76 replies to this topic

[Deathlike]

PhoenixFire55, on 17 April 2017 - 04:15 AM, said:

So much sarcasm, but sadly you've entirely missed the point of my post.

Creating a matchmaker isn't simple, nor did I "solve" it anywhere. However, saying that low population is the reason your MM isn't working is a lame excuse, which is what I've proven.

Low population IS the core reason for this though.

Think of it this way.

Let's assume that Tier 1 is filled with mostly actual good players, and players that play the game too much (which is kind of the case right now).

If Tier 1 is 50% of the active population AT THE TIME of the MM, then you're going to have issues getting balanced matches. Good T1 players are mixed with bad T1 players, and consequently you're going to get screwed up skill levels based on the players selected.

But let's really talk about the actual flaws, because you're forgetting a few aspects about PGI's MM.

1) Tier 1 is Tier 1. Some people may be at the lower end of Tier 1, but there's a pretty likely chance that PGI's definition is just "is this guy in Tier 1" and forgets about the actual values involved (whether that person is at the highest end of T1 or lowest end of T1). It's just easier to simplify w/o math.

2) Teams are constructed as is for the MM waiting for "the equivalent side" to be created. What I mean by this is that let's say it decides to group an 8-man with a 4-man. Well, the MM doesn't bother "deconstructing" that group that's waiting in the queues to say an 8-man and 2 2-mans, or replace the 8-man with a 5-man and 3-man to make a different by close approximation of the group sizes. In essence, the MM doesn't attempt to reshuffle the deck in order to make a "better match". This is problematic for just basic MM principles because if you're that group that perfectly matches with an SJR/EmP deathsquad, you may have won the lotto (or not because you ended up on the other side).

3) The group queue doesn't factor in tonnage (it's only enforced in group creation, not in MM factoring), which is primarily responsible for tonnage mismatches. But worse is that the group queue tends to have players in MULTIPLE TIERS, and to attempt matching "that grouping" is difficult... whether it is a group with a 4-man deathsquad or a 8-man group of PUGs (yes, they exist in the queues).. you will have trouble just matching "perfectly" with that kind of setup.


In summation, point #2 is the bigger/biggest problem in the MM, where constructing a team does not get any reshuffling consideration (maybe outside of the group quitting out of the queues) and tries to "magically" fit everything together by those construction rules (and doing a crap job of it most of the time).

[MrJeffers]

PhoenixFire55, on 17 April 2017 - 10:02 AM, said:

Except two of the three you mentioned aren't part of any criteria and aren't part of the MM at all. And TBH I'm not sure time spent in queue is either.

Once again, you are not reading and thus missing the entire point. I'm not building a matchmaker, I am simply demonstrating how blaming any sort of MM fails on low population is entirely false and nothing but a lame excuse.

You're wrong on all counts. The first criteria to seed the match is time in queue.
From there it has to build a match using that first person as the seed, so their tier and mech class are used then to start populating teams in a round robin format. It tries to build a match first matching tier level while matching the 4x3 mech class limits, that is the first relief valve, and if the match still can't be made the second relief valve is tier ranking.
https://mwomercs.com...93#entry3695793


Your "proof" is not showing anything except that you don't understand how the match maker works. It is in fact trying to do what you are saying, but there are other criteria that make it exponentially more difficult to do so. Which is exactly why *low population* is the problem with creating matches. There are not enough people waiting in the queue that meet all the criteira to build a "perfect" match.

[PhoenixFire55]

Deathlike, on 17 April 2017 - 11:17 AM, said:

Low population IS the core reason for this though.

Think of it this way.

Let's assume that Tier 1 is filled with mostly actual good players, and players that play the game too much (which is kind of the case right now).

If Tier 1 is 50% of the active population AT THE TIME of the MM, then you're going to have issues getting balanced matches. Good T1 players are mixed with bad T1 players, and consequently you're going to get screwed up skill levels based on the players selected.

No. Nothing to do with population. I've given an example of a perfectly balanced match with absolutely random players of any skill levels. For that perfect balance I needed exactly 24 people. Any people.

Now if as I've also said the criteria that assigns "skill" levels to players is faulty (which is clearly the case with tiers and PSR), then the unbalance does start to show up, however, that is a case of a broken criteria, which has nothing to do with player population.

