Jump to content

Why Elo Doesn't Work Here


633 replies to this topic

#381 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 25 January 2014 - 09:02 AM

View PostNightfire, on 25 January 2014 - 01:21 AM, said:


Oh! My! God! Can you straw-man any harder!? I hate sports games and yet you somehow manage to translate that into my mental conduct and psychological state. You have no idea what I am like and furthermore, sports games not the pinnacle of online sportsmanship and respect of ones adversary to the exclusion of any other team game. To infer as much is pure arrogance. To infer that I lack these qualities because of my dislike for sports games boarders on liable. Restrain yourself Sir!

The PC was the ONLY viable platform for sports games pre-PS1. Some sports games existed on early consoles but the PC is where the bastion of processing and graphic resources lay. As this changed, so did where Sports games were published. It is far more sociable to play in the lounge on a TV screen with people than in front of a PC alone. I have never said Sports Gaming on the PC wasn't viable, simply:



and I stand by that statement! The PC was the only VIABLE platform at the time. It ceased to be the ONLY viable platform and sports games IN GENERAL moved to the better suited platform. To infer that I stated PCs were not viable platforms for Sports Gaming means you are either intellectually dishonest or you lack critical reading skills. Your entire rant to this point lies on the premise of the straw-man you have built to vilify me.


MWO is a sport! Its not the roleplaying game you wanted. get over it. Most of the mentalities here is why the game is getting even less popular then it already is.

I don't need to quote any of your further comments. This thread is getting spammy now.


You've already explained it all in your previous post when you said you hate sports games and your glad they are not on the pc anymore. ....WHich is basic proof of how pc's are unpopular nowadays. Most of society likes sports I'm sorry. Nerds who don't are still a minority.

Playing tiger woods 2008 with players from around the world was very sociable. And I keep explaining, it was a golf game that made the pc more popular for sports games over the atari and intellivision, in 1986 mainly Because of the graphics. But also all the other abilities, like mapmaker, mods, etc. Many sports games had community sites that updated rosters and player attributes, stadiums, player models, etc... Graphics are still better on the PC. And PC's are still better for multiplayers and online communities. You can follow the evolution of graphics on the pc through Golf games on the computer up until 2008.

Most Sports games are no longer on the pc, mainly because of piraters, hackers and cheaters. Soccer being the exception since its mostly Americans abandoning the internet nowadays. But its a fact most gamers on the pc nowadays, feel like you. Its basically only strategy games and roleplaying games that are popular on pc nowadays. People don't want to deal with cheaters and technical problems. Most people left on the pc are computer nerds or work in the industry, and like you, they don't like sports.

Most RPG games are losing money, and cost way too much to develop. Which is probably why PGI went the arcade arena shooter route. Sorry that ruins it for all you long time fans of the Genre. But there is no reason to selfishly drive off newcomers, who do enjoy sporstmanship, by encouraging PGI to sabotage their own game. :P

Again, basing the ranking/rating system on wins alone will drive more players away when:

1. its a random team game

2. the game is based on cbills.

Edited by RichAC, 25 January 2014 - 09:30 AM.


#382 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 25 January 2014 - 09:50 AM

View PostAbivard, on 25 January 2014 - 12:36 AM, said:

Intelligent, well educated, ignorance.
You are parroting ideas without understanding them.

You keep confusing correlation with causation.

BTW, M.I.T. has no connection to Mr Elo whatsoever! In fact, M.I.T. has very little to do with theoretical mathematics, it is an engineering school, not a math school.


So show me the math that shows how Elo doesn't work. MIT was brought up because TrueSkill, the current 'peak' of matchmaking systems which is also based on Elo, was reviewed and evaluated by a peer group at MIT. Roland linked in a paper from them some pages ago, that's where the reference was.

You're attempting to represent little witticisms and soundbites as actually having some technical basis or value. Why don't you put up a mathematical model representing how the Elo formula can not work. Show me the complexity breakdown where suddenly statistical modeling ceases to function when you get into 12 v 12 representations. Show some sort of actual useful data that clarifies that somehow you can not solve for one factors contribution in a roughly weight and skill matched 12 v 12 environment over sufficient sample sizes.

Here is the link to the MIT article on TrueSkill again. TrueSkill uses Elo and then uses Bayesian probability to apply that Elo score across different games and formats. At its core though, as that article shows, it uses the Elo equation to establish the persons actual performance. Where it goes further and in similar function hopefully MW:Os matchmaker will go further in the future, is digging into the telemetry of the makeup of the environment in which they player won to effectively predict how he would perform in otherwise completely different environments. For example taking your Elo in MW:O and using it to predict how you would rank in a CoD FFA match.

