71 replies to this topic

[Daurock]

justcallme A S H, on 21 November 2018 - 11:59 AM, said:

So fixed means, perfect? Lol

You can absolutely make changes to the MM to, potentially, improve matches. It's not that hard to do and is a rather easy 2-Step process. PGI just refuse to come to that way if thinking unfortunately.

Right now just dumping 24 into a match based on "Tier" is all that happens. That means there is a huge room for improvement because the fundamental system it is based on is utterly broken. How you fail to comprehend that is mind-blowing.

Even if there was double the current population (so higher than it's been the last 2 years) there would be no perfect 'fix' for the MM which is what you seem to be after.

To make it even noticeably better (Not perfect, but noticeably different than just what is done today), it would have to be a far more complicated MM than just a tier or ELO rating, and sorting around that. The reason behind that revolves around the fact that people play different mechs, and the MM has to still balance that out. Some people are just awful in lights, yet amazing in assaults, others the other way around. Throw in the fact that sometimes you'll have a great player on each team, one with a light, and the opposing one in an assault, and have to balance out the entirety of each team around that, too. Some maps are going to play to specific builds, which may or may not be present on a given team, even if the skill IS balanced out, so there's that to think about too. I could continue here for a long time on the list of "things that could/should be sorted around to make an even game" but i'd probably be here for a while.

Even the simple sounding attempt to weave in a simple "ELO sorting" is going to have issues, because you have to decide whether the mech the player brought, or the player's ELO is more important, and that answer is going to change on a game-by game basis, based on the zillion other factors out there. No single algorithm is even going to begin to be able to answer question more often than not. At absolute best, you could maybe sort each set of players in each weight class in each game by ELO, but that's going to only have a vague impact on the actual games themselves.

The bottom line here is, there's a TON of noise in this particular game. Between builds, various mechs, and every other variable out there, there is so much noise that besides sorting for weight class, which is already done, there's not a lot else that is going to make an appreciable difference in ensuring a balanced match. The challenge of making a noticeably better MM is monumental in this particular case, simply due to the multitude of factors involved.

Lastly, couple that with the meme of "Why is PGI putting me on terrible teams all the time?" This simple meme would actually be the Reality for good players at that point, since if they sorted by skill rating, they ACTUALLY WOULD be getting the bad teams all the time, and be expected to carry nearly all the time. Is that really what we want to do?

[justcallme A S H]

Daurock, on 21 November 2018 - 01:43 PM, said:

To make it even noticeably better (Not perfect, but noticeably different than just what is done today), it would have to be a far more complicated MM than just a tier or ELO rating, and sorting around that. The reason behind that revolves around the fact that people play different mechs, and the MM has to still balance that out. Some people are just awful in lights, yet amazing in assaults, others the other way around. Throw in the fact that sometimes you'll have a great player on each team, one with a light, and the opposing one in an assault, and have to balance out the entirety of each team around that, too.

In the grand scheme - it will balance out. Someone with a 450AMS is going to do well, in most instances, no matter the mech chosen. They do not have a 450AMS piloting their worst class with the worst mech with the worst build.

As I said it is two parts.

• Fix the Tier structure so it accurately reflects player skill as opposed to an experience bar where even the bottom 40% of the playerbase can end up in Tier 1
  
• Fix the MM itself such that once 24 players are located for a lobby. It then balances the lobby to the best of its ability based on WLR / AMS.

It's not perfect but given what we have now is basically a non-existent match maker that forms matches, essentially, on how long you've played/grinded - it's going to be a darn sight better than what we have.

What you are talking about is a perfect world solution. That'll never happen and it is literally impossible to code for. So you take the improvements you can get rather than just flatly ignore it.

BIOHAZARD, on 21 November 2018 - 01:04 AM, said:

Spoiler

I am sure this has been mentioned a million times before, and there is a huge PUG population out there that will cry at the top of their lungs against it.

BUT

This is the only way to get more experienced players to WANT to stay in the game. Nothing else will work because the alternatives are dev intensive. We are talking about rebuilding this game instead, which is simply not going to happen.

PGI!

Just untie individual payouts after the match from team performance already!

This is a team game, in which we cannot choose our team. Half the time you wouldn't even WANT some of the people you end up with, to be on your team. And when you carry, and you loose, this makes the experience simply not worth your time.

Just untie what we get after the match! Pay everyone based on their individual performance, as if this was a won match every time. Or at least to the players, on the loosing side, that performed well enough to get the green arrow up score. Then the real tier 1 will not care anymore about pugstomps.

