too, is all.

You can keep repeating that mantra all you want but it doesn't excuse the way the system works right now.
Match after match that's a frustrating mess is not an enjoyable experience, especially for someone who's used to have a fun experience a majority of the time. Sacrificing good players' fun for the sake of bad players is a crappy way to "balance" things. It's not really balance at all, it's bringing everyone down to a lowest common denominator, in this case 50:50 or 1:1. Plainly, anyone accustomed to doing much better than that is simply not going to enjoy the experience as much.

I absolutely get where you're coming from. Winning is fun. The problem is that regardless of their skill at the game everyone has an equal right to 'have fun' in the game. Just because you're better doesn't mean you get to have more fun at the expense of newer or less skilled players.

Think of it like just about any sports. Football in the US for example. They don't get to just go play full tilt against High School football teams for kicks. It would be horrible and while hilarious from a 'look at those poor ba$tards getting pummeled' standpoint it's not really doing anything for anyone save the pro team that gets to flex their muscles against people they clearly out-match.

Every sport and contest out there has leagues to separate people by skill and match comparable to comparable, if only within a certain range. That's all Elo is. A way to separate players out into bands of comparably skilled players to prevent pro league folks from brutalizing bush league folks.

Play more too. It helps. The more games you play the better settled your Elo will get. As in like 1,000 games since the stats changed, mostly in the same weight class. It makes a BIG difference.

[Covered]

First, stop calling it ELO. Its the Elo Ranking system. Its a guys name, not an acronym.

Second, the Elo ranking system has been proven, in theory and in empirical evidence, to work. Its not the ranking system, or at least in theory. Yes everyone wants to say that they dropped into a game with a bunch of putz's and that is why they lost. Sorry, there are mountains of evidence that will literally smash you into jelly (if you were to actually print them out and try to stack them on yourselves...) that indicate your anecdotal evidence as to the broken nature of the Elo Ranking system is an isolated statistical anomaly.

Your implicit assumption is that you are always being matched as a high rated player in a low rating game. Hence you are matched with a lot of low rated players and EVERY time you queue you get average players who stomp you into nothingness because 6v1 isn't fair. Lets, for a second, ignore the fact that you can as easily be pitted against a another team in the exact same boat, ie a high rated player with a lot of low rated players. The simple fact that you lose, brings your ranking down, by quite a lot might I add if it is a low rated game, and hence brings the player level more in line. As in the next time you queue you pop into a game with players who are rated higher (assuming the same starting conditions, you are the anomaly artificially inflating the ranking). If you win, your ranking climbs with theirs, in fact yours climbs less, so they get brought even MORE into line with your ranking.

Now, where the implementation is broken for MWO, tonnage and value based drops. Sorry by a Spider is not the same as a Raven, same with an Atlas and an Awesome. But this adds more levels of filters onto the queuing system.

Another element of broken Elo Ranking system is the way average game ranking value is calculated, or decided. Unless there is something missing from the dev posts as to how they calculate it, you will invariably get ranked games at the average player ranking currently queuing up. And judging from where you say you are located, you are probably playing in a more off hours time, which can make the queue rank fairly... volatile.

Simple solution to fixing this "problem" is not "ELO is ****** get rid of it", since it is not the Elo Ranking system, it is the implementation. First is to add BV/Tonnage/C-Bill cost drop limits, which according to their development plan is eventually going to be pushed out. Second is to break the ranking system into tiers, so that you only get paired with people of similar ranking, now people will probably complain about that as well since their K/D and W/L ratios are going to drop to near 1.0, and queue times are going to grow drastically for the top and bottom percentage.

Just brainstorming, but they could implement a dynamic tiering system that looks at the average queue ranking over a period of time and develops tiers based on the distribution and player population.... ie, break the population into 3 equal tiers, or 3 tiers of "playable" pop sizes, meaning queue times are appropriately short, and allow match making within those tiers and not cross tier...

But in reality, there is no "Elo hell" and no, Elo isn't a broken ranking system, it works as its intended.

[Corwin Vickers]

MischiefSC, on 01 April 2013 - 07:45 PM, said:

I'm not so sure about that.