Ironically that paper goes into how for chess they use a logistical curve for charting k factor and ranking distribution but for anything more complex (like multiplayer teams) you need a Gaussian distribution, as was shown in the development of Glicko.

In the end of section 1, as it discusses why Elo is important and the basis for their calculation, it brings up two critical points, and I quote:

Quote

1. game outcomes often refer to teams of players yet a skill rating for individual players is needed for future matchmaking.

2. More than two players or teams compete such that the game outcome is a permutation of teams of players rather than just a winner and a loser.


It then moves into section two, where it literally presents the Bayesian equation that reflects how one extracts the value of an individual player from the performance of the team.

Where TrueSkill differs from a straight Elo system is the introduction of a Bayesian dynamic factor where it calculates how a players performance varies based on the actual ranking and variables of the team they play with and against. Again, let me clarify that I think something akin to TrueSkill would be freaking brilliant for MW:O. We don't have the player base to exploit all the detail it would bring but it's pretty amazing.

MW:O does *not* use a strict Elo system any more than TrueSkill does. It uses Elo as a basis for its calculation and valuation, then applies a k factor (a modifier to your score) based on relative performance against other players. It also attempts to account for mech weights and the like.

TrueSkill is a matchmaker, not just an Elo system. It uses Elo in its matchmaking - just like the MW:O matchmaker does. Is there room for improvement in the MW:O matchmaker? Absolutely.

Primarily we can benefit from the same changes that TrueSkill and Glicko made (if they haven't been made already. Remember, we're speculating about a system we don't have access to).

1. Gaussian distribution instead of logical one.

2. Isolate the players Elo score as finely as possible - at the very least split pug and premade. Someday having an Elo by chassis, even loadout would be incredible. Modifiers to your relative Elo score based on how you've performed with specific weapons and factors. For example if you do well with an AC20 or with JJs it'll alter your Elo by chassis when you've got those weapons. If you suck on ice with pulse lasers it'll do the same.

3. Match players in a range, not high/low to target. As you get into larger player groups on a team (not FFA) the relative impact of higher Elo declines while lower Elo increases. The same multiplication of force that creates roflstomps as soon as one side as a 3 or 4 mech advantage is exacerbated when you've got 3 or 4 outmatched people on one team.

Currently however we seem to lack the population density to take advantage of change 2 - you're lucky to get roughly weight-balanced matches within your approximate Elo score so refining your Elo score to be more precise isn't going to change much. #1 however will make #2 easier and #3 will make the increase in outliers caused by #1 and #2 less significant.

What's critical to understand here is that at the core of it the only factors that matter at the core of establishing your base ranking in the matchmaker is win/loss. The Elo equation is how that's represented, which is then modified (as it is in MW:O) based on the environment it's being used in. At no point is damage, kills, etc. going to be useful, valuable or intelligently applicable in a matchmaker.

If you want to debate the validity of Elo as the basis for a matchmaker in MW:O you need to show, mathematically, why. Not just throw out some soundbite. Given that there isn't a way to debate that you have no choice but to use ad hominem and argument from incredulity, also called 'the divine fallacy'. "I cannot imagine how this could be true, therefore it must be false."

That's not an argument. Trying to couch it in insulting witticisms doesn't change that.

I've presented you links, quotes and excerpts showing the basis of my point. At this point the burden of proof for you to prove that Elo isn't actually a valid basis for a matchmaker in Elo is going to involve you disproving research that has been done by some of the best professional statisticians in the industry, then peer reviewed and affirmed by some of the most respected experts on the topic and the premier universities that teach it.

Best of luck with that. It's your chance to be famous!

#383 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 25 January 2014 - 10:17 AM

Wow you wrote such a long post, that you must be correct!!

Professional statisticians, that have nothing to do with sports!!

I notice Statistician has the word STAT in it...lol STATISTICS!!!! you know, the thing your so highly against?

Statistician = Theoretical or applied. Does not really mean proven bud.

The MIT guy i would vote for, is the one who worked for the Oakland A's.

"Bill James defined sabermetrics as "the search for objective knowledge about baseball." Thus, sabermetrics attempts to answer objective questions about baseball, such as "which player on the Red Sox contributed the most to the team's offense?" or "How many home runs will Ken Griffy hit next year?"



If only we get one of these guys in the sports industry to write a program designed for fps team deathmatch type games, that will rate or rank people over a short time period based on stats and wins.

"While playing for the Blatimore Orioles in the early 1970s, Davey Johnson used an IBM System/360 at team owner Jerry Hofberger's brewery to write a FORTRAN baseball computer simulation, and using the results unsuccessfully proposed to manager Earl Weaver that he should bat second in the lineup. He wrote IBM BASIC programs to help him manage the Tidewater Tides, and after becoming manager of the New York Mets in 1984 arranged for a team employee to write a dBASE II application to store opposing teams' statistics.[5]"



I never knew Davey Johnson was a computer nerd! The other nerd on the 1986 mets, was catcher Gary Carter. The dude was a computer himself. If players had questions about their performance, or another team. they asked Gary. One of the best baseball teams in history.