P.S. I will also take the 'I don't ever want to be with this dude on the same team again' button.

Goes to Jarls list to have a look. I see the issue.

You need to carry harder. WLR and AMS will naturally rise if you improve, plenty of room for that from what I see.

Or keep posting / asking PGI to fix the Tier system. That will mean you'll end up in the middle of Tier 2 though because your payouts are definately a reflection of your ingame performance.

[Nightbird]

AMS isn't really that useful either since it doesn't reflect a player's contribution to a team. Even on the first page of Jarl's you can see the top 100 AMS players have a W/L that swings from 1.5 to 4. That's basically as bad as the tier system.

[Navid A1]

How about fixing the goddamn PSR system and defining new thresholds for Tiers and Tier jumps.

It baffles the mind why something so stupidly easy is not done already.

[justcallme A S H]

Nightbird, on 21 November 2018 - 02:36 PM, said:

AMS isn't really that useful either since it doesn't reflect a player's contribution to a team. Even on the first page of Jarl's you can see the top 100 AMS players have a W/L that swings from 1.5 to 4. That's basically as bad as the tier system.

Match Score is what the core number that the Tier system uses. Useful or not - we aren't about to see PGI totally reinvent the wheel to some complex system. You could in-lobby balance on WLR as a sub-set easily if you wanted or other such easy tweaks/add ons.

As for top 100 on Jarls - They have a high AMS, they win more often than they lose - therefore having a positive affect on a battle / team they are in.

Navid A1, on 21 November 2018 - 02:43 PM, said:

How about fixing the goddamn PSR system and defining new thresholds for Tiers and Tier jumps.

It baffles the mind why something so stupidly easy is not done already.

Yep. When you can have 170 AMS and end up in Tier 1, which is below the MEDIAN of the entire active population - that is a giant red flag

Edited by justcallme A S H, 21 November 2018 - 02:53 PM.

[Grus]

justcallme A S H, on 21 November 2018 - 11:59 AM, said:

So fixed means, perfect? Lol

You can absolutely make changes to the MM to, potentially, improve matches. It's not that hard to do and is a rather easy 2-Step process. PGI just refuse to come to that way if thinking unfortunately.

Right now just dumping 24 into a match based on "Tier" is all that happens. That means there is a huge room for improvement because the fundamental system it is based on is utterly broken. How you fail to comprehend that is mind-blowing.

Even if there was double the current population (so higher than it's been the last 2 years) there would be no perfect 'fix' for the MM which is what you seem to be after.

Agreed, just throwing X teir with X teir dosnt work. There needs to be a more in-depth look from the MM to determan a closer match up.

Somthing i would add would be what mech is person A going to bring. ( lets say light ) check person stats; likelyhood A will be able to carry a match in said light; match 11 others accordingly.

Are there pilots that can pilots lights and do run ham and get a ton of kills, yes. and that is, and what should be counted.

I can say for my self i love a good match that is challenging. Not a match where 11 pugs are dead in the first 2 min and im still shoiting going... what in the hell happened?

Edited by Grus, 21 November 2018 - 03:23 PM.

[Nightbird]

justcallme A S H, on 21 November 2018 - 02:53 PM, said:

Match Score is what the core number that the Tier system uses. Useful or not - we aren't about to see PGI totally reinvent the wheel to some complex system. You could in-lobby balance on WLR as a sub-set easily if you wanted or other such easy tweaks/add ons.

As for top 100 on Jarls - They have a high AMS, they win more often than they lose - therefore having a positive affect on a battle / team they are in.

Well, Tier 1s are better players than Tier 5s. So what? Match Score, assigning arbitrary weights to certain actions without any attempt to figure out how much that action benefits the team. Calling AMS a horribly inaccurate measuring stick is an insult to all the bad measure sticks out there.

[Nightbird]

^ imagine taking a test with 100 questions, getting 99 right and 1 wrong and feeling good. Then when you see your grade, it's a C. You're like, WTF, I got 99%, and the prof says, no, that 1 question you got wrong is worth 30% of your grade. You say, BS, that question is the same length and difficulty as the other ones. Prof says too bad. That's match score folks, arbitrary as all hell.

[Anomalocaris]

Nightbird, on 21 November 2018 - 04:04 PM, said:

^ imagine taking a test with 100 questions, getting 99 right and 1 wrong and feeling good. Then when you see your grade, it's a C. You're like, WTF, I got 99%, and the prof says, no, that 1 question you got wrong is worth 30% of your grade. You say, BS, that question is the same length and difficulty as the other ones. Prof says too bad. That's match score folks, arbitrary as all hell.