In terms of the matchmaker, at least according to Paul in a relatively old post, it's going to average you guys out and fill your team with people of a skill level equal to your average. Then do the same with the other team.

The matchmaker is specifically NOT designed to try and balance good players with bad on the same team to get an average that's equal to the other teams average. Here's the quote from Paul:



The idea that a good player will be pulled into a team full of bad players to match against a team of all above average players is a myth. It doesn't work that way.

All Paul says is an average. I seem to remember him describe it as a target goal average or something.

In any case if we have two guys at 1000 points and one guy at 500 and one at 1500 there is only one way to match these players up.

The two average players with 1000 Elo will be put on one team and the other two on the other. Given limited pools of players in the queue at the same time and instances like this are bound to happen.


And my example of 4 mans sticking you with lower ability randoms is based on the 4 players being awesome as everyone always states. If your 4 man is just the same average distribution then you shouldn't notice a difference between pugging and 4 mans.

Also, I don't think you can say it doesn't happen (high + low players) because I see it happen a lot but I try to work with it. I see it less when I drop solo.

Covered, on 01 April 2013 - 09:25 PM, said:

Now, where the implementation is broken for MWO, tonnage and value based drops. Sorry by a Spider is not the same as a Raven, same with an Atlas and an Awesome. But this adds more levels of filters onto the queuing system.

I don't think that is right. Let's assume for a minute that assaults are a superior factor to whether your team wins or not compared to lights. That would mean you'd have a worse win/loss record on your light mechs than your assault mechs. That would mean you have a lower Elo rating on your light class than your assault class and that would mean you would face weaker opponents on your light mechs than on your assault mechs which should average out your wins for each class.

With cap mechanics and the atrocious assault pilots out there I don't necessarily think assaults are more valuable for winning. Also factor in that most people are trying to kill steal and wrack up damage and couldn't give two ***** about winning.

I am not reading closely enough tonight. I don't think it's tonnage that differentiates an Atlas and an Awesome, it's the hardpoints so a tonnage based breakdown wouldn't help. Something BV-esque maybe but it would have to be subjective since the system is too complex for an objective measure.

Edited by Corwin Vickers, 01 April 2013 - 10:15 PM.

[MischiefSC]

Corwin Vickers, on 01 April 2013 - 10:10 PM, said:

All Paul says is an average. I seem to remember him describe it as a target goal average or something.

In any case if we have two guys at 1000 points and one guy at 500 and one at 1500 there is only one way to match these players up.

The two average players with 1000 Elo will be put on one team and the other two on the other. Given limited pools of players in the queue at the same time and instances like this are bound to happen.


And my example of 4 mans sticking you with lower ability randoms is based on the 4 players being awesome as everyone always states. If your 4 man is just the same average distribution then you shouldn't notice a difference between pugging and 4 mans.

Also, I don't think you can say it doesn't happen (high + low players) because I see it happen a lot but I try to work with it. I see it less when I drop solo.

In the quote he specifically says that it's a target, not an average. So with entry Elo being 1300 (that's newbie Elo) let's say your Elo is 1800. It won't drop you with someone who's 800 to get an average of 1300. It'll populate the match with 16 players with as close to 1800 Elo each as possible.

Where this gets screwy is premades. Suppose you have 1800 and you are dropping with 2 friends, each with Elo scores of 1300. It'll count you as 1467 (possibly, Garth did say in another conversation that group averages would be 'weighted' as in it might skew towards the higher level player) Elo and drop you in a match with 13 people with an Elo of around 1467.

The result being that you, with an 1800, would gain little for winning the match and lose quite a bit if your team lost the match. Your teammates would, conversely, gain a lot from wins but drop little from losses.

Make sense?

[Corwin Vickers]

Makes sense but at least with the selection, I don't think it's working the way you describe. Or at least it's definition of "close" to the the target number isn't the same definition of "close" as we understand it. I see the same guys who run to the top of the hill, stop moving, and then start shooting LRMs at people at 170 meters that everyone is complaining about. Not as often as when they first put Elo in but it still happens.

And that effect is amplified for 4 mans because of short queues and very limited pools of players.