Edited by RichAC, 25 January 2014 - 10:36 AM.


#384 Sarsaparilla Kid

    Member

  • PipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 664 posts
  • LocationGold Country

Posted 25 January 2014 - 10:33 AM

View PostRichAC, on 25 January 2014 - 09:02 AM, said:


MWO is a sport! Its not the roleplaying game you wanted. get over it.

Most RPG games are losing money, and cost way too much to develop. Which is probably why PGI went the arcade arena shooter route. Sorry that ruins it for all you long time fans of the Genre. But there is no reason to selfishly drive off newcomers, who do enjoy sporstmanship, by encouraging PGI to sabotage their own game. :P



A bit off-topic, but it's sad that even RPGs aren't what they used to be. Most of the RPG games used to be filled primarily with role-players that wanted to immerse themselves in a different world and pretend to be a different person. Towards the end of my career in Everquest 2, on the only remaining RP server no less, I had people in my group yelling at the tank to stop role-playing because it was getting on their nerves and they were just there to run the dungeon as fast as possible to get phat lewt... :( I was enjoying the role-play myself, but apparently, RP is a dying art among the current generation of gamers, even in MMORPGs like Everquest, WoW and Rift! They've all became competitions to be the first to kill dungeon boss X and get phat lewt item Y to wear as proof of being ub3r l33t.

All previous versions of Mechwarrior had an immersive solo campaign that you could do on your own in addition to the multiplayer module, which gave it more of an RPG feel. Even in the multiplayer chat, we immersed ourselves into the roles of a military unit. Perhaps Community Warfare will help bring out the RPG aspects that the MW franchise has been known for in the past. Unfortunately, those that seek to play the game only as a backdrop for a sports arena or for all that glitters will generally not be the ones that will stick with MWO when the next shiny new combat game comes out.

#385 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 25 January 2014 - 10:40 AM

View PostSarsaparilla Kid, on 25 January 2014 - 10:33 AM, said:


A bit off-topic, but it's sad that even RPGs aren't what they used to be. Most of the RPG games used to be filled primarily with role-players that wanted to immerse themselves in a different world and pretend to be a different person. Towards the end of my career in Everquest 2, on the only remaining RP server no less, I had people in my group yelling at the tank to stop role-playing because it was getting on their nerves and they were just there to run the dungeon as fast as possible to get phat lewt... :P I was enjoying the role-play myself, but apparently, RP is a dying art among the current generation of gamers, even in MMORPGs like Everquest, WoW and Rift! They've all became competitions to be the first to kill dungeon boss X and get phat lewt item Y to wear as proof of being ub3r l33t.

All previous versions of Mechwarrior had an immersive solo campaign that you could do on your own in addition to the multiplayer module, which gave it more of an RPG feel. Even in the multiplayer chat, we immersed ourselves into the roles of a military unit. Perhaps Community Warfare will help bring out the RPG aspects that the MW franchise has been known for in the past. Unfortunately, those that seek to play the game only as a backdrop for a sports arena or for all that glitters will generally not be the ones that will stick with MWO when the next shiny new combat game comes out.


I hear ya dude. One of the main issues with pc gaming nowadays, that has been getting worse for years. Is people just want to exploit the games flaws, for mostly phat loot as you described, but also for many reasons.

Noone wants to play the way the developers intended them to play. Thats what it really boils down to imo.

All games are also more dumb'd down as well. Noone wants to take the time to practice a game. This game keeps getting dumb'd down, but it still has a higher learning curve then most.

And its not so much just in RPG's, its the same story even in arena shooters. There are no more games like quakelive, where the skill gap is almost as great as in pro athletic sports, literally.

But i understand what your saying, because When I play this game and people get mad me for trying to win, or because i put my Atlas on a base, I feel the same way you feel about wanting to roleplay in an rpg. Unless its PvP, thats exactly what your supposed to do.

Unfortunately PGI didn't make an RPG or singleplayer, they probably didn't have the money. And after all it would just get pirated by hackers anyway and they wouldn't make any money. This is a sad truth of the times.

I worry about CW because this community already refuses to play 12 mans, because they don't like losing. Ironically thats why there is not enough teams in the first place, which would actually make it more fun! Yet at the same time they hypocritically claim they want to just play for fun. Unfortunately I don't see why CW would be any different for anybody.

Edited by RichAC, 25 January 2014 - 11:05 AM.


