Jump to content

Matchmaker Adjustment - 06/05/2014


292 replies to this topic

#161 Cimarb

    Member

  • PipPipPipPipPipPipPipPipPip
  • Caladbolg
  • Caladbolg
  • 3,912 posts
  • Twitter: Link
  • Twitch: Link
  • LocationA hop, skip and jump from Terra

Posted 20 May 2014 - 04:16 PM

View PostWM Xitomatl, on 20 May 2014 - 02:52 PM, said:

Niko, I have a question about this thing:

Posted Image

Is the data gathered from mechs currently in matches, or is the data from people waiting to match? I ask because if it is from users currently in matches then it is bad data to go by, because then if it says that mediums are the best to take, then everyone waiting for a match chooses mediums and it takes forever for most of them to find a match.

And the reason I would possibly suspect it is because I just had 2 matches where I chose the recommended weight class and had to wait 5 minutes each time, then I tried selecting the "worst" one and got a match in less than one minute.

There is such a small difference between the weight classes, at least according to your screenshot, that it won't matter what you pick - it's purely based upon Elo and other matchmaker criteria.

#162 Roland

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 8,260 posts

Posted 20 May 2014 - 06:16 PM

View PostRoland, on 13 May 2014 - 07:45 PM, said:

Niko, could you please clarify these numbers that were provided by Paul?

In the reference, it mentions "number of drops". Was this just a master of confusing phrasing?

Specifically, in that data set, did a group of 4 players in a premade team who launched count as a single drop, or four drops?

Clarification of this point should be easily made, and would help us understand the composition of the player base.

Thanks.

Niko, did you ever find the answer to this?
Thanks

#163 Fatal25

    Member

  • PipPipPipPipPip
  • The 1 Percent
  • 120 posts

Posted 21 May 2014 - 01:26 AM

Can someone tell me the purpose of the matchmaker? Is it to prevent too many assault mechs or too many light mechs? Personally, I would be more concerned about 6+ light mechs in a match over assaults. I also had an issue today where my buddy and I, both in assaults, would get into a match, but in different lances. I hope that was a bug and not how it is intended to work. We get into a group so we can play together and use the strengths of each others mechs to our advantage, it is called tactics.

#164 Reza Malin

    Member

  • PipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 617 posts
  • LocationUK

Posted 21 May 2014 - 03:57 AM

View PostSoHxPaladin, on 20 May 2014 - 12:59 PM, said:


Being in a lance on comms isn't cheating first off. Second, if you are just a walking target for the premades, JOIN A PREMADE! This is supposed to be a team game, get with the curve! Third, if you are just a walking target for premades, CHANGE YOUR BUILDS! they gave you almost every tool they could *almost key word there* for the IS mechs to change everything in their loadout to fit the pilot. Fourth, after all those, if you still suck against premades, then you aint that great of a pilot, live with it and stop your bitching


Dude, stfu, you have no idea outside your own bubble. You sound like one of those arrogant douchebags that think because you are in a premade everyone should be like you, and that you are awesome. Newsflash: even crap pilots can be half decent with comms, i can just as easily say to you that you might be one of those crap pilots right? Its pretty much the same as you saying this guy is a bad pilot for not being able to beat a premade on his own. I mean what planet do you live on?

The guy is right to a certain extent about being a walking target for premades without some kind of matchmaker. If you like your premade so much, go do some private matches against other premades, or are you one of those premades that can only do well against PUG teams?

There is no defence against good premades other than a good team. Build has barely any bearing whatsoever. Oh yes i know, ill just change that XL for a Std, and now i can withstand 1000's of AC rounds from 2 different angles whereas before i got melted. No.

So yeah you are right in one sense, that teaming up should increase his odds. However they could just make a matchmaker that eliminates that need.

Not everyone is from the US/Canada, not everyone has all day every day to play, and not everyone thinks this game is amazing. These are reasons why not everyone is in a premade.

Maybe you are one of these types who played all the other mechwarrior games and devotes large amounts of his free time pretending to be a pilot with all the other battletech nerds, and not being in a premade is unthinkable for you. Well done. Not everyone is like that, for some this game is just a FPS mech sim shooter.

If you have nothing constructive to add, don't comment. Sick of hearing elitist ****** around here.

Edited by Fade Akira, 21 May 2014 - 04:01 AM.


#165 PhoenixFire55

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 5,725 posts
  • LocationSt.Petersburg / Outreach

Posted 21 May 2014 - 05:26 AM

View PostNikolai Lubkiewicz, on 20 May 2014 - 11:15 AM, said:

As you may be aware, 3/3/3/3 has been re-enabled with this latest patch.
Wait times have extended in some cases farther than expected. As a result, we are removing the one-premade per team limit at this stage, while retaining 3/3/3/3.


In other words ... there are enough lights and mediums to fill all matches but not enough solo droppers?

84% indeed ... ggclose.