Given that it's already casting a very wide net on occasion, and that it sometimes "fails to find a match" that means that there were not 16 players in the queue at that time. If that is true then ideas that split the community could be game killing.

I am assuming that it casts a wide net because some of these guys have to have someone help them tie their shoes. I mean seriously, what are people doing with LRMs these days, much less ONLY LRMs with no backup weapons, probably not even TAG.

[Hekalite]

It's almost impossible to have a constructive discussion on this subject because everyone's experience with the system will depend on their skill level and their play style. I personally find the system to be abysmal at finding a fun and fair match during peak hours, but it's not too bad in the morning. I've been dropping solo 99% of the time since my group evaporated. During prime time, it seems like the system almost always puts you on one side or the other of a stomp.

Anyone who thinks the system is "working" right now is delusional. While it isn't totally broken, it does need to be heavily tuned. Fortunately, the devs have stated they will continue to adjust the system. Unfortunately, they move at a slugs pace.

For the 4-man teams out there who are frustrated with their solo teammates, I have to concur with what was posted earlier in the thread. You guys need to communicate in chat more if you expect help. I hardly ever see anyone dictating strategy, and it's even more rare that I see communication past the initial let's do X phase. Sometimes what you interpret as a solo player running off to get killed is that person pushing forward expecting the three or four mechs behind him to follow, only to discover they had other ideas. I have not yet unlocked the ESP module, so please help me out.

[FerretGR]

Mad props to MischiefSC for being a drop of reason in a pool of confusion. I've tried to fight this fight and it frustrates me. You're consistently giving this perspective and you're consistently calm in the face of folks repeating the same misinformation and arguing that misinformation with you. You're a better man than I

[Elfman]

Since the Elo patch the general fun factor for MWO for me at least has certainly decreased to a point that I am seriously thinking about taking a break till they fix this mess.

Been playing since early closed beta and the present match making is the worst (in my opinion) since they introduced weight balancing. Most of the matches are a complete wash 8-0 8-1 and it seem I have to work at the game to balance out the pure hopeless players I get teamed up with on a regular basis.

Other than when doing 8 mans on clan nights pugging or 4 mans seems to more often than not end up with 1 hot shot 2-3 average players and 4 either new players (not there fault) or solo Rambo do what they want and die leaving the team 4-0 down within 2 mins of game starts.

Seems the win or loose depends now more on the level on the 4 worst players on each side and unless I take an assault or a full cheese build its generally a bad match.

When you do 500-600 damage get 4-5 kills and still loose the match as the other randoms have all died and left you to solo the other team its why do I bother and recently I have just stopped bothering to log on and play. It wouldn't be so bad if they at least brought some weight class balancing back in.

And from the guys i normally play with there defiantly less of them logging on and with the same complaints. 8 mans are fun and all but when you have had a 12-14 hour difficult day at work and don't need the hard graft of 8 man's then these days I go watch TV or play something else

[RussianWolf]

Tezcatli, on 31 March 2013 - 07:53 PM, said:

This has been a recurring theme in threads. They need to implement tonnage limits on matchmaking. Even if someone is highly skilled, they're still fighting an uphill struggle against heavies and assaults. It'd be like sticking a light weight fighter against a heavy weight fighter. Sure he could be skilled enough to beat him, but he would still be at a considerable disadvantage.

go watch some of the MMA fights before they broke into weight divisions. Gracey was dominant at about 200 pounds over guys weighing 300+

Skill will overcome size nearly every time.

[Yokaiko]

RussianWolf, on 02 April 2013 - 06:01 AM, said:

go watch some of the MMA fights before they broke into weight divisions. Gracey was dominant at about 200 pounds over guys weighing 300+

Skill will overcome size nearly every time.

Not when you are consistantly left with 3-1 odds.

[FerretGR]

What is the mechanism by which that's consistently happening to you and not to your opponents?

[Yokaiko]

FerretGR, on 02 April 2013 - 06:17 AM, said:

What is the mechanism by which that's consistently happening to you and not to your opponents?

4 games in a row today, 3 kills is the best if seen, I get teamkilled chasing an enemy light, and 3 other times when I was either capping on conquest or covering a basecap attempt on assualt I look up and see 2-5 or 2-6 on the scoreboard.


