50 replies to this topic

[Modo44]

IraqiWalker, on 20 September 2014 - 01:37 AM, said:

Not to mention that K/D isn't a good description of your contribution in a game. Nope, assists, ECM coverage, TAGing, NARCing, Spotting covering wtih AMS ... etc. are better descriptors of your contribution to your team, instead of individual glory hounding.

It goes further. No value set represents your abilities fully. As weak as Elo is, it actually provides the best approximation.

[IraqiWalker]

Modo44, on 20 September 2014 - 02:33 AM, said:

It goes further. No value set represents your abilities fully. As weak as Elo is, it actually provides the best approximation.

Sad but true. The biggest problem is that our player base is small enough that we need huge buckets, and that causes problems for the MM. Even with close averages where the difference between the 2 teams is only 10 points. That could mean a god-like player and a new/bad player on the same team.

[Appogee]

IraqiWalker, on 20 September 2014 - 01:37 AM, said:

queue even more rushing to get kills, and even stupider mistakes

Which is EXACTLY what I experienced all yesterday PUGging, as I tried to contrive team wins, only to have selfish or clueless players in equal number simply refuse to fight together as a team within a cohesive strategy.

But at least an Elo based on individual performance would start to reflect some kind of individual skills level based on contribution, instead of the current one which stupidly creates your Elo based on whether you happen to mostly be in a team which derps less than the other.

IraqiWalker, on 20 September 2014 - 01:37 AM, said:

Not to mention that K/D isn't a good description of your contribution in a game. Nope, assists, ECM coverage, TAGing, NARCing, Spotting covering wtih AMS ... etc. are better descriptors of your contribution to your team,

Of course it's not.

But it's a HELL of a lot better indicator than "whether my team won" regardless of how well I played or how badly other players derped.

I'd be delighted if PGI wanted to create a more advanced Elo formula which includes effective support. I proposed a simpler solution because PGI only ever implements the simplest and easiest solutions.

Modo44, on 20 September 2014 - 02:33 AM, said:

It goes further. No value set represents your abilities fully. As weak as Elo is, it actually provides the best approximation.

No, for many players, it approximates how well their regular team plays, not them as an individual.

You know this. You've seen this.

Edited by Appogee, 20 September 2014 - 03:16 AM.

[Modo44]

Appogee, on 20 September 2014 - 03:14 AM, said:

No, for many players, it approximates how well their regular team plays, not them as an individual.

To put it precisely, Elo approximates how hard you personally can carry. It is a better win potential approximation than literally any other value set. You can go show your KDR e-peen elsewhere, it will not help build better matches.

[Appogee]

Modo44, on 20 September 2014 - 03:16 AM, said:

It is a better win potential approximation than literally any other value set. You can go show your KDR e-peen elsewhere, it will not help build better matches.

Stop trying to portray me as being personally epeening about k/d. That is not the case at all, and it's personally insulting, given that I'm the guy who will always die to try to create a team victory.

I am making the case that k/d + average damage per match + assists and other personal contributions are a far better indcator of an individual player's abilities than how often they are on a team which wins or loses.

Yes, a better player can carry a poor team more often than a less good player can. But on the flipside, it also makes no sense that the players being carried have their Elos inflated because they happened to be lucky enough to have some good players on their team.

This cuts both ways.

Edited by Appogee, 20 September 2014 - 03:25 AM.

[Modo44]

Appogee, on 20 September 2014 - 03:23 AM, said:

Stop trying to portray me as being personally epeening about k/d. That is not the case at all, and it's personally insulting.

I am making the case that k/d + average damage per match + assists and other personal contributions say far more about a player's abilities than how often their team wins or loses.

Except they do not. That Atlas that died "for free", bringing the entire enemy team out of position. That light mech with 0 damage, 0 assists, 0 kills, who won the game by living and capping. Those are just the most blatantly obvious examples that would remain completely ignored by your idiotic idea.

[Appogee]

Modo44, on 20 September 2014 - 03:16 AM, said:

it will not help build better matches.

PUG matches with players of more equal ability would be far better than what we have at them moment, a random crapshoot about whether personal Elo will be 'balanced' by having 9 noobs assigned to your team.

Modo44, on 20 September 2014 - 03:25 AM, said:

That Atlas that died "for free", bringing the entire enemy team out of position. That light mech with 0 damage, 0 assists, 0 kills, who won the game by living and capping. Those are just the most blatantly obvious examples that would remain completely ignored by your idiotic idea.

And yet, the current idiocy of assigning a personal Elo based on a collection of team scores apparently escapes you. Go figure.

Add capping assists to my formula. That would strengthen it, not invalidate it.

Edited by Appogee, 20 September 2014 - 03:35 AM.

[DeathlyEyes]

IraqiWalker, on 20 September 2014 - 01:37 AM, said:

yes, because we don't already have a lot of new players thrown into the shark pit with 3000+ games played guys like myself. You want a tier system. Kind of like the ones in LoL, StarcraftII ... etc. Where Bronze players are matched against only bronze players (maybe low silver with high bronze) ... etc. No Diamond players matched against bronze players. That can help with the MM a bit.