#166 Roland

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 8,260 posts

Posted 22 May 2014 - 10:32 AM

View PostNikolai Lubkiewicz, on 13 May 2014 - 08:35 PM, said:


Good question. I'll try to get an answer for you on this one by the end of week or early next week.

Nikko, did you find the answer to that question? It's already past the early point of the week.
Thanks

#167 Roadbeer

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • 8,160 posts
  • LocationWazan, Zion Cluster

Posted 22 May 2014 - 10:36 AM

So...
Just checking in, did they give up on that 1 group per team thing?
Seems to me that if they can't get that to work, they should focus on one group being MOST of the team.... just sayin.

#168 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 22 May 2014 - 10:59 AM

View PostRoadbeer, on 22 May 2014 - 10:36 AM, said:

So...
Just checking in, did they give up on that 1 group per team thing?
Seems to me that if they can't get that to work, they should focus on one group being MOST of the team.... just sayin.

According to this, http://mwomercs.com/...ost__p__3394760 , they haven't given up on it yet.

#169 Cimarb

    Member

  • PipPipPipPipPipPipPipPipPip
  • Caladbolg
  • Caladbolg
  • 3,912 posts
  • Twitter: Link
  • Twitch: Link
  • LocationA hop, skip and jump from Terra

Posted 22 May 2014 - 12:25 PM

View PostBilbo, on 22 May 2014 - 10:59 AM, said:

According to this, http://mwomercs.com/...ost__p__3394760 , they haven't given up on it yet.

I am still hoping that rule continues to fail miserably, at least long enough for them to realize it is failing because that 84% stat is extremely incorrect and there are far more people grouping than they think.

#170 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 22 May 2014 - 12:29 PM

View PostCimarb, on 22 May 2014 - 12:25 PM, said:


I am still hoping that rule continues to fail miserably, at least long enough for them to realize it is failing because that 84% stat is extremely incorrect and there are far more people grouping than they think.

I don't have the numbers but I can't see them beating this thing to death repeatedly if their numbers didn't show that it should work. Will they get it to work? No idea.

#171 Roadbeer

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • 8,160 posts
  • LocationWazan, Zion Cluster

Posted 22 May 2014 - 12:31 PM

View PostBilbo, on 22 May 2014 - 12:29 PM, said:

I don't have the numbers but I can't see them beating this thing to death repeatedly if their numbers didn't show that it should work. Will they get it to work? No idea.

Didn't they beat Elo to death repeatedly (and continue to do so) even after it was proven to not work for... oh... 18 months now?

#172 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 22 May 2014 - 12:38 PM

View PostRoadbeer, on 22 May 2014 - 12:31 PM, said:


Didn't they beat Elo to death repeatedly (and continue to do so) even after it was proven to not work for... oh... 18 months now?

I'm not really convinced that Elo isn't working.

#173 Roadbeer

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • 8,160 posts
  • LocationWazan, Zion Cluster

Posted 22 May 2014 - 12:45 PM

View PostBilbo, on 22 May 2014 - 12:38 PM, said:

I'm not really convinced that Elo isn't working.

When you widen the spread to put a new player against a high Elo player in the interest of mitigating wait time, to say that Elo is working is to say that a successful launch proves Elo is working. What was the spread? 1400, that basically only keeps the super good and the super derp apart.

Or are we using buckets now? so 1 bucket from 1-1000 one from 1001 to 1499 and one for 1500 and above. That's really not making Elo work, that's just pigeon holing your players into 3 levels. You could just as easily do that without an overly complex algorithm.

#174 John Norad

    Member

  • PipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 524 posts

Posted 22 May 2014 - 12:49 PM

View PostSoHxPaladin, on 20 May 2014 - 12:59 PM, said:

Second, if you are just a walking target for the premades, JOIN A PREMADE! This is supposed to be a team game, get with the curve!

Dude, there are always going to be new players and players who prefer to play casual/solo. If this is all the solution you are willing to offer them, then you'd better stop talking about the topic altogether. Because being ignorant is not a good business practice. It's a horrible one, and I hope PGI doesn't think along those lines. Please don't try to make them.

View PostSoHxPaladin, on 20 May 2014 - 12:59 PM, said:

Fourth, after all those, if you still suck against premades, then you aint that great of a pilot, live with it and stop your bitching

Maybe you should stop your bitching and elitism, when MWO is struggling at best, get off that high horse and try to be constructive and start thinking about what's best for the game.
Hint: every F2P game needs players. Lots of them. Yes, even the unskilled ones. Skill bears no meaning or importance for the developer. Payments do.
So by all means stop discounting 'noobs' or players below your perceived level.

-edit-
A bit on topic: Elo should really come last. First you need a simple matchmaker that works. The devs should've used the Beta time to lay the foundations for all this, tweaking numbers and finding out what's important for matchmaking and what's not.
Then implement some Elo on top of that. They're doing it wrong. going from a so-so basic matchmaker directly to Elo was pretty much bound to fail.
My advice: back to square one. Build a solid matchmaker without Elo. Then proceed.

Edited by John Norad, 22 May 2014 - 12:58 PM.