#386 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 25 January 2014 - 10:49 AM

Rich? Baseball is a pretty unique sport. You can have a pitcher throw a shutout game and only have 1 person on 1 team win 1/2 the game. It's literally more rigid and structured than chess. It has teams but each player effectively only does one thing. Most of them don't even hit.

You're creating a false equivalency to go with your ad hominem.

I linked a peer reviewed article discussing, in detail with formula, why Elo works for games like MW:O. You're talking about why people like tracking different statistics in baseball and saying that because people like tracking statistics in baseball, Elo doesn't work for MW:O. To say that it's an irrational point is, to put it gently, an understatement. Then you attempt to say that sine MW:O is like a sport, it's like baseball.

Soccer, football and several other pro sports do use Elo. You know who else uses Elo?

Matchmakers in the gaming industry. Do you know what the most popular, successful and well developed matchmaker for multiplayer games is? TrueSkill. Know what it's based on?

That's right. Elo.

Just for giggles Rich, just for you, here.

It's an Elo rater for baseball players. Play around with it, by all accounts it's actually pretty accurate.

#387 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 25 January 2014 - 11:10 AM

View PostMischiefSC, on 25 January 2014 - 10:49 AM, said:

Rich? Baseball is a pretty unique sport. You can have a pitcher throw a shutout game and only have 1 person on 1 team win 1/2 the game. It's literally more rigid and structured than chess. It has teams but each player effectively only does one thing. Most of them don't even hit.

You're creating a false equivalency to go with your ad hominem.

I linked a peer reviewed article discussing, in detail with formula, why Elo works for games like MW:O. You're talking about why people like tracking different statistics in baseball and saying that because people like tracking statistics in baseball, Elo doesn't work for MW:O. To say that it's an irrational point is, to put it gently, an understatement. Then you attempt to say that sine MW:O is like a sport, it's like baseball.

Soccer, football and several other pro sports do use Elo. You know who else uses Elo?

Matchmakers in the gaming industry. Do you know what the most popular, successful and well developed matchmaker for multiplayer games is? TrueSkill. Know what it's based on?

That's right. Elo.

Just for giggles Rich, just for you, here.

It's an Elo rater for baseball players. Play around with it, by all accounts it's actually pretty accurate.


Why are linking me something from some 3rd party site or Fan, that has nothing to do with how MLB works? Just like you linked me to QL ranks which has nothing to do with how ID ranks players in their team games. These are amateurs my friend. These are not the "professionals" in the said industry. I don't care if they guy was tutored by Einstein. Hes just a hobby.

First, all those sports have static teams, Which is why many posters have told you ELO only works in leagues and tournaments first of all. Not random pugs. What is hard to understand about that?

Second, baseball is still a team game. I don't even know what your getting at here. Did you forget about the catchers? Also A-rod is worth more money then any pitcher. Home runs pay more then shutout innings.

Third, Managers and players and Fans, all take stats into account. Its what makes all those games worth money. Not just wins. Its why fans still follow losing teams.

The reason you need to add STATS, or Match score into the equation, especially in MWO.....Is because people care about stats more then they do wins, which you yourself even admitted. MWO is losing players, which means they are losing money, so to just ignorantly rank people up by wins, when people care about stats, you will never satisfy a playerbase who thinks the match maker or ELO system is not working right, and you are hurting the game.

No matter how mistaken you think they are. What you are not understanding is W/L ratios are not the only issue here my friend. Or else noone would be complaining because I'm willing to bet most peoples w/l are not that bad. I hope you soon realize this.

1. its a random team game.

2. Its a game based on cbills.

Edited by RichAC, 25 January 2014 - 11:22 AM.


#388 Nightfire

    Member

  • PipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 226 posts
  • LocationAustralia

Posted 25 January 2014 - 11:52 AM

View PostRichAC, on 25 January 2014 - 09:02 AM, said:

MWO is a sport! Its not the roleplaying game you wanted. get over it. Most of the mentalities here is why the game is getting even less popular then it already is.


Let's have a look at this shall we? You're talking about telling comments, how about we really look at what you're saying.
  • Sorry, MWO isn't a sport. Every definition in the context you are looking for involves physical exertion.
  • Not the role-playing game I wanted huh? Oh, the double whammy! Here we have the shaming language, this is where ideologues run when they have no real argument. This also is an attempt to pigeon hole me in another box since the last one didn't take. RPG ... I'm guessing nerd / geek stereotype.
  • Get over it huh? Is that something like "Man up!"? More shaming language.
  • Now blame everyone else who disagrees with him. I'd even go so far to call this "splitting" in psych circles.


Quote

I don't need to quote any of your further comments. This thread is getting spammy now.


Got it. I called you on your evasive techniques and so you're going to bail but in a way where you don't have to admit you were wrong and blame me and everyone else for it.