I used to like pugging I'm about to say the hell with it again.

Just looked 4-11 since yesterday afternoon, and all since I decided to master out my Jenner-K (I know)

Edited by Yokaiko, 02 April 2013 - 06:24 AM.

[FerretGR]

Your tiny amount of anecdotal evidence doesn't answer the question I asked.

[El Bandito]

Personally, I never liked to face a group when pugging as it usually results in steamrolling.

That is all. Let groups queue against groups only.

Edited by El Bandito, 02 April 2013 - 07:06 AM.

[Elfman]

I have the feeling its a time zone thing I wonder if the people not noticing the issue are in a time zone with more players on and us who notice this issue are in a more sparse play base time zone.

And for me at least a main problem is getting at least 2-3 people not connect or disconnect as soon as the match starts or when I do get an 8 v 8 its 8 of us in mediums, heavies vs 4 ravens and 4 assaults

Just bring back flipping weight class matching as well as Elo.

[Yokaiko]

No idea, but my w/l dropped from 3:1 pre ELO ******** and is down to 1.14:1.

Four DAYS at a time winning 1:4 blows, period and seeing 3+ people under 100 damage per game blows harder.

[Whompity]

I generally seem to do a bit "better" (KD-wise) earlier in the evening... 9pm-10pm EST. After that, I seem to go downhill a bit... maybe more people are online so the ELO matchups are tighter then. At least, that's my assumption for why my kills go from 2-3 to 0-2 and my deaths seem to inch up a bit.

[Yokaiko]

Elfman, on 02 April 2013 - 07:05 AM, said:

I have the feeling its a time zone thing I wonder if the people not noticing the issue are in a time zone with more players on and us who notice this issue are in a more sparse play base time zone.

And for me at least a main problem is getting at least 2-3 people not connect or disconnect as soon as the match starts or when I do get an 8 v 8 its 8 of us in mediums, heavies vs 4 ravens and 4 assaults

Just bring back flipping weight class matching as well as Elo.

Yeah, a lot of that ****.

Olivia Maybach, on 02 April 2013 - 07:10 AM, said:

I generally seem to do a bit "better" (KD-wise) earlier in the evening... 9pm-10pm EST. After that, I seem to go downhill a bit... maybe more people are online so the ELO matchups are tighter then. At least, that's my assumption for why my kills go from 2-3 to 0-2 and my deaths seem to inch up a bit.

I only get to play am EDT.

So I guess im just hosed, because its no fun anymore.

[TexAce]

If ELO really just adds the rankings together (without any borders) to get an average and then tries to find a team with same average, it's not a good design.

How I picture it to be right now (with fantasy numbers)

#1: 10 ELO points
#2: 2 points
#3: 2
#4: 2
#5: 4
#6: 2
#7: 1
#8: 3
= 24 ELO points
-> tries to find a team with average 24 ELO points.

how I think would be better:

- Same 8 people as above
- ELO removes player #1 because his ELO is too high for the others and tries to find someone with an ELO of 1-5
- lets say it finds someone with a group ELO of 2

new group would have not 24 ELO points but 16
-> find a team with 16 ELO points to match them against so all "bad/new" players would play together.

And the one player with an ELO of 10 who got removed should be placed in a group with ELOs of 8-12 resulting in a group ELO of ~80 where everybody is equally good. Then the MM should find another group with an ELO of ~80 to match them against.

Perhaps the ELO right now works like I think it should be BUT we just don't have enough players for it to shine.

Edited by TexAss, 02 April 2013 - 07:31 AM.

[Whompity]

The more I think about it, the more I think the issue is with the number of players available at any given time (as others have stated). I suspect that in the early evening, with fewer online, the MM is pulling in people with a wider range of ELO, essentially giving me a higher number of NEW players to shoot at (no offense intended). That said, it SHOULD also pull in a larger number of better-than-me players in... why doesn't it seem to be doing that? I know I'm not that good... in theory the high and low should balance out.

Later in the evening, when there are more of each ELO to draw from, the games should be "tighter", which they appear to be, at least to me.