134 replies to this topic

[Kunae]

Assertion: The Elo system is very flawed, polluted with stats from both grouping and solo, and with their mechanic of balancing high-Elo with a proportion of very low Elo, in matches, to "even things out", it hurts the game more than it helps.

I would suggest that PGI turns off Elo for one week or patch-cycle so everyone can see how it plays. Just use a rough weight-class matching, and let it go at that.

It is my belief that the resultant matches will be the same or better than ones using Elo. I would just like to be able to test this theory.

[Roland]

I have to agree at this point. The current matchmaking certainly isn't able to match based on skill, as I'm consistently in matches with some of the top players, while also having folks who obviously have no idea how to play the game yet running around in trial mechs and stuff.

It would be better to at least get the tonnage even, so when the skill disparity on the two teams is high, the better team at least doesn't have a huge tonnage advantage as well.

A quick point to make here, from a game design perspective:
If one team wins because they are better, but the match itself was even from the perspective of equipment taken, then this is always going to be viewed more favorably by the losing team compared to a game where one team had a large advantage in equipment like mech tonnage.

In the case where skill does not match up, the game was at least fair. It's like playing basketball against a pro.. You expect to lose, and the game may be lopsided, but the game was at least FAIR.

In the case where equipment does not match up, then the player is going to feel cheated. He will feel like the deck is stacked against him, and this is one of the worst feelings that a game can inspire in a player. It's what makes players quit a game.

If you have to choose between even tonnage matching and skill matching (and it seems like you do in fact have to make that choice), then you need to err on the side of tonnage matching, because it is more important in making games appear fair to your playerbase.

[Satan n stuff]

The biggest problem is that there is a huge difference between the skill level of players in a match. That could easily be solved if the matchmaker would just adjust the target value based on the elo score of those already selected. So if the first player has a high elo score, the target value goes up and the matchmaker will try to find players closer to the first one, making it more likely that the teams will consist of equally skilled players.

[Ghogiel]

I wouldn't say no to a week of stat padding.

Edited by Ghogiel, 22 October 2013 - 06:14 AM.

[Joseph Mallan]

You know me... I am game for battling all comers, Win or lose!

[Blast Thickneck]

Tonnage limits on groups is incoming, no need to take steps backwards to soothe bruised egos. Chin up little fella, it will get better,

[Felio]

The last few nights, I have lost about 9/10 games. It's demoralizing. I don't really want to play anymore.

But I'd rather see changes made to improve ELO than see it removed.

[Kunae]

Blast Thickneck, on 22 October 2013 - 06:16 AM, said:

Tonnage limits on groups is incoming, no need to take steps backwards to soothe bruised egos. Chin up little fella, it will get better,

This isn't about ego. This is about throwing new players, in trial mechs, in with people who've played 7000 matches... on purpose. This is what the current, broken, Elo system is doing. And, as Roland pointed out, it's doing this in preference to balancing the weight-classes between sides.

At least with no Elo, you will get a much more varied and mixed distribution of players in any given match.

[dario03]

How does the system work right now? Isn't it by filling or mostly filling one team and then filling the other team with a similar elo score? Wouldn't it be better if it just grabbed 24 players and then divided them so that not just the average but the range was similar? Or is it actually doing it like that now? Basically instead of a team that has everyone with a elo of 8 fighting a team of 6's and 10's, both teams would have equal numbers of 6's, 8's, 10's

[Kunae]

Felio, on 22 October 2013 - 06:17 AM, said:

The last few nights, I have lost about 9/10 games. It's demoralizing. I don't really want to play anymore.

But I'd rather see changes made to improve ELO than see it removed.

Removal is an improvement. Elo is designed for 1v1 ranking/matchups. It was never designed to balance teams(premade or not).

dario03, on 22 October 2013 - 06:26 AM, said:

How does the system work right now? Isn't it by filling or mostly filling one team and then filling the other team with a similar elo score? Wouldn't it be better if it just grabbed 24 players and then divided them so that not just the average but the range was similar? Or is it actually doing it like that now? Basically instead of a team that has everyone with a elo of 8 fighting a team of 6's and 10's, both teams would have equal numbers of 6's, 8's, 10's

It's doing what I underlined, although to greater extremes.

[Chavette]

The test server doesn't have elo MM on. Jump in next time.

[Kunae]

Chavette, on 22 October 2013 - 06:30 AM, said:

The test server doesn't have elo MM on. Jump in next time.

Unfortunately, as they only have it up for very brief periods with limited populations, and typically when I am not available, that's not really a good option.

[A banana in the tailpipe]

I love how [people] will go so far as to create threads demanding players step up to thier level, when most players are more than capable of smoking the average pro, they just chose to ignore the meta out of spite to PGI or don't care because the game's too broken balance wise to bother. There are also too many other outside factors nobody gives due credit to, such as casual playtime, alcohol, narcotic influence, digital distraction such as watching multiple monitors with movies/tv ect, pets, children, telephones, and any other sort of distractions that might incluence any given match.