Quote

You've already explained it all in your previous post when you said you hate sports games and your glad they are not on the pc anymore. ....WHich is basic proof of how pc's are unpopular nowadays. Most of society likes sports I'm sorry. Nerds who don't are still a minority.


Oh I like this, this is insult is a little more nuanced. I hate sports games (which remember, MWO is in RichAC's opinion.) and I'm glad they are not on the PC now (I am!) but RichAC -likes- sports games! So now I am evidence of why PC's are unpopular because real men like sports. You know, most of society! The point of this convoluted insult is to state that I am not a normal part of society and that I belong to the class of people directly responsible for the downfall of PC gaming. See how that is both splitting and shaming language? Brilliant!

Quote

Playing tiger woods 2008 with players from around the world was very sociable. And I keep explaining, it was a golf game that made the pc more popular for sports games over the atari and intellivision, in 1986 mainly Because of the graphics. But also all the other abilities, like mapmaker, mods, etc. Many sports games had community sites that updated rosters and player attributes, stadiums, player models, etc... Graphics are still better on the PC. And PC's are still better for multiplayers and online communities. You can follow the evolution of graphics on the pc through Golf games on the computer up until 2008.


What!? Something of actual substance!? I better address that then!
If you really want to get into the popularity of gaming culture it is actually a quite complex and nuanced discussion but I'll touch on the relevant highlights for you.
  • Sports gaming (that is, so we are clear, video games that emulate a real sport) has long been favored by ... Sports Fans! in the majority.
  • There has long been a balance, a tipping point if you will of computing resources vs will to use. PCs always had the resources early on and the ease of development. PC Sports gaming was how it was because you had a cross section of people who both liked sports and knew how to use a computer at least at the basic level.
  • When consoles came to a point where graphics and performance hit said tipping point, it really didn't matter how much better it got on the PC. See, the demographic changed! Most players of said games didn't and really didn't want to learn how to use and maintain a PC. They wanted, and got, a turn key device that allowed them access to their games. The majority of these players would never be swayed to play on a PC because it's "too complicated to just have fun!" These new players we kind of like the basic PC users who wouldn't touch Linux. Their money was just as good though and there was more of them.
  • PC's may be better for online communities but the majority of the players of Sports Games are never going to convert so ... what are you going to do?
This may not mesh with the narrative you've told yourself but hey, reality is often duller than almost anything else. Often ...


Quote

Most Sports games are no longer on the pc, mainly because of piraters, hackers and cheaters. Soccer being the exception since its mostly Americans abandoning the internet nowadays. But its a fact most gamers on the pc nowadays, feel like you. Its basically only strategy games and roleplaying games that are popular on pc nowadays. People don't want to deal with cheaters and technical problems. Most people left on the pc are computer nerds or work in the industry, and like you, they don't like sports.


See, you focus on this epidemic of cheaters, and pirates and hackers and ... whatever else you want to blame for your gaming world not being as good as your nostalgia remembers. RTS games are pretty popular (you can't argue against Starcraft and C&C) but RPGs popular on the PC? I don't think so! I mean you can point to a few tall poppies like WoW (it's not really an RPG but most will call it such) but no, there are many other popular genres.

And you were doing so well but now you delve back into the shaming insults.

Quote

Most RPG games are losing money, and cost way too much to develop. Which is probably why PGI went the arcade arena shooter route. Sorry that ruins it for all you long time fans of the Genre. But there is no reason to selfishly drive off newcomers, who do enjoy sporstmanship, by encouraging PGI to sabotage their own game. :P


Ok, I don't like sports games but you obviously have something about RPGs. BTW, RPGs are NOT the most expensive games to develop. AAA titles in various areas cost far more than they probably should and I'd put MMO's right up there in terms of development costs but then again, MMOs are designed to have a far extended revenue generation life. Games like MWO draw players by virtue of their Intellectual Property. If you don't implement the flavor of the IP, then you don't draw the die hard fans. If you lose those, your game has lost its champions. Every game needs something beyond core mechanics or we'd all be still playing pong. Sorry if that ruins it for you but if you don't like the IP of a particular game, perhaps that game just isn't for you?

So, since you are addressing me you're going to have to be a little more specific on how I have recommended PGI drive off newcomers and how do I not engage in sportsmanship (which is the implied shaming language there) which the newcomers obviously do.

Quote

Again, basing the ranking/rating system on wins alone will drive more players away when:

1. its a random team game

2. the game is based on cbills.