Match score isn't perfect. Some things (like hiding in cover, etc.) are rewarded that mean little (if you're doing something that helps you survive, that is rewarded in the end with a win, survival, etc.). But as ASH pointed out it's still a performance based metric and, on average, a player with a higher AMS is better than one with a lower AMS. If you sort by a combo of WLR and AMS you're going to do a much better job of balancing teams.

Look, we all know how PGI works. With the age of MWO and their future development priorities, this game is going to get minimal resource investment. Asking them to add sorting based upon WLR and AMS, two existing, archived statistics, is about as much as anyone could expect at this point. Asking for a new ranking system, etc. is pissing in the wind IMO.

[Warning incoming Humble Dexterer]

BIOHAZARD, on 21 November 2018 - 01:04 AM, said:

I am sure this has been mentioned a million times before... Just untie individual payouts after the match from team performance already!

I believe payout is ALREADY untied from team performance : You can score great, really great, even if your team sucks and loses.

So in my case, I said the opposite : Tie individual payouts to your team's performance.

But by that I mean that if you rank in the top spots of your own 12-man team, then you deserve a top payout, even if your whole team (including yourself) got stomped so bad it barely damaged the enemy team.

Because some times your team is so bad you're better off leaving the game... but no matter how unbalanced the teams are it's always possible to score the top position of your own team.

Edited by Humble Dexter, 21 November 2018 - 05:04 PM.

[Grus]

Humble Dexter, on 21 November 2018 - 04:59 PM, said:

I believe payout is ALREADY untied from team performance : You can score great, really great, even if your team sucks and loses.

So in my case, I said the opposite : Tie individual payouts to your team's performance.

But by that I mean that if you rank in the top spots of your own 12-man team, then you deserve a top payout, even if your whole team (including yourself) got stomped so bad it barely damaged the enemy team.

Because some times your team is so bad you're better off leaving the game... but no matter how unbalanced the teams are it's always possible to score the top position of your own team.

I wonder if we could convince Quickybaby to do a XVM for MWO?..

It would be nice to see what i have to work with in game so i can better judge how i need to play.

[Nightbird]

Anomalocaris, on 21 November 2018 - 04:53 PM, said:

Match score isn't perfect. Some things (like hiding in cover, etc.) are rewarded that mean little (if you're doing something that helps you survive, that is rewarded in the end with a win, survival, etc.). But as ASH pointed out it's still a performance based metric and, on average, a player with a higher AMS is better than one with a lower AMS. If you sort by a combo of WLR and AMS you're going to do a much better job of balancing teams.

Look, we all know how PGI works. With the age of MWO and their future development priorities, this game is going to get minimal resource investment. Asking them to add sorting based upon WLR and AMS, two existing, archived statistics, is about as much as anyone could expect at this point. Asking for a new ranking system, etc. is pissing in the wind IMO.

Perfect isn't the issue, it's not a tool made for measuring skill, it's just gives out ribbons for people to feel good about doing something other than standing still.

[Khobai]

justcallme A S H, on 21 November 2018 - 02:30 PM, said:

As I said it is two parts.

• Fix the Tier structure so it accurately reflects player skill as opposed to an experience bar where even the bottom 40% of the playerbase can end up in Tier 1
  
• Fix the MM itself such that once 24 players are located for a lobby. It then balances the lobby to the best of its ability based on WLR / AMS

all that would accomplish is massively increasing queue times. youd be waiting all night for enough players to fill a game.

a better idea would just be to ban 99%ers from quickplay. If they cant play quickplay they cant screw up quickplay for everyone else. and they also cant complain about being too good for quickplay.

[El Bandito]

justcallme A S H, on 21 November 2018 - 11:59 AM, said:

So fixed means, perfect? Lol

You can absolutely make changes to the MM to, potentially, improve matches. It's not that hard to do and is a rather easy 2-Step process. PGI just refuse to come to that way if thinking unfortunately.

Right now just dumping 24 into a match based on "Tier" is all that happens. That means there is a huge room for improvement because the fundamental system it is based on is utterly broken. How you fail to comprehend that is mind-blowing.

Even if there was double the current population (so higher than it's been the last 2 years) there would be no perfect 'fix' for the MM which is what you seem to be after.