Deathlike, on 17 April 2017 - 11:17 AM, said:

But let's really talk about the actual flaws, because you're forgetting a few aspects about PGI's MM.

We've been over this countless times, I don't want to go over it all again. The only point I was proving is that blaming low population is nothing but a lame excuse. 24 people of any skill level, potatoes, ubercomp pros, doesn't matter, is all you need in terms of numbers to make a balanced match. Anything else is a developer failure to figure out a proper skill-evaluating criteria, failure to make a proper matching algorithm or whatever else that has nothing to do with the amount of players.

[PhoenixFire55]

MrJeffers, on 17 April 2017 - 11:40 AM, said:

Your "proof" is not showing anything except that you don't understand how the match maker works ...

You fail at reading over and over again. Since you clearly don't care about what others are trying to say feel free to keep talking to yourself and stop bothering me.

[MrJeffers]

PhoenixFire55, on 17 April 2017 - 11:52 AM, said:

You fail at reading over and over again. Since you clearly don't care about what others are trying to say feel free to keep talking to yourself and stop bothering me.

Sorry, the reading comprehension fail is on your end, since you don't understand there is more involved in creating a match than just the skill level rating. You can keep repeating that it's simple and not a population issue, but it will never make it true if you do the math. So I'll ask you to stop spouting misinformation that bothers the rest of us.

[BLOOD WOLF]

PhoenixFire55, on 17 April 2017 - 10:02 AM, said:

Except two of the three you mentioned aren't part of any criteria and aren't part of the MM at all. And TBH I'm not sure time spent in queue is either.



Once again, you are not reading and thus missing the entire point. I'm not building a matchmaker, I am simply demonstrating how blaming any sort of MM fails on low population is entirely false and nothing but a lame excuse.

The problem is that you are incorrect that the MM doesn't have the other criteria that jeffers mentioned. which I am sure it has. From what I remember, when time in que is increase the MM relaxes restrictions. Meaning it changes it's criteria to meet the match player requirement. So its gonna pull 2 dire wolf's more than necessary if it chooses. These are extra dimensions.

You are right as far as blaming low population is not even remotely accurate and is pretty much a lame excuse to put as an explanation to anything.

MrJeffers, on 17 April 2017 - 11:58 AM, said:

but it will never make it true if you do the math. So I'll ask you to stop spouting misinformation that bothers the rest of us.

well sire, building models is something I can actually do with mathematics. I can make any number of scenarios in which a mm will act. I can use any number of people also as long as there is minimum of 2(although that would be silly).

Edited by BLOOD WOLF, 17 April 2017 - 12:03 PM.

[PhoenixFire55]

MrJeffers, on 17 April 2017 - 11:58 AM, said:

Sorry, the reading comprehension fail is on your end, since you don't understand there is more involved in creating a match than just the skill level rating. You can keep repeating that it's simple and not a population issue, but it will never make it true if you do the math. So I'll ask you to stop spouting misinformation that bothers the rest of us.

Nothing of what I've shown in my example is changed in any way regardless if any sort of other stuff like time-in-queue, mechs, ranks, or whatever else you come up with. Rename the "skill" level from said example into "matching criteria" or "overall criteria" or "magic value no.33", it doesn't change a single thing.

As for misinformation ... go and study up yourself before discussing things you don't understand nor trying to.

BLOOD WOLF, on 17 April 2017 - 11:58 AM, said:

The problem is that you are incorrect that the MM doesn't have the other criteria that jeffers mentioned. which I am sure it has. From what I remember, when time in que is increase the MM relaxes restrictions. Meaning it changes it's criteria to meet the match player requirement. So its gonna pull 2 dire wolf's more than necessary if it chooses. These are extra dimensions.

Irrelevant. Read just above.

[MrJeffers]

PhoenixFire55, on 17 April 2017 - 12:08 PM, said:

Nothing of what I've shown in my example is changed in any way regardless if any sort of other stuff like time-in-queue, mechs, ranks, or whatever else you come up with. Rename the "skill" level from said example into "matching criteria" or "overall criteria" or "magic value no.33", it doesn't change a single thing.