I actually sort of agree, unfortunately.
  • I believe there are better suited algorithms and the current implementation of Elo is not tied to any metric directly impacted by the player.
  • I also believe the current implementation of Elo / Matchamker, if it ever works as intended, requires a sample size of matches so large that (on average) no casual player will ever stay long enough to get there.
So, now you have demonstrated Straw-manning, quote mining, shaming language, splitting, vilifying and general insults.
Would you actually care to discuss something with intellectual honesty and rigor?

#389 Nightfire

    Member

  • PipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 226 posts
  • LocationAustralia

Posted 25 January 2014 - 12:00 PM

View PostMischiefSC, on 25 January 2014 - 09:50 AM, said:

I quote:


It then moves into section two, where it literally presents the Bayesian equation that reflects how one extracts the value of an individual player from the performance of the team.

Where TrueSkill differs from a straight Elo system is the introduction of a Bayesian dynamic factor where it calculates how a players performance varies based on the actual ranking and variables of the team they play with and against. Again, let me clarify that I think something akin to TrueSkill would be freaking brilliant for MW:O. We don't have the player base to exploit all the detail it would bring but it's pretty amazing.

MW:O does *not* use a strict Elo system any more than TrueSkill does. It uses Elo as a basis for its calculation and valuation, then applies a k factor (a modifier to your score) based on relative performance against other players. It also attempts to account for mech weights and the like.

TrueSkill is a matchmaker, not just an Elo system. It uses Elo in its matchmaking - just like the MW:O matchmaker does. Is there room for improvement in the MW:O matchmaker? Absolutely.

Primarily we can benefit from the same changes that TrueSkill and Glicko made (if they haven't been made already. Remember, we're speculating about a system we don't have access to).

The scary thing for me here is, I can actually agree with this.

#390 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 25 January 2014 - 12:20 PM

View PostNightfire, on 25 January 2014 - 11:52 AM, said:


Let's have a look at this shall we? You're talking about telling comments, how about we really look at what you're saying.
  • Sorry, MWO isn't a sport. Every definition in the context you are looking for involves physical exertion.
  • Not the role-playing game I wanted huh? Oh, the double whammy! Here we have the shaming language, this is where ideologues run when they have no real argument. This also is an attempt to pigeon hole me in another box since the last one didn't take. RPG ... I'm guessing nerd / geek stereotype.
  • Get over it huh? Is that something like "Man up!"? More shaming language.
Now blame everyone else who disagrees with him. I'd even go so far to call this "splitting" in psych circles.


No reason to go past this first quote. tldr.

You don't respect MWO like you would an athletic sport. This is whats telling. This is all I need to know about you. This is what has always held back pc multiplayer games. and e-sports. To me they are not that different. And I treat the game with the same respect I would any sport. It doesn't make a diff to me that it is a video game. Its not any less mature, Its not any less competitive. Its not any less sporting then playing chess or football.

LoL is the first e-sport on tv, in over 30 years of gaming, not only because strategy games attract more of the brain over brawn types, where aim doesn't matter as much and is less exploitable. (even though closer to athletic sports imo) But because Koreans are more honorable and more respectful to their game. There is a good sense of sportsmanship in that community.


To keep it on topic, ranking/rating and matching players based only on wins, when players care just as much about other stats like K/D And dmg done is not going to quell any complaints.

1. its game with random teams

2. its a game based on cbills

Edited by RichAC, 25 January 2014 - 12:30 PM.


#391 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 25 January 2014 - 12:33 PM

View PostNightfire, on 25 January 2014 - 12:00 PM, said:

The scary thing for me here is, I can actually agree with this.


Where people get caught up is in basing off an Elo system. You have to; as the peer review article from MIT puts it you must have a pure variable from which to base matchmaking or you introduce either a duplicative dynamic (something reflected twice, like if you made a score based off someones damage PLUS match score when match score is already modified by damage) or a non-corollary dynamic (including kills when how many kills someone gets doesn't actually relate to how likely their team is to win - kill stealing, etc) ergo win/loss is the only trustworthy basis for a matchmaker.

Nerd-words aside, this means that if you try to develop a matchmaker based off anything BUT win/loss it's inherently unreliable. Just win/loss alone however isn't precise enough.

So you include the stuff that is associated with their performance. In MW:O this would be things like what weapons you use, what your chassis is. Are you premade or pug, who are you dropping with. What sort of team did you drop with. What was the other teams makeup of people, mechs and loadouts. What map, what mode.

These are not direct metrics that can be gamed but are instead the CAUSE of those metrics. You play better with PPCs and JJs, so you win more. What's so critical is that it doesn't matter how you win. More damage, more kills, distraction, whatever. Those metrics are symptomatic, not causal or even truly representative. For you, specifically, some things will help you drive a victory for your team better than others.

So *that* is what you modify Elo with. The actual factors, the STUFF that actually helps you drive the RESULT. What mech, weapon, loadout, team design, what works best for YOU to drive a WIN.

