Jump to content

If Elo Casts Such A Wide Net...


54 replies to this topic

#41 Joseph Mallan

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 35,216 posts
  • Google+: Link
  • Facebook: Link
  • LocationMallanhold, Furillo

Posted 13 March 2013 - 05:09 AM

View PostSerapth, on 12 March 2013 - 09:00 PM, said:

... that we end up with 3 or 4 players on the team doing well under 100 damage...

And it is completely disregarding mech weight like it is currently...


Why, oh why, did I just have a 8v6 match... again? How hard is it to find players for a match when the criteria seem to be non-existent??? At least pre-Elo you matched based on weight.


Seriously, what happened since last patch... before then, Elo matches seemed quite balanced, I never had a match worse then 8v7, and even the weight mismatches didn't seem quite so bad. The skill balance certainly seemed better. Has the community shrunk so you have so few players to choose from, or does the matchmaking algorithm really suck that bad now?

it is possible you had two crashes during loading. I have friends who almost make it into the game but freeze just before the drop. That could be tying up a drop space... right? :)

#42 Star Captain Obvious Kerensky

    Member

  • PipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 500 posts

Posted 13 March 2013 - 05:18 AM

If the primary purpose of Elo is to protect the new players from the veterans, there are much simpler systems to implement. Rather than an Elo matchmaker, PGI would only need to use the Elo score as a measuring stick:

"Everyone with Elo Above 2000 goes in this bucket (veterans), everyone below 2000 goes in this bucket (newbies)". The number of buckets could then be expanded into more categories as needed and server population warrants.

#43 Yokaiko

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 6,775 posts

Posted 13 March 2013 - 05:36 AM

View PostEldragon, on 13 March 2013 - 05:18 AM, said:

If the primary purpose of Elo is to protect the new players from the veterans, there are much simpler systems to implement. Rather than an Elo matchmaker, PGI would only need to use the Elo score as a measuring stick:




Yeah that **** doesn't work, when I play in the morning (US time) I see the occasional trial floating around. It tapers off and the games get better as the afternoon progresses.

#44 Star Captain Obvious Kerensky

    Member

  • PipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 500 posts

Posted 13 March 2013 - 05:41 AM

View PostThontor, on 13 March 2013 - 05:26 AM, said:

And they may or may not do that... But they should would try to make it work as it is though... Give them a chance to make adjustments, they haven't even made any yet... See if it gets better. Just don't use the one sides matches as a measure because those would happen even with "perfect" matchmaking.

I don't know why you would be against matching teams as evenly as possible... In Elo rating and weight... Without making the wait too long.


I'm not outright against it, but rather I believe the game lacks the server population to make it work, and the complex nature of this team game makes calculating a per player Elo score questionable at best.

I fear individualized Elo scores with so many uneven matches (in both weight and skill) are only serving to pollute the Elo scoring system.

Elo was originally designed for Chess, and was applied to MW:O. But to carry that analogy further:

MW:O is a chess by mail game where you have 8 players per team and each player gets to bring their own piece to the board. Sometimes you have a game of 3 knights and 4 queens versus a team of Rooks and Bishops. Some teams communicate by telephone and play together daily, some team communicate by carrier pigeon and have never played together before.

So I am exceptionally incredulous that any kind of per player Elo score would be accurate enough to form a reliable matchmaker by comparing scores directly.

Edited by Eldragon, 13 March 2013 - 05:42 AM.


#45 Glythe

    Member

  • PipPipPipPipPipPipPipPip
  • 1,566 posts

Posted 13 March 2013 - 05:57 AM

View PostAethon, on 12 March 2013 - 09:36 PM, said:


ELO was not the answer, though. It was never needed. I do not need or want some computer artificially regulating my win/loss ratio, just to hold everyone's hand. It also penalizes groups of good players. The better the players I end up with in my 3-4 man groups, the worse the matches end, because the other 4 people hardly deal any damage before they run out and die by themselves. Just because I know some special little snowflake is going to claim that I think all pugs suck, am an elitist so-and-so, etc., I had better put in the usual disclaimer: I have seen good and bad pugs...but ELO balances the good players by putting really bad players on their team, which pretty much turns the match into 4v8. It also is not very fair for these players, because they are being dragged up into matches against players of a much higher average skill level.

If you want pugs to have a fair shake, all you have to do is give them a 'pug only' queue, so they do not have to deal with organized groups of pugstompers on teamspeak. This is fair, this is simple, and this does not penalize the people who have 2-3 people who want to have fun together.

As it presently stands, I have to drop solo to get any good matches, which kind of defeats the purpose of being in a unit. Sure, we can drop 8-man groups, but we only have 8 of us online at certain times of day; the rest of the time, we pretty much have to pug, or try to suffer through the crappy matchmaking system if we want to have fun playing together...which really is not much fun.


Everything this guy is saying is dead on if you are on the upper end of the Elo spectrum. I couldn't have said it better myself.