As for misinformation ... go and study up yourself before discussing things you don't understand nor trying to.



Irrelevant. Read just above.

Actually when you put the "match making criteria" into a multi-dimentional array it does completely change everything. So maybe you should go study more.

EDIT; Take your spread of skill numbers and then randomly append to each one with a number representing mech class. 1, 2, 3, 4, e.g. (493:1) is a 493 skill person in a light mech.
Now 1s are 10% of the skill numbers, 2s 25% of your population, 3s are 35%, and 4s are 30%, just for rough numbers. Now try and bulid your perfect match.

Edited by MrJeffers, 17 April 2017 - 12:16 PM.

[BLOOD WOLF]

PhoenixFire55, on 17 April 2017 - 12:08 PM, said:

what?.....no dude. its you who doesn't understand

MrJeffers, on 17 April 2017 - 12:09 PM, said:

Actually when you put the "match making criteria" into a multi-dimentional array it does completely change everything. So maybe you should go study more.

yep.

Edited by BLOOD WOLF, 17 April 2017 - 12:13 PM.

[Deathlike]

PhoenixFire55, on 17 April 2017 - 11:49 AM, said:

No. Nothing to do with population. I've given an example of a perfectly balanced match with absolutely random players of any skill levels. For that perfect balance I needed exactly 24 people. Any people.

Now if as I've also said the criteria that assigns "skill" levels to players is faulty (which is clearly the case with tiers and PSR), then the unbalance does start to show up, however, that is a case of a broken criteria, which has nothing to do with player population.

That assumes that you can accurately rate the players (and PSR doesn't even remotely do that). Sure you can literally do the algorithm as listed, but that would ONLY apply to solo drops. I'm talking a lot more about group drops, where you can't really deconstruct the numbers and magically it would just work.

One number will not magically describe a player's skill, let alone the fools that want BV for mechs. It doesn't just work as is.

Quote

We've been over this countless times, I don't want to go over it all again. The only point I was proving is that blaming low population is nothing but a lame excuse. 24 people of any skill level, potatoes, ubercomp pros, doesn't matter, is all you need in terms of numbers to make a balanced match. Anything else is a developer failure to figure out a proper skill-evaluating criteria, failure to make a proper matching algorithm or whatever else that has nothing to do with the amount of players.

I think you totally overestimate how an MM works. I mean, if you break it down to its finer points... it may "look easy", but reality/practicality says that one specific number literally doesn't describe a player's skill level. You have to remember that this same number is being used in BOTH solo and group... which both have completely different dynamics and usually are not indicative based on the specific queue (some people are better group players than solo and vice versa - not that there can't be people that are good or bad at both).

In the literal sense, you do need a significant enough population to have an effective MM. You could try and "implement" your concept but you're still going to run into the same problems (if not worse) than what we already experience.



You have to remember that there is a multiplicative/force factor where more skilled players in a group will have a greater impact than just random players (even if they are Tier 1) that are in different actual groups. I'm seriously saying that you cannot boil an MM down to a "simple" math equation or algorithm. It can realistically get close, but again... you have to have the population at the moment in question to service the balance in teams.

Edited by Deathlike, 17 April 2017 - 12:35 PM.

[PhoenixFire55]

MrJeffers, on 17 April 2017 - 12:09 PM, said:

Actually when you put the "match making criteria" into a multi-dimentional array it does completely change everything. So maybe you should go study more.

EDIT; Take your spread of skill numbers and then randomly append to each one with a number representing mech class. 1, 2, 3, 4, e.g. (493:1) is a 493 skill person in a light mech.
Now 1s are 10% of the skill numbers, 2s 25% of your population, 3s are 35%, and 4s are 30%, just for rough numbers. Now try and bulid your perfect match.

LOL, that's the best you could come up with? Once again clearly showing that you don't understand a thing ... Missing the one simple point over and over and over again.

