18 replies to this topic

[MischiefSC]

I'll start by saying I'm a fan of the Elo idea and I think in general it's a good one. Necessary even and I'm not trying to knock it. This isn't a flame thread by any means. I'm a big fan of discussion for a productive end, not just to have something to cry over.

However, Elo ranking has a couple of concerns tied to it.

Your Elo score is a chain that is very difficult to escape. Here is my prediction of a timeline that's going to follow Elo release, give or take.

1. Elo is released. Rainbows puppies and kittens for everyone. Thank you threads come out as people talk about the better challenge involved and better balance much like the lagshield fix.

2. The higher tiers of team play quickly shake out the pug farmers from the elite and high Elo play becomes like 8v8s are today - the best builds for teams will show themselves and dominate and in turn reduce variety in the same way that all competitive sports do. Everyone either goes with what works or they lose.

3. People will complain that they play the game for fun, not constant tournament style play. People and teams with 80% win rates will drop to about 30% or 40% as the tactics that work well against disorganized players fail against practiced organized opponents.

4. A brief interlude of alt accounts mixing with team accounts to lower the average Elo score of the team for a drop. Two 'newbies' in the form of alternate accounts for experienced players will drop with two regular experienced players allowing 4 experienced players (or 8) to drop against less experienced opponents in the hopes of increasing win rate and edging up Elo scores for under-performing teams on the margin. This fails when it results in the teams doing so getting brutally brutally humped by the even higher Elo teams they now come into contact with.

5. Rise of the alt accounts. Experienced players on organized teams will make alts to drop with together 'to just play for fun' since there is no way for them to play at their Elo level with existing characters without it being an ECM/missile/sniper fest requiring precision coordination and teamwork not to get rolled by hyper-competitive teams.

6. Teams of experienced players dropping in groups of alts with a neutral Elo will pubstomp newbies under the auspice of 'playing for fun' (which isn't an unreasonable reason to play a game) and badly skew the lower spectrum of Elo advancement, crushing new and less skilled players from gaining Elo with much speed and destroying their enjoyment of the game. Que pugstomp threads and many tears.

7. Brief and failed attempt to get PGI to make MC, or mechs or perks like Founders carrying between accounts 'since it's the same person playing' as PGI effectively gets people to pay more than once for the same things (paint, hero mechs, etc) and artificially inflates subscriber levels. For PGI, this is not a bad scenario.

8. Top tier Elo population declines for the same reason 8v8 declined, not everyone wants to play hard core competitive games all the time but once your Elo score gets high enough that's all you drop against.

9. Mid to high level elite pugs have a field day as suddenly everyone they drop with is like them - experienced pugs who are good at teamwork by practice and not voice chat, focusing fire and relying on hand-eye coordination and brutal experience. Unfortunately they hit the 'glass ceiling' of hardcore competitive team play and can not advance further without joining a comparable team and conversely losing a lot of the casual aspect of the game that motivated them to play. They can either up-convert or quit.

10. ?

I don't know what happens at this point.

So how do you separate those experiences? How do you let people with high Elo scores just play for fun without effectively defeating what the Elo is in place to do; place people with comparable opponents.

Winning is funner than losing. People like to win. When you have people winning 80% of the time and suddenly they're at less than 50%, they're going to look for a way to get back to the winning all the time experience.

Suggestions? This isn't a flame. It's a concerned prediction of some possible problems coming with Elo matchmaking. Just being aware of it as a potential isn't a bad thing but to some greater or lesser degree these activities will probably exist and having a high Elo score is going to, for some people, make the game less fun. The point is to ask how do you fix that?

[Mackman]

The obvious solution is to have ranked and unranked play.

EDIT: I agree with you, people will try to game the system, and no system is foolproof. But any system is better than the nothing we have now.

Edited by Mackman, 06 February 2013 - 10:59 AM.

[Sir Wulfrick]

While this is fairly speculative, I suppose it may be a reasonable deduction from the published plans surrounding ELO.

Personally I like the idea of one player = one account. How this could be achieved technically is however much more difficult since IP =! person.

Perhaps there could be separate groups of matchmaker queues: a competitive queue where match-making is calculated strictly by ELO rating; and then a 'casual' queue where the match making could be a bit more fuzzy in it's decissions, thus providing everone with challenging games but not so tough that you could make boots out of it. Just a thought...