#46 Yokaiko

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 6,775 posts

Posted 13 March 2013 - 05:59 AM

View PostGlythe, on 13 March 2013 - 05:57 AM, said:


Everything this guy is saying is dead on if you are on the upper end of the Elo spectrum. I couldn't have said it better myself.



Except that I'd wager there isn't a population to support a group only queue, not and get games to launch. Ever try to do 8s outside of US prime? I have never seen a game launch.

#47 Deathlike

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 29,240 posts
  • Location#NOToTaterBalance #BadBalanceOverlordIsBad

Posted 13 March 2013 - 06:04 AM

You will always be lumped in with low ELO players.

The MM system "hopes" your overall skill would help them out, but sometimes it just won't happen. Some people will still suck, and some people will still pwn everything.

The difference now is that you're getting less newbies and actually get more competitive matches. Unless.. winning tougher matches is too hard for you? Winning and losing in PUG games should not be detrimental to you ego. If that is the case, you have more issues that we care to know of.

#48 SgtKinCaiD

    Member

  • PipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 1,095 posts
  • LocationBordeaux

Posted 13 March 2013 - 06:19 AM

View PostThontor, on 13 March 2013 - 05:26 AM, said:

I don't know why you would be against matching teams as evenly as possible... In Elo rating and weight... Without making the wait too long.

I'm totally for it, but actually it's not working and end up with worse balanced matchs.

And as i said in another post, PGI will spent a lot of time to tune it (and balance the game accordingly), time that will not be spent on CW, although everybody was asking for CW and none for ELO ratings.

Plus, even if one day we have the perfect matchmaking system, it's still utter nonsense to test mech build, farm cbills, grind xp in ranked matchs. Ranked matchs means you play with your best mech already tuned, a group already organised to play with, and so on.

I totally support the idea of Aethon to have a 'pug only' queue. It was a quick way to resolve the problem of pugstompers.

#49 Glythe

    Member

  • PipPipPipPipPipPipPipPip
  • 1,566 posts

Posted 13 March 2013 - 09:25 AM

View PostDeathlike, on 13 March 2013 - 06:04 AM, said:

You will always be lumped in with low ELO players.

The MM system "hopes" your overall skill would help them out, but sometimes it just won't happen. Some people will still suck, and some people will still pwn everything.

The difference now is that you're getting less newbies and actually get more competitive matches. Unless.. winning tougher matches is too hard for you? Winning and losing in PUG games should not be detrimental to you ego. If that is the case, you have more issues that we care to know of.


Nope it doesn't work like that.

4 great players+ 4 epic fail players really doesn't really have a fair chance against 8 good players.

It's about hoping for a fair match. If we bring 4 good players the match maker will not give us equal skilled opponents 90% of the time. When it works you get a fantastic match..... and it fails miserably the rest of the time.

Most games have an unranked ladder and a ranked ladder. Unranked 4 man group would be fantastic.

#50 Genewen

    Member

  • PipPipPipPipPipPip
  • 355 posts

Posted 13 March 2013 - 09:27 AM

View PostDeathlike, on 13 March 2013 - 06:04 AM, said:

The difference now is that you're getting less newbies and actually get more competitive matches. Unless.. winning tougher matches is too hard for you? Winning and losing in PUG games should not be detrimental to you ego. If that is the case, you have more issues that we care to know of.

I wish I had as much competitive matches. Way more fun than roflstomping newbs.
Instead, I experience what several others here have already said: I get lumped together with a bunch of people who cannot discern their elbows from their butts. Matches in which several assault mechs and heavy mechs don't do more than 30dmg. Matches in which 6/8 teammates are under 50 dmg/each.

Even if someones elo was high enough to mathematically make up for it, a team is more than the sum of it's part. One or two individuals with high elo and terribad teammates will have less chances to win than a team only consisting of medium players. Especially since those bad players often not only have problems with managing their mechs, they also mostly suck in the tactics compartment as well and rarely listen to others.

If the matchmaker worked fine to actually give out mainly competitive matches, that'd be great. But right now, it does not. There are huge discrepancies in skill, tonnage and, also important in my book, ecm capabilities. 0 ECM vs. 3 ECM don't make a match easier.

#51 NRP

    Member

  • PipPipPipPipPipPipPipPipPip
  • Fire
  • Fire
  • 3,949 posts
  • LocationCalifornia

Posted 13 March 2013 - 09:56 AM

I have to admit I have been experiencing the same thing as Serapth. After they initially implemented ELO, I had some of best matches ever, regardless if I won or lost. But recently, the quality of the matches has really taken a steep down turn. Either people have figured out how to game the system, or PGI changed something behind the scenes.

What I think I'm experiencing is this: there is too much variance in skill on a given team. Lately, the teams I've been on usually has one or two superstars who carry the rest of us. I find I am getting slaughtered even when my team wins.

It's weird. It's screwy. And it is not fun. I find I can only stomach a few matches before I quit and do something else. I hope it is something PGI can fix.