However, we have too many modes for the current player base. If this game had a player base of over 1 million, maybe. Right now, we have almost 600,000 registered members, and we know for a fact not all of them are still playing this game.


queue even more rushing to get kills, and even stupider mistakes made in the name of trying to bump Elo up. Not to mention that K/D isn't a good description of your contribution in a game. Nope, assists, ECM coverage, TAGing, NARCing, Spotting covering wtih AMS ... etc. are better descriptors of your contribution to your team, instead of individual glory hounding.

We have new players thrown into high ELO games because the matchmaker is compensating for High ELOs. It sees a game and either the ELO of one team is way low or high and it throws missmatched players in to compensate. Sadly I do not count for 6 trial mechs when playing against a team of average to above average players.

If you remove ELO the chances of that happening go down. Sure every now and then a new player is thrusted into a game with high elo players. We know this game doesn't have a large base of new players. Rus said this isn't who they are targeting with the game. Removing ELO removes a system that is actually pushing them into high ELO games.

If ELO is to stay in I would want it based off Damage per match.

Edited by DeathlyEyes, 20 September 2014 - 03:52 AM.

[IraqiWalker]

Modo44, on 20 September 2014 - 03:25 AM, said:

Appogee, on 20 September 2014 - 03:34 AM, said:

both of you need to cool off a but. I've interacted with appogee over the forums a lot, and in-game a few times as well, and I can safely say he's not the guy that flaunts his e-peen.

Modo is also one of the sharper analytical players I've come across.

Both of you have fair points about your side of the argument. How about instead of yelling at each other, just simplify what you are proposing for your side of the argument, and try and work together on finding a middle ground/optimal solution?

DeathlyEyes, on 20 September 2014 - 03:48 AM, said:

If you remove ELO the chances of that happening go down.

This part is wrong worse. If you remove elo you have no matching and it becomes even more random than now.

Also, what are you replacing it with? If you don't replace it with anything, the game becomes a bigger Cr**shoot than it already is.

DeathlyEyes, on 20 September 2014 - 03:48 AM, said:

If ELO is to stay in I would want it based off Damage per match.

That's also not a good choice, as mentioned above, and discussed in detail on half a dozen threads, that is not a good indication of good play. What about support builds, the guys that actually play AS A TEAM. How about them?

I've seen a raven-3L pull all the stops in a match. ECM covered his allies, TAG'ed and NARC'd all day long, UAVs, Arty strikes and all. He didn't deal much damage because he was busy winning the game for his team. Why should he have a lower elo score than the Atlas that died like an idiot 2 minutes in after dealing only 100 damage (for the record, that Atlas ended up with the same match score as the Raven that practically won the game for his team solo)

Edited by IraqiWalker, 20 September 2014 - 04:05 AM.

[Modo44]

IraqiWalker, on 20 September 2014 - 03:58 AM, said:

Both of you have fair points about your side of the argument.

The argument was repeated a couple of times -- in full -- in Karl Berg's feedback thread. It always shows more obvious, bigger issues with workarounds than with straight up Elo, so when supposedly smart people bring it up, I treat it as deliberate flaming.

[Appogee]

Modo44, on 20 September 2014 - 04:07 AM, said:

The argument was repeated a couple of times -- in full -- in Karl Berg's feedback thread. It always shows more obvious, bigger issues with workarounds than with straight up Elo, so when supposedly smart people bring it up, I treat it as deliberate flaming.

First "flaming" is what you did when you described someone else's idea as "idiotic" and completely mischaracterise their motivations as "e-peen".

Second, "flaming" is different to "trolling" which is what you probably meant.

Third, to state the obvious, not everyone on the forums has read every other thread on the forums, and that's why flaming them when they propose a well-intentioned is not appropriate.

For my part, I will go read the Karl Berg thread. Because I'm interested in advancing the game's crap matchmaking, not in defending myself against people who would rather flame than debate ideas.

Edited by Appogee, 20 September 2014 - 04:38 AM.

[Clit Beastwood]

Appogee, on 20 September 2014 - 01:31 AM, said:

I have an idea...

Replace the current individual Elo, which is calculated on how often their team wins and loses,
with
New Elo, based on personal kill/death ratio and personal average damage per match.

Your individual contribution to a game, in terms of personal kills and damage you cause, is a better indication of your skill, than whether you were or weren't grouped with or against a bunch of randoms.

This is nonsensical. KDR as a metric has little to do with performance as a player. You should try playing with a decen group some time. 12-1 stomp, all members of your team do 350-450 dmg and 1-2 kills. A good group will have a really even spread... The perfect collaborative effort. Weighting KDR more heavily would just further encourage players to "crowd" targets trying for the killshot. In a drop with a good team, everyone just defers to the person closest for the kill shot, etc., because the players realize individual performance does not matter. Strip enemy, move on, lights bat clean up.

[Clit Beastwood]

DeathlyEyes, on 20 September 2014 - 03:48 AM, said:

If ELO is to stay in I would want it based off Damage per match.