Edited by Sir Wulfrick, 06 February 2013 - 11:01 AM.

[MischiefSC]

Unranked play is a good idea. What's funny is that ranked play will become more of a haven for new players while unranked play will involve team drops of varying skill, while pug players will stick to ranked play in that situation.

How do you keep alts dropping in premade teams to stomp newbies though?

[FrDrake]

Yea, it'll have to go the way of LoL with ladder matches and hidden Elo matches.

[Vlad Ward]

Playing against challenging opponents is fun for me.

I can tell you right off the bat, though, that the vast majority of teams who quit the 8v8 queue in a stormy rage were never close to the level of the teams that beat them, tonnage or no tonnage. They aren't going to be "kicked out" of the top ELO range because they won't approach it to begin with.

There is a wide variety of team compositions that can succeed in 8v8s. They simply rely on both individual and team skill. All you're seeing is the aftermath of a bunch of pugstompers getting a major reality check when their targets started shooting back.

Regarding Alts, you either have to devote a bunch of time to building them up (which would increase your ELO anyways...) or dump a bunch of real cash to buy them MC toys to stomp with. Neither solution is very sustainable for the metaphorical premade.

[Natasha Kerensky]

MischiefSC, on 06 February 2013 - 10:53 AM, said:

I'll start by saying I'm a fan of the Elo idea and I think in general it's a good one. Necessary even and I'm not trying to knock it. This isn't a flame thread by any means. I'm a big fan of discussion for a productive end, not just to have something to cry over.

However, Elo ranking has a couple of concerns tied to it.

Your Elo score is a chain that is very difficult to escape. Here is my prediction of a timeline that's going to follow Elo release, give or take.

1. Elo is released. Rainbows puppies and kittens for everyone. Thank you threads come out as people talk about the better challenge involved and better balance much like the lagshield fix.

2. The higher tiers of team play quickly shake out the pug farmers from the elite and high Elo play becomes like 8v8s are today - the best builds for teams will show themselves and dominate and in turn reduce variety in the same way that all competitive sports do. Everyone either goes with what works or they lose.

3. People will complain that they play the game for fun, not constant tournament style play. People and teams with 80% win rates will drop to about 30% or 40% as the tactics that work well against disorganized players fail against practiced organized opponents.

4. A brief interlude of alt accounts mixing with team accounts to lower the average Elo score of the team for a drop. Two 'newbies' in the form of alternate accounts for experienced players will drop with two regular experienced players allowing 4 experienced players (or 8) to drop against less experienced opponents in the hopes of increasing win rate and edging up Elo scores for under-performing teams on the margin. This fails when it results in the teams doing so getting brutally brutally humped by the even higher Elo teams they now come into contact with.

5. Rise of the alt accounts. Experienced players on organized teams will make alts to drop with together 'to just play for fun' since there is no way for them to play at their Elo level with existing characters without it being an ECM/missile/sniper fest requiring precision coordination and teamwork not to get rolled by hyper-competitive teams.

6. Teams of experienced players dropping in groups of alts with a neutral Elo will pubstomp newbies under the auspice of 'playing for fun' (which isn't an unreasonable reason to play a game) and badly skew the lower spectrum of Elo advancement, crushing new and less skilled players from gaining Elo with much speed and destroying their enjoyment of the game. Que pugstomp threads and many tears.

7. Brief and failed attempt to get PGI to make MC, or mechs or perks like Founders carrying between accounts 'since it's the same person playing' as PGI effectively gets people to pay more than once for the same things (paint, hero mechs, etc) and artificially inflates subscriber levels. For PGI, this is not a bad scenario.

8. Top tier Elo population declines for the same reason 8v8 declined, not everyone wants to play hard core competitive games all the time but once your Elo score gets high enough that's all you drop against.

9. Mid to high level elite pugs have a field day as suddenly everyone they drop with is like them - experienced pugs who are good at teamwork by practice and not voice chat, focusing fire and relying on hand-eye coordination and brutal experience. Unfortunately they hit the 'glass ceiling' of hardcore competitive team play and can not advance further without joining a comparable team and conversely losing a lot of the casual aspect of the game that motivated them to play. They can either up-convert or quit.

10. ?

I don't know what happens at this point.