But no.... it must be every player is bad but you, and unless [people] get thier own special playground full of like-minded individuals they cry a river in chat and on the forums. Just stop beating around the bush and ask PGI for arenas and ladder matches already. It's what you secretly want. Stomp matches will always exist due to the distractions I've mentioned and the ELO will never be perfect.

/deal with it.

Edited by miSs, 22 October 2013 - 09:13 AM.

[dario03]

Kunae, on 22 October 2013 - 06:28 AM, said:

Removal is an improvement. Elo is designed for 1v1 ranking/matchups. It was never designed to balance teams(premade or not).


It's doing what I underlined, although to greater extremes.

Ok that's what I thought. Well then yeah, why not change it so that it splits everyone after getting 24 players. Try to divide it up so that the teams average elo and elo range. Probably also try to keep premades split evenly too if theres more than one in a match and maybe have a multiplier if they can't be even.

[Ghogiel]

Kunae, on 22 October 2013 - 06:26 AM, said:

This isn't about ego. This is about throwing new players, in trial mechs, in with people who've played 7000 matches... on purpose. This is what the current, broken, Elo system is doing. And, as Roland pointed out, it's doing this in preference to balancing the weight-classes between sides.

At least with no Elo, you will get a much more varied and mixed distribution of players in any given match.

Cool, this way I won't have to have 4 failed to find matches in a row for every match I do get with my assaults AND I will be able to roflstomp 90% of my matches while I premade with 3 other vets.

dario03, on 22 October 2013 - 06:26 AM, said:

How does the system work right now? Isn't it by filling or mostly filling one team and then filling the other team with a similar elo score? Wouldn't it be better if it just grabbed 24 players and then divided them so that not just the average but the range was similar? Or is it actually doing it like that now? Basically instead of a team that has everyone with a elo of 8 fighting a team of 6's and 10's, both teams would have equal numbers of 6's, 8's, 10's

It picks a target Elo number for a match and then places players who are within the Elo range constraints of that target number on the teams. Then assigns a probability who will win based on the total Elo of the teams, the team with the higher Elo is more likely to win. If the team with the higher Elo rating wins their Elo rating would be increased less than if the team with the lower Elo rating would have won. The math for how the Elo adjustments are done are in the command chair post.

And nah. Having scrubs anywhere near good players in a game ends with them getting pwnt.

[Kunae]

dario03, on 22 October 2013 - 06:39 AM, said:

Ok that's what I thought. Well then yeah, why not change it so that it splits everyone after getting 24 players. Try to divide it up so that the teams average elo and elo range. Probably also try to keep premades split evenly too if theres more than one in a match and maybe have a multiplier if they can't be even.

While that would be a barely functional "workaround", why are we trying to work around a system that is unsuited, from the start?

Why try to modify your Volkswagen Golf so it can carry 4'x8' sheets of plywood, when you can just get a truck for that?

[Kunae]

Ghogiel, on 22 October 2013 - 06:39 AM, said:

Cool, this way I won't have to have 4 failed to find matches in a row for every match I do get with my assaults AND I will be able to roflstomp 90% of my matches while I premade with 3 other vets.

And even without Elo in force, you're just as likely to find yourself facing a like 4-man, or maybe 2 of them. Randomness is the spice of life, in a game like this.

Ghogiel, on 22 October 2013 - 06:39 AM, said:

It picks a target Elo number for a match and then places players who are within the Elo range constraints of that target number on the teams. Then assigns a probability who will win based on the total Elo of the teams, the team with the higher Elo is more likely to win. If the team with the higher Elo rating wins their Elo rating would be increased less than if the team with the lower Elo rating would have won. The math for how the Elo adjustments are done are in the command chair post.

And nah. Having scrubs anywhere near good players in a game ends with them getting pwnt.

They have specifically stated that the MM will balance out high Elo players with low Elo players, on a side, to try to approach this "average" target. It's silly.

[mike29tw]

I predict no noticeable difference after they remove ELO matchmaking.

[dario03]

Kunae, on 22 October 2013 - 06:41 AM, said:

While that would be a barely functional "workaround", why are we trying to work around a system that is unsuited, from the start?

Why try to modify your Volkswagen Golf so it can carry 4'x8' sheets of plywood, when you can just get a truck for that?

Well I don't think they are just going to drop elo so I'm suggesting something that I think would be a improvement without needing a entirely new system. Its not perfect but I think it would be a little better. Or if somebody has a better system then lets do that, but I don't think they are just going to turn elo off (but I'm fine trying that too).

Edited by dario03, 22 October 2013 - 06:56 AM.

[Ghogiel]

Kunae, on 22 October 2013 - 06:45 AM, said:

And even without Elo in force, you're just as likely to find yourself facing a like 4-man, or maybe 2 of them. Randomness is the spice of life, in a game like this.

I'm cool with it. I'll just go back to having 10 KDRs and W/L like back before Elo.