That doesn't make sense either- an LRM boat hiding in the back refusing to engage directly gets an elo boost for being a lazy coward.. Drop after drop, elo going up... Now they're used as the high elo player on a team to balance a bunch of noobs, and they're matched against a bunch of midrange players... How's that going to fare? Basing it on wins is, at a macroscopic level, the most sensible average to factor. It's hard to "game" - over the course of 10,000 drops, you can't cheat your WLR... But you can easily fluff damage and kills by meat shielding and kill stealing. I don't want those people on my team. I want the people that are going to stay at my side and provide a unified front.

[Modo44]

This is what I was talking about. Coming up with ever dumber examples, which can go on forever. Until you arrive at this: How do you measure the ability to direct team mates when playing solo. That lance you told to not go away, and won the match. Those people you told where to gather? Elo measures that, literally nothing else does. GG.

Edited by Modo44, 20 September 2014 - 09:58 AM.

[Clit Beastwood]

Modo44, on 20 September 2014 - 09:58 AM, said:

This is what I was talking about. Coming up with ever dumber examples, which can go on forever. Until you arrive at this: How do you measure the ability to direct team mates when playing solo. That lance you told to not go away, and won the match. Those people you told where to gather? Elo measures that, literally nothing else does. GG.

It's a team game, I'd think pugs are secondary to groups.

[Modo44]

Fierostetz, on 20 September 2014 - 10:20 AM, said:

It's a team game, I'd think pugs are secondary to groups.

Yeah, communication. That part nothing but Elo tracks. Quite the thing in group drops, I hear.

[Appogee]

Fierostetz, on 20 September 2014 - 09:11 AM, said:

KDR as a metric has little to do with performance as a player.

Ultimately, after we've all been good team players and done our share of flanking, tanking, spotting etc, ultimately, one of us has to actually kill the enemy. So, as I explained, k/d is usually one contributing indicator of a good player.

Fierostetz, on 20 September 2014 - 09:11 AM, said:

You should try playing with a decen group some time.

I do, almost every day, but nice try with the gratuitous insult, person I've never seen on the battlefield.

Fierostetz, on 20 September 2014 - 09:11 AM, said:

A good group will have a really even spread... all members of your team do 350-450 dmg and 1-2 kills.

Actually, that's probably not a "good team". In a "good team", players have designated roles and it's accepted that the snipers (or LRM boats, if they're part of the drop deck) will get more kills than the Narcers, squirrels etc.

What you described is an "equally capable" team, which is actually what I is exactly what I would like the matchmaker to create, instead of the current way it puts a few good players with 9 noobs and considers that a workable average.

Edited by Appogee, 20 September 2014 - 10:49 AM.

[Appogee]

Modo44, on 20 September 2014 - 09:58 AM, said:

This is what I was talking about. Coming up with ever dumber examples, which can go on forever. Until you arrive at this: How do you measure the ability to direct team mates when playing solo. That lance you told to not go away, and won the match. Those people you told where to gather? Elo measures that, literally nothing else does. GG.

Your attempted use of the perfect solution fallacy is even dumber.

However, I remain genuinely surprised that you are so satisfied with the idea that individual Elo should be determined by team outcomes, that you think it's the best solution available.

Perhaps you never drop solo and experience how flawed the current system is in the solo queue.

Team-based Elo makes sense for large groups playing the group queue. It is less good when assembling smaller groups into one team, where it creates an average of the teams' Elos. And it is woeful and nonsensical when for the solo queue.

Edited by Appogee, 20 September 2014 - 10:52 AM.

[DeathlyEyes]

Fierostetz, on 20 September 2014 - 09:18 AM, said:

That doesn't make sense either- an LRM boat hiding in the back refusing to engage directly gets an elo boost for being a lazy coward.. Drop after drop, elo going up... Now they're used as the high elo player on a team to balance a bunch of noobs, and they're matched against a bunch of midrange players... How's that going to fare? Basing it on wins is, at a macroscopic level, the most sensible average to factor. It's hard to "game" - over the course of 10,000 drops, you can't cheat your WLR... But you can easily fluff damage and kills by meat shielding and kill stealing. I don't want those people on my team. I want the people that are going to stay at my side and provide a unified front.

I guess a total team contribution system is best. Factor in average kills average damage and support in order to come up with some sort of number. Right now though I am sick of being expected to account for a bunch of terribads and make up for them in the grand scheme of things when dropping in a small group. I would much rather just making sure groups of similar sizes drop against each other. Let the better players win and the bad players lose.

[Clit Beastwood]

DeathlyEyes, on 21 September 2014 - 07:04 AM, said:

I guess a total team contribution system is best. Factor in average kills average damage and support in order to come up with some sort of number. Right now though I am sick of being expected to account for a bunch of terribads and make up for them in the grand scheme of things when dropping in a small group. I would much rather just making sure groups of similar sizes drop against each other. Let the better players win and the bad players lose.

I just don't see many options without a drastic increase in the size of the player base.