So how do you separate those experiences? How do you let people with high Elo scores just play for fun without effectively defeating what the Elo is in place to do; place people with comparable opponents.

Winning is funner than losing. People like to win. When you have people winning 80% of the time and suddenly they're at less than 50%, they're going to look for a way to get back to the winning all the time experience.

Suggestions? This isn't a flame. It's a concerned prediction of some possible problems coming with Elo matchmaking. Just being aware of it as a potential isn't a bad thing but to some greater or lesser degree these activities will probably exist and having a high Elo score is going to, for some people, make the game less fun. The point is to ask how do you fix that?

The whole tread is one big assumption.

[Kdogg788]

Since this is a team game, and not an FPS or head to head strategy game, I'm willing to bet in 4 mans at least, it will come to some sort of average of your total team score, where it could be possible to have a team of elites and noobs together against the more average ELO scores. I doubt the MM score will be a hard cap to where you won't be able to find matches unless you're in an 8 man in some cases, only a way of ordering teams and drops, and premade groups in a more balanced manner.

-k

[OneEyed Jack]

If PGI makes the mistake of not having un-ranked play available for when competitive players want to play just for fun, then they will make alt accounts. They'll play whatever the hell they feel like on those accounts and not worry about rising in ranking. If they're really that good, sure, they'll rise up. But they won't be forced to begin using the most competitive builds just to play, they'll just reach a balance where they're good enough to win about half their games in whatever they feel like playing, and since they're not worried about the rank on that account, that's good enough.

This is not a big deal.

Nothing to see here, move along.

[Mackman]

MischiefSC, on 06 February 2013 - 11:01 AM, said:

Unranked play is a good idea. What's funny is that ranked play will become more of a haven for new players while unranked play will involve team drops of varying skill, while pug players will stick to ranked play in that situation.

How do you keep alts dropping in premade teams to stomp newbies though?

I would make it so in ranked play, everyone on the team must be within <CALCULATED NUMBER HERE> Elo of each other. Say, 100 or so. In unranked play, anything goes, so that you can play with your friend who just started playing. Another would be to simply skew the average much more towards the highest Elo on the team.

[IceSerpent]

I don't see the point of "gaming the system" like this - matching by equal ELO implies roughly 50/50 chance of winning (by definition of ELO), if you drop your ELO by throwing a couple of matches you would also end up with roughly 50/50 win/loss ratio (you lose 10 matches, then steamroll 10 matches - your ELO is back to where it was).

[MischiefSC]

Vlad Ward, on 06 February 2013 - 11:02 AM, said:

Playing against challenging opponents is fun for me.

I can tell you right off the bat, though, that the vast majority of teams who quit the 8v8 queue in a stormy rage were never close to the level of the teams that beat them, tonnage or no tonnage. They aren't going to be "kicked out" of the top ELO range because they won't approach it to begin with.

There is a wide variety of team compositions that can succeed in 8v8s. They simply rely on both individual and team skill. All you're seeing is the aftermath of a bunch of pugstompers getting a major reality check when their targets started shooting back.

Regarding Alts, you either have to devote a bunch of time to building them up (which would increase your ELO anyways...) or dump a bunch of real cash to buy them MC toys to stomp with. Neither solution is very sustainable for the metaphorical premade.

What I'm talking about is 4 man drops with trial mechs who are, based on a neutral Elo, going to be dropping largely against brand new players in trial mechs. It'll be far more one-sided than anything we see today since it will literally be 4 man organized teams of skilled players against people with less than 20 matches under their belts.

I agree that alts will probably not fill into the middle Elo range but some might. Just decide to 'start over' with your friends and work your way back up. For people who enjoy winning more than hardcore competition the ladder of working your way back up through less experienced (but probably better equipped) players might be a lot of fun - just not for the people they're chewing through. Rinse and repeat until you get high enough in Elo that you start running into the same issue again and.... back to step 2.

What's important to keep in mind though is that people playing back through with alts are not going to pug. They're going to play with their same teams and effectively recreate the situation that Elo is trying to remove. Putting experienced and competitive team players in games against new and inexperienced players or casual players for a very one-sided experience.

Absolutely the whole thing is an assumption, it has to be. I think the polite industry term is 'predictive analysis'.