#175 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 22 May 2014 - 12:55 PM

View PostRoadbeer, on 22 May 2014 - 12:45 PM, said:


When you widen the spread to put a new player against a high Elo player in the interest of mitigating wait time, to say that Elo is working is to say that a successful launch proves Elo is working. What was the spread? 1400, that basically only keeps the super good and the super derp apart.

Or are we using buckets now? so 1 bucket from 1-1000 one from 1001 to 1499 and one for 1500 and above. That's really not making Elo work, that's just pigeon holing your players into 3 levels. You could just as easily do that without an overly complex algorithm.

How long are you willing to wait for a match? They apparently have a target time in mind. Maybe the should adjust that.

#176 Cimarb

    Member

  • PipPipPipPipPipPipPipPipPip
  • Caladbolg
  • Caladbolg
  • 3,912 posts
  • Twitter: Link
  • Twitch: Link
  • LocationA hop, skip and jump from Terra

Posted 22 May 2014 - 01:00 PM

View PostBilbo, on 22 May 2014 - 12:29 PM, said:

I don't have the numbers but I can't see them beating this thing to death repeatedly if their numbers didn't show that it should work. Will they get it to work? No idea.

The problem is that they saw that 84% number and assumed wrongly on what it meant. It didn't mean 84% of the population WANTED to drop solo - it meant that 84% of the population had gotten so fed up or lost so many friends moving to other games that they had no other choice than to play solo. Both of those are assuming that 84% is even an accurate number, as I don't think PGI has even clarified whether groups were being counted as a single drop or set of individuals dropping...

#177 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 22 May 2014 - 01:04 PM

View PostCimarb, on 22 May 2014 - 01:00 PM, said:


The problem is that they saw that 84% number and assumed wrongly on what it meant. It didn't mean 84% of the population WANTED to drop solo - it meant that 84% of the population had gotten so fed up or lost so many friends moving to other games that they had no other choice than to play solo. Both of those are assuming that 84% is even an accurate number, as I don't think PGI has even clarified whether groups were being counted as a single drop or set of individuals dropping...

Without the data used to calculate the percentages given, I can't really make any judgments about their use/misuse of them.

#178 Roadbeer

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • 8,160 posts
  • LocationWazan, Zion Cluster

Posted 22 May 2014 - 01:05 PM

View PostBilbo, on 22 May 2014 - 12:55 PM, said:

How long are you willing to wait for a match? They apparently have a target time in mind. Maybe the should adjust that.

See, now that's where Matchmaker breaks with Elo.
As a solo player, you should have "Instant Action" there shouldn't be a wait, but now with the class restrictions put in, you have to wait based off of what mech you bring, a solo player, no matter how good isn't going to skew the battlefield so much that their having to be matched off of their Elo is worth the hassle.

As a group, I'm prepared to wait a little longer to get into a match, and PUGs could fill those holes. But then they over complicate it by taking individual Elo and averaging it to create a group "Elo" and try to match with that, when we all know that skill (as represented by Elo) isn't even a force multiplier when compared to coordination (VoIP) and cohesiveness (how well you communicate and mesh as a team), two factors that CAN'T be represented by Elo.

So, in those circumstances, I'm firmly of the mind that Elo is a myth, like the unicorn, or the clitoris.

#179 PropagandaWar

    Member

  • PipPipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 2,495 posts
  • LocationCalifornia

Posted 22 May 2014 - 01:17 PM

View PostZeriniel, on 06 May 2014 - 01:49 PM, said:

That's exactly what they did. In their calculations they counted a 2-4 man team as one drop, vs a solo player also being one drop.

So yeah. Horrible misleading math is misleading.

Misleading! PGI! Nah Never! Only George W. Bush is Misleading! Off to my affordable healthcare and Pre-Made vs. Premade combat. Good Day.

#180 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 22 May 2014 - 01:26 PM

View PostRoadbeer, on 22 May 2014 - 01:05 PM, said:


See, now that's where Matchmaker breaks with Elo.
As a solo player, you should have "Instant Action" there shouldn't be a wait, but now with the class restrictions put in, you have to wait based off of what mech you bring, a solo player, no matter how good isn't going to skew the battlefield so much that their having to be matched off of their Elo is worth the hassle.

As a group, I'm prepared to wait a little longer to get into a match, and PUGs could fill those holes. But then they over complicate it by taking individual Elo and averaging it to create a group "Elo" and try to match with that, when we all know that skill (as represented by Elo) isn't even a force multiplier when compared to coordination (VoIP) and cohesiveness (how well you communicate and mesh as a team), two factors that CAN'T be represented by Elo.

So, in those circumstances, I'm firmly of the mind that Elo is a myth, like the unicorn, or the clitoris.

I agree that grouping offers advantages. I don't believe Elo breaks because of them. Your Elo might skew high if you predominantly drop groups but that's because the inherent advantages of grouping are duly reflected.





12 user(s) are reading this topic

0 members, 12 guests, 0 anonymous users