#52 Aethon

    Member

  • PipPipPipPipPipPipPipPipPip
  • Mercenary
  • 2,037 posts
  • LocationSt. Louis, Niles, Kerensky Cluster

Posted 13 March 2013 - 10:17 AM

View PostThontor, on 13 March 2013 - 04:39 AM, said:

There haven't been any changes to the matchmaking system since they implemented phase 3 on February 19.

Matthew Craig said right in those quotes I posted that they haven't done anything yet.

I certainly haven't noticed any difference since then. Didn't see any change with the March 5 patch.




I think Matthew Craig explains that better than I could:



Matt again:





I don't see how adding a friend adds "boat anchors" to your team. Unless your friends are boat anchors ;)
I definitely don't see how adding a friend could possibly make you lose more...



Its pretty obvious really, Elo ratings main purpose is to protect new players from the highly skilled veterans. To attempt to place similarly skilled players into a match together. Keep in mind, as Matthew said, even a match with equally skilled teams can end in a one sides victory, while the next one might end in a one sided victory for the other team.

Its not perfect, there are kinks, the only way they could even see and iron out those kinks is by testing it on live servers.


You can quote devs all you want. You can tell me how much better you think everyone's experience is all you want. You can even tell me how much better the game experience is for the new players. But the truth of the matter is that my experience in MWO has gone straight down the *******. No devquote is going to change that fact, or make me want to spend any money on this game until that changes.

You want a list of the idiotic stuff I have seen out of the pugs on my teams lately? I will give you a couple very brief examples, but I could spend all day typing a list. It was NEVER this bad before; we would usually get a 50/50 mix, on average, of good/bad pugs. Now, if we have a 4-man group, our pugs drool on themselves for a bit, run out to the middle of the map, and just stand there shooting at people out in the open until they die...and this is not just Atlases. I had a Commando on my team last night who hid in a corner the entire match...and then, when he was found, he did not even turn his chassis; he just stood there blazing away at the 7 people left on the other team (gee, big surprise there) with the awesome firepower of 3 flamers and an SRM-4. I also had an LRM60 Awesome run out and facehug someone with his LRM's until he died. He ended the match with 0 damage.

So, yes. This was a massive step backward for MWO, from my point of view. I cannot play with my Clanmates outside of 8-man groups (when we have enough people online) anymore, because it ends up being 4v8 every time.

View PostEldragon, on 13 March 2013 - 05:18 AM, said:

If the primary purpose of Elo is to protect the new players from the veterans, there are much simpler systems to implement. Rather than an Elo matchmaker, PGI would only need to use the Elo score as a measuring stick:

"Everyone with Elo Above 2000 goes in this bucket (veterans), everyone below 2000 goes in this bucket (newbies)". The number of buckets could then be expanded into more categories as needed and server population warrants.


THIS. This is what WOT has been doing, and it WORKS.

When new players join the game, they only play against other new players for their first 10-20 matches. That gives them time to learn, before they are dumped into the deep end of the pool. If we had this, but not the rest of our current ELO system, it would solve the new player experience, without screwing the more dedicated players.

#53 Serapth

    Member

  • PipPipPipPipPipPipPipPip
  • 1,674 posts

Posted 13 March 2013 - 11:47 AM

View PostDvergar, on 13 March 2013 - 04:48 AM, said:

In addition, groups need to be matched. Twice recently I've been in games where 8 pugs get put up against two 4-mans. How is that remotely good game design? Group drops should be mirrored. 2man/6pug v 2man/6pug, 4man/4pug v 4man/4pug, 4man/2man/2pug v 4man/2man/2pug, etc.


This is one of those ongoing infuriating parts of MWO/PGI, they refuse to acknowledge there is a huge advantage to being in a group, especially a group on comms. With Elo getting horrifically bad again, at least for me, once again being able to choose 3 of your lancemates is a gigantic advantage. Simply averaging a teams ELO rating is a downright absurd solution.

#54 Serapth

    Member

  • PipPipPipPipPipPipPipPip
  • 1,674 posts

Posted 13 March 2013 - 11:50 AM

View PostEldragon, on 13 March 2013 - 05:18 AM, said:

If the primary purpose of Elo is to protect the new players from the veterans, there are much simpler systems to implement. Rather than an Elo matchmaker, PGI would only need to use the Elo score as a measuring stick:

"Everyone with Elo Above 2000 goes in this bucket (veterans), everyone below 2000 goes in this bucket (newbies)". The number of buckets could then be expanded into more categories as needed and server population warrants.



If I am honest, I thought this was exactly what MWO did. It certainly felt like that for a while... you were grouped with skilled teammates and faced skilled competition. Then... something changed. Now trial mechs are back, as are lopsided match balances.

#55 SgtKinCaiD

    Member

  • PipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 1,095 posts
  • LocationBordeaux

Posted 13 March 2013 - 04:09 PM

Another wasted evening thanks to the matchmaking ...





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users