So unranked play, is that going to essentially just be how it works right now just hopefully without trial accounts and new players? Will it separate premade vs pug or attempt to balance populations on both sides of drops? How do you keep unranked play from becoming exactly what Elo is being put in to fix or avoid?

[Tikkamasala]

MischiefSC, on 06 February 2013 - 10:53 AM, said:

2. The higher tiers of team play quickly shake out the pug farmers from the elite and high Elo play becomes like 8v8s are today - the best builds for teams will show themselves and dominate and in turn reduce variety in the same way that all competitive sports do. Everyone either goes with what works or they lose.

3. People will complain that they play the game for fun, not constant tournament style play. People and teams with 80% win rates will drop to about 30% or 40% as the tactics that work well against disorganized players fail against practiced organized opponents.

8. Top tier Elo population declines for the same reason 8v8 declined, not everyone wants to play hard core competitive games all the time but once your Elo score gets high enough that's all you drop against.

9. Mid to high level elite pugs have a field day as suddenly everyone they drop with is like them - experienced pugs who are good at teamwork by practice and not voice chat, focusing fire and relying on hand-eye coordination and brutal experience. Unfortunately they hit the 'glass ceiling' of hardcore competitive team play and can not advance further without joining a comparable team and conversely losing a lot of the casual aspect of the game that motivated them to play. They can either up-convert or quit.

So how do you separate those experiences? How do you let people with high Elo scores just play for fun without effectively defeating what the Elo is in place to do; place people with comparable opponents.

Winning is funner than losing. People like to win. When you have people winning 80% of the time and suddenly they're at less than 50%, they're going to look for a way to get back to the winning all the time experience.

2. & 3. & 8. The highest elo play will most likely involve optimised builds and strategies. If you don't want to be a part of that, e.g. you like to play fun variants/mechs, you will drop in elo (if you were up there to begin with) and start to be matched with players of similar skill with fun variants or less skillful opponents in optimised builds. How is that a problem? Isn't that exactly what you'd want in that situation? The actual elo number is of no real value in that case anyway.

9. Same for the mid-high elo tier of players: there is no glass ceiling if all you want is a fun and challenging game as you get matched with players of similar skill.

As for the win-loss-ratio: if the matchmaking works properly that ratio will converge on 0.5 eventually.

[Vlad Ward]

I think it's a tad farfetched, to be honest. It assumes a lot about human nature and that's never a safe thing to do. Eddie Murphy may be able to pull it off in the big time Hollywood movies, but real world predictive analysis relies on numbers, math, and data collection.

[MischiefSC]

Vlad Ward, on 06 February 2013 - 11:17 AM, said:

I think it's a tad farfetched, to be honest. It assumes a lot about human nature and that's never a safe thing to do. Eddie Murphy may be able to pull it off in the big time Hollywood movies, but real world predictive analysis relies on numbers, math, and data collection.

I absolutely agree, though I wouldn't say it's far fetched. I deal with predictive analysis (not gathering it, just playing with the results) in a customer facing industry. 82% of customer complaints and escalations come from the same ~5% of the customer population, that's pretty consistent. I know that in the days when I worked in the gaming industry about 10% of the games population accounted for almost all the griefing, cheating and exploit reports.

The problem is that abusive behavior has an exponential impact on the game. It promotes 'copycat griefing' from other players, etc.

I like the idea of unranked play. It's a game, you don't want to just play competitive ranked matches all the time do you?

Maybe none of it will come up and everyone will just be happy to keep playing at their Elo rank, though again I strongly suspect it'll be pretty top heavy and recreate the problems that depopulated 8v8 games.

Tikkamasala, on 06 February 2013 - 11:16 AM, said:

2. & 3. & 8. The highest elo play will most likely involve optimised builds and strategies. If you don't want to be a part of that, e.g. you like to play fun variants/mechs, you will drop in elo (if you were up there to begin with) and start to be matched with players of similar skill with fun variants or less skillful opponents in optimised builds. How is that a problem? Isn't that exactly what you'd want in that situation? The actual elo number is of no real value in that case anyway.

9. Same for the mid-high elo tier of players: there is no glass ceiling if all you want is a fun and challenging game as you get matched with players of similar skill.

As for the win-loss-ratio: if the matchmaking works properly that ratio will converge on 0.5 eventually.