Seems you just wanna talk for the sake of talking, rather than to make a point. I already said in a previous post that even with 10x population, the MM cannot be fixed. It can be only improved, and I never said that I was against its improvement. You should read my posts with bit more care.

[Nightbird]

El Bandito, on 21 November 2018 - 06:14 PM, said:

Seems you just wanna talk for the sake of talking, rather than to make a point. I already said in a previous post that even with 10x population, the MM cannot be fixed. It can be only improved, and I never said that I was against its improvement. You should read my posts with bit more care.

https://www.dictionary.com/browse/fix

fix:
1. to repair; mend.
2. to put in order or in good condition; adjust or arrrange:
3 to make fast, firm, or stable.

The MM is broken, and if corrective measures are taken, it will be fixed.

[justcallme A S H]

Khobai, on 21 November 2018 - 05:45 PM, said:

[/list] all that would accomplish is massively increasing queue times. youd be waiting all night for enough players to fill a game.

a better idea would just be to ban 99%ers from quickplay. If they cant play quickplay they cant screw up quickplay for everyone else. and they also cant complain about being too good for quickplay.

As usual, horribly incorrect.

If the bell curve has the bulk of players in Tier 3 (which it should) there would be no real impact based on how the MM currently looks up/down 2 tier levels to find players.

So you either have no idea how it works now, or how a system should work, or both. Good thing we both know that it's both based on you literally making non stop unfounded claims without proof.

Do I really need to highlight them all, yet again? Still waiting for proof to I think 3-4 threads now that you bail on when facts come into a discussion.

El Bandito, on 21 November 2018 - 06:14 PM, said:

Seems you just wanna talk for the sake of talking, rather than to make a point. I already said in a previous post that even with 10x population, the MM cannot be fixed. It can be only improved, and I never said that I was against its improvement. You should read my posts with bit more care.

Nightbird covered that by simply going to the dictionary.

Edited by justcallme A S H, 21 November 2018 - 07:08 PM.

[justcallme A S H]

Nightbird, on 21 November 2018 - 03:58 PM, said:

Well, Tier 1s are better players than Tier 5s. So what? Match Score, assigning arbitrary weights to certain actions without any attempt to figure out how much that action benefits the team. Calling AMS a horribly inaccurate measuring stick is an insult to all the bad measure sticks out there.

Totally agree, definitely not great at all.

It's bad because of the things it weighs as important. Like a win or a loss should not count as much as it does. Attention to it would also let roles actually means stuff (like point capping, scouting etc).

AMS could be a reasonable tool, if attention was given to it. Given we already have it it is better than ground up and better chance of getting attention this late in the game life cycle.

Edited by justcallme A S H, 21 November 2018 - 07:12 PM.

[Grus]

justcallme A S H, on 21 November 2018 - 07:05 PM, said:

As usual, horribly incorrect.

If the bell curve has the bulk of players in Tier 3 (which it should) there would be no real impact based on how the MM currently looks up/down 2 tier levels to find players.

So you either have no idea how it works now, or how a system should work, or both. Good thing we both know that it's both based on you literally making non stop unfounded claims without proof.

Do I really need to highlight them all, yet again? Still waiting for proof to I think 3-4 threads now that you bail on when facts come into a discussion.



Nightbird covered that by simply going to the dictionary.

Why do i suddenly hear someone chanting *Jerry*Jerry *Jerry*

[Anomalocaris]

Nightbird, on 21 November 2018 - 05:26 PM, said:

Perfect isn't the issue, it's not a tool made for measuring skill, it's just gives out ribbons for people to feel good about doing something other than standing still.

Don't let the perfect be the enemy of the good. AMS is still a reasonable (if broad) indicator of the kind of player you're getting. As a modifier to WLR, it's far and away better than what we use now in the MM. Within the existing stat structure we have in MWO, you need a modifier to WLR because many people don't play enough matches each month to flatten out the impact of overly good or bad teams and random distribution.

[Grus]

Anomalocaris, on 21 November 2018 - 07:28 PM, said:

Don't let the perfect be the enemy of the good. AMS is still a reasonable (if broad) indicator of the kind of player you're getting. As a modifier to WLR, it's far and away better than what we use now in the MM. Within the existing stat structure we have in MWO, you need a modifier to WLR because many people don't play enough matches each month to flatten out the impact of overly good or bad teams and random distribution.

I think you have to play like 100to150+ per season to get a actual read on how well you are doing. Sadly that means 100 to 150 games in QP... *cringe*

It sucks if the main mode you play is FW..