You get paid off of score because, well, it's immediate. Win/loss may modify how much you get paid but you get paid off symptoms and byproducts of your performance like damage, kills, assists, spotting. It's not that they don't have value, they're very valuable. They are however not reliable nor are they truly representative of your complete contribution.

A straight Elo system wouldn't work for MW:O. Which is fine because MW:O doesn't use a straight Elo system. It does intelligently use Elo in the matchmaker along with other factors.

What needs to be discussed is what other factors are good to use and how to use them and how much weight they deserve.

Tonnage is, of course, a great example but is it truly representative and reliable? I would ten times rather face Wispy in an 80 ton Awesome than in a 35 ton Jenner. That's not the case for everyone though. So while tonnage is important, what's more important is how that specific player manages that tonnage.

BV is one I see thrown around but it's also neither representative or reliable. The Battle Value for an LB10X in tabletop is way higher than an AC10. Same with ERPPCs vs PC, ERLL vs LL, etc. XLs worth way more than standard engines. It just doesn't translate anything like realistically.

What we really need and fortunately the nature of the system (server side recording of performance) is to modify Elo by the player and by the match based on what they're bringing and how they handle it.

The bigger problem then becomes dropship mode. How can you effectively rank players for matches when after you've got both teams picked people can change up what they're bringing, what map and what mode?

You can't. So that brings us full circle to a sort of 'general' Elo score for the player. What the specific ranking will do is after you drop into the match if suddenly one team ends up with a score advantage over the other it will change what the rewards are for winning for both teams. The team with the higher score will get less for winning and lose more for losing.

#392 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 25 January 2014 - 12:51 PM

@Mischief,

How in the heck are you supposed to determine who is better with what weapons? Having a separate ELO for each weapon? totally ridiculous and not needed. Your going into left field buddy. is your goal to try and make it as complicated and impractical as possible?

People already know they are going to suck more with certain weights or weapons, depending on the player, their competition because of it shouldn't change. If they keep practicing with builds or mechs they suck with, it will just automatically get avgd in the end result.

What people are going to find out with CW and weightlimits. IS that its not going to be much different. And they still will be crying.

Again your worried about gaming the system. This is the whole basis for your whole argument. This is the only thread you have left to cling to. This is your last argument to stand imo. And I already told you this is a free to play game. So NO matter what you do they can just go make a new account.

So you will always have those players that don't spend a dime on the game, that could care less, and will be able to game ANY system no matter how complicated.

Noone is saying to base it on stats only, wins are still part of the equation.

Now that being said, I doubt most people are going to game themselves out of cbills, to be placed in a bracket they can't even see or determine...hahaha

People already try to get as much dmg as they can without winning. Noone cares about winning until they get rewarded for doing so. Don't you realize this?

Edited by RichAC, 25 January 2014 - 12:55 PM.


#393 Abivard

    Member

  • PipPipPipPipPipPipPipPip
  • Shredder
  • 1,935 posts
  • LocationFree Rasalhague Republic

Posted 25 January 2014 - 01:00 PM

Mischief,you keep posting stuff that refutes your position, but yet you stick by what you THINK they are saying instead of what they are really saying.

Elo is not magic,math is not magic, it doesnt have some little gremlin that accounts for all these unknowns and filters them out with out anyone's help.

If we keep it simple, it flys over your head, if we go into detail you straw argument it away.

Are you deliberately being obtuse?

Do you actually read what you link to?

#394 Abivard

    Member

  • PipPipPipPipPipPipPipPip
  • Shredder
  • 1,935 posts
  • LocationFree Rasalhague Republic

Posted 25 January 2014 - 01:11 PM

Mischief.how many times will you present evidence that Elo alone can not and will not ever be able to correctly create a players ranking by itself in a multi player team environment before you have that epiphany?

Walls of text do not prove points.

All of these systems you mention are attempts to rate players in multi player team based games BECAUSE Elo by itself, as it is, will not do it!

But you seem to think you are the only person in the whole world that understands the trick to making straight Elo work at anything in the universe! lol

#395 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 25 January 2014 - 01:14 PM

View PostAbivard, on 25 January 2014 - 01:00 PM, said:

Mischief,you keep posting stuff that refutes your position, but yet you stick by what you THINK they are saying instead of what they are really saying.

Elo is not magic,math is not magic, it doesnt have some little gremlin that accounts for all these unknowns and filters them out with out anyone's help.

If we keep it simple, it flys over your head, if we go into detail you straw argument it away.

Are you deliberately being obtuse?

Do you actually read what you link to?