All excellent points. The only question I have is how will it work to deal with

A) people who are not in the mood for ranked games

people who try to make alts and go feast on the sweet sweet flesh of newbies

Unranked play is a cool idea. Is newbie hunting with friends in alt characters going to be 'alright'?

[Tikkamasala]

MischiefSC, on 06 February 2013 - 11:36 AM, said:

All excellent points. The only question I have is how will it work to deal with

A) people who are not in the mood for ranked games

B) people who try to make alts and go feast on the sweet sweet flesh of newbies

Unranked play is a cool idea. Is newbie hunting with friends in alt characters going to be 'alright'?

A) If you are not in the mood often than your elo will reflect that and you get to play with like-minded or lower rated players. Your performance will vary according to your mood, but that's not really too different from everybody else. If you are not trying to keep a certain elo value (why would you?) i don't really see a problem with that.

B) It's a bit like cheating in a solo player game with the added important difference that your actions might negatively impact other players enjoyment of the game.

[Kekrebos]

Even though this is all theory, I would say its a plausible outcome if there is no unranked play option.

There are a LOT of people obsessed with stats, ladders, and ELO. I know that in World of Tanks when I started playing with a group of rather good players I started becoming obsessed with stats, and getting mine better. To the point that I wouldn't play a match unless I was platooned with at least one other player. You sometimes get competitive even with your friends.

I think an unranked play mode would have a lot of appeal to new players that are just starting out, people messing around with builds, your casual solo player that doesn't want to be bothered, and even your 'pro' player that just wants to have fun for a few matches when his buddies aren't around.

[MischiefSC]

Tikkamasala, on 06 February 2013 - 11:53 AM, said:

A) If you are not in the mood often than your elo will reflect that and you get to play with like-minded or lower rated players. Your performance will vary according to your mood, but that's not really too different from everybody else. If you are not trying to keep a certain elo value (why would you?) i don't really see a problem with that.

It's a bit like cheating in a solo player game with the added important difference that your actions might negatively impact other players enjoyment of the game.

The problem being however keeping your Elo in line with your group. One thing I'm curious to see is how it'll handle not enough people of a comparable Elo available. Does it wait longer or just populate with people further up or down the ladder? How about team play, will it take disparity of skill into account at an individual level when calculating advances in Elo?

Rallog, on 06 February 2013 - 11:54 AM, said:

Even though this is all theory, I would say its a plausible outcome if there is no unranked play option.

There are a LOT of people obsessed with stats, ladders, and ELO. I know that in World of Tanks when I started playing with a group of rather good players I started becoming obsessed with stats, and getting mine better. To the point that I wouldn't play a match unless I was platooned with at least one other player. You sometimes get competitive even with your friends.

I think an unranked play mode would have a lot of appeal to new players that are just starting out, people messing around with builds, your casual solo player that doesn't want to be bothered, and even your 'pro' player that just wants to have fun for a few matches when his buddies aren't around.

Hence why I'm suddenly very fond of an unranked play option. Ideally balanced for premade and pug populations on each side when possible. Newer players would be better suited to stick with ranked play as they'll have a better experienced while unranked play can be a playground for more experienced players.

[Tikkamasala]

MischiefSC, on 06 February 2013 - 12:04 PM, said:

The problem being however keeping your Elo in line with your group. One thing I'm curious to see is how it'll handle not enough people of a comparable Elo available. Does it wait longer or just populate with people further up or down the ladder? How about team play, will it take disparity of skill into account at an individual level when calculating advances in Elo?

I assume the matchmaker widens the matching threshold if it cannot fill both teams or the devs tune the initial threshold until on average a match starts after an acceptable wait time. The teams elo is calculated with a simple arithmetic mean iirc so it doesn't care about disparities and can be easily influenced by outliers. From the example in the dev matchmaking post i got the impression that elo changes are calculated per player and not per team.

MischiefSC, on 06 February 2013 - 12:04 PM, said:

Hence why I'm suddenly very fond of an unranked play option. Ideally balanced for premade and pug populations on each side when possible. Newer players would be better suited to stick with ranked play as they'll have a better experienced while unranked play can be a playground for more experienced players.

Iirc in LoL the unranked queue is actually handled by a matchmaker using its own separate elo values. Is that what we are talking about here as opposed to a totally random matchmaking?