So, let me sum up your response with a quote from what I said to you earlier:

Quote

You're attempting to represent little witticisms and soundbites as actually having some technical basis or value. Why don't you put up a mathematical model representing how the Elo formula can not work. Show me the complexity breakdown where suddenly statistical modeling ceases to function when you get into 12 v 12 representations. Show some sort of actual useful data that clarifies that somehow you can not solve for one factors contribution in a roughly weight and skill matched 12 v 12 environment over sufficient sample sizes.


You say I refute my own position, show me where. Show me where you've actually given a simple or detailed argument that actually refutes anything in fact, other than witticisms, ad hominems and the like. You say Elo doesn't work, but don't say why or how other than 'it just doesn't work because it's 12 v 12'. I showed you a link to a peer reviewed study from MIT talking about not just the fact that it works but how it works with the formula to back it up.

#396 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 25 January 2014 - 01:21 PM

View PostAbivard, on 25 January 2014 - 01:11 PM, said:

Mischief.how many times will you present evidence that Elo alone can not and will not ever be able to correctly create a players ranking by itself in a multi player team environment before you have that epiphany?

Walls of text do not prove points.

All of these systems you mention are attempts to rate players in multi player team based games BECAUSE Elo by itself, as it is, will not do it!

But you seem to think you are the only person in the whole world that understands the trick to making straight Elo work at anything in the universe! lol


Neither MW:O nor any place I've linked you uses a strict Elo system. Nobody ever said they did. It is however the basis, just like it is for MW:O. It all uses win/loss as the basis for the ranking system, modified by things like who was on your team, who was on the other team, what sort of game mode it was.

Never damage, score, kills, anything like that.

Which I've said. Repeatedly. Yes, Elo is the only viable system to use for MW:O, or any similar game for that matter as the core scoring component of the matchmaker. That Elo is then modified by a k factor, it's also matched for tonnage. Ideally in the long run your Elo score will be modified by what loadout and chassis you're using based on how you've performed in terms of win/loss with that loadout and chassis before which is, again, using the Elo system.

#397 Abivard

    Member

  • PipPipPipPipPipPipPipPip
  • Shredder
  • 1,935 posts
  • LocationFree Rasalhague Republic

Posted 25 January 2014 - 01:25 PM

Again not a single person has said Elo does not work for what it was designed for.

We are saying it doesn't apply here.

Your MIT says it doesn't work in this environment, every source you have linked to or referenced has said it will not work in this type of environment.

No matter what anyone else says, you refuse to comprehend.

#398 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 25 January 2014 - 01:37 PM

View PostAbivard, on 25 January 2014 - 01:25 PM, said:

Again not a single person has said Elo does not work for what it was designed for.

We are saying it doesn't apply here.

Your MIT says it doesn't work in this environment, every source you have linked to or referenced has said it will not work in this type of environment.

No matter what anyone else says, you refuse to comprehend.


No. Read the link. It says Elo works great - that's why the Elo equation is used in TrueSkill. It then uses modifiers to apply the results of it across multiple games.

TrueSkill uses the Elo equation in the exact same manner that the MW:O matchmaker does. To establish the players probability of winning against a set opponent based off their win/loss record as a member of other teams in other games and matches. It then modifies that ranking via a Bayesian equation to convey it to other games and other team sizes.

#399 Abivard

    Member

  • PipPipPipPipPipPipPipPip
  • Shredder
  • 1,935 posts
  • LocationFree Rasalhague Republic

Posted 25 January 2014 - 01:44 PM

View PostMischiefSC, on 25 January 2014 - 01:37 PM, said:


No. Read the link. It says Elo works great - that's why the Elo equation is used in TrueSkill. It then uses modifiers to apply the results of it across multiple games.

TrueSkill uses the Elo equation in the exact same manner that the MW:O matchmaker does. To establish the players probability of winning against a set opponent based off their win/loss record as a member of other teams in other games and matches. It then modifies that ranking via a Bayesian equation to convey it to other games and other team sizes.



OMFG! are you for real! Trueskill is not Elo under a different name, My sports car is an automobile, it runs on wheels, The fact that it uses wheels does not mean it is a rickshaw, even if the rickshaw came first.

If you try to race that rickshaw at the indy 500 what do you think would happen?

#400 Roland

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 8,260 posts

Posted 25 January 2014 - 01:45 PM

That second part of applying additional modifications is pretty key though.

However, that is besides the point, since trueskill would likely also fail to perform good matchmaking in this game, due to the significant differences in capabilities stemming from the equipment rather than the player.

That is, you could take 24 pilots with identical skill, put half of them all on a team of hunchbacks, and half on a team of highlanders, and that game would not be even remotely close, even if elo suggested they each had equal chance of winning.

A market based value system could address this fairly easily, in a way that would effectively self balance.





17 user(s) are reading this topic

0 members, 17 guests, 0 anonymous users