It doesn't matter who the players in question are, doesn't matter from what "ranks" or "tiers" they are, doesn't matter what mechs they bring and what loadouts they use and whatever else you can think of. Any player in any mech has a limited amount of impact he can have on a match. That amount is determined by his skill, by his mech, by his loadout, by whatever. The important thing is that it all translates into one thing, one amount. Properly evaluating that amount is not a trivial task, but never did I claim that I provided or even mentioned anything about how it should be done. That is a task that any developer creating a MM system solves, or in our case fails to. That is why matches are unbalanced. Because as long as there is a proper way of evaluating the impact a player makes, the amount of players becomes irrelevant and all it takes for a balanced 12v12 match is 24 people of any skill, in any mechs and whatever else.

In your silly example with mech classes you can simply assign a multiplier for each mech class, or each mech chassis, or each mech variant, or even each mech loadout that together with the "skill" level will combine into "magic value no.33" that the MM will in turn use. These multipliers can be even calculated by the MM itself based on average performance of said mechs, their variants or loadouts, or individual performance of any particular player in said mechs, variants and loadouts. It takes nanoseconds for modern PCs to perform such operations. But once again, this isn't what I am talking about here at all.

BLOOD WOLF, on 17 April 2017 - 12:10 PM, said:

what?.....no dude. its you who doesn't understand

yep.

Oh great, now you are going to pretend that you understand all about multi-dimensional arrays as well.

For starters give me a definition of it. Then provide an MWO-relevant example of such multi-dimensional array of criteria that can be used for MM. Then maybe we can talk.

[BLOOD WOLF]

PhoenixFire55, on 17 April 2017 - 12:47 PM, said:

Oh great, now you are going to pretend that you understand all about multi-dimensional arrays as well.

For starters give me a definition of it. Then provide an MWO-relevant example of such multi-dimensional array of criteria that can be used for MM. Then maybe we can talk.

your really talking functions. pieces of information such as {0,1..............................{x,x} {x,x,x,x,x,x}, any number really, or words. All we need to know is that it represents something.

that has nothing to do with anything.

or is it necessary for anyone to understand the basic principles of a matchmaker. as you can use simpler mathematical principles to build one without relying on functions of variables. They can make it as simple or complex as they want, barring its capable within the logic

Edited by BLOOD WOLF, 17 April 2017 - 01:03 PM.

[PhoenixFire55]

Deathlike, on 17 April 2017 - 12:32 PM, said:

That assumes that you can accurately rate the players (and PSR doesn't even remotely do that). Sure you can literally do the algorithm as listed, but that would ONLY apply to solo drops. I'm talking a lot more about group drops, where you can't really deconstruct the numbers and magically it would just work.

I can't. PGI can't either. But that is their job, not mine. And yes, everything that is being discussed here is discussed for solo queue and 12v12 matches.

Deathlike, on 17 April 2017 - 12:32 PM, said:

One number will not magically describe a player's skill, let alone the fools that want BV for mechs. It doesn't just work as is.

What is MM trying to do? ... Match players according to skill. Skill is one thing, one value. How many other values you use in order to figure out this one is irrelevant. Player individual skill, his ability to work in a team, his mechs and loadout, it all translates into one thing ... the impact this player makes in a match. That impact must be evaluated and translated into a single value MM uses. Anything else is not a skill-based MM to begin with.

Deathlike, on 17 April 2017 - 12:32 PM, said:

In the literal sense, you do need a significant enough population to have an effective MM. You could try and "implement" your concept but you're still going to run into the same problems (if not worse) than what we already experience.

Once again, no. Not as long as I have a proper skill-evaluating criteria. If I do need significant population it simply means I don't have any sort of an even close to proper criteria.

Deathlike, on 17 April 2017 - 12:32 PM, said:

It can realistically get close, but again... you have to have the population at the moment in question to service the balance in teams.

I need 24 people, since there is no match unless there are at least 24 people it means that population at the moment in question (or any moment) is irrelevant for balancing.

[PhoenixFire55]

BLOOD WOLF, on 17 April 2017 - 12:53 PM, said:

that has nothing to do with anything.

or is it necessary for anyone to understand the basic principles of a matchmaker

That has everything to do with everything. Also, basic school education is required to understand principles of a matchmaker or any principles at all.

FYI ... computers and algorithms do not operate in random terms like "skill" and other fancy words you keep saying, they operate with values, numbers as you put it. That goes for that little thingy you call MWO as well. Surprise surprise.

[BLOOD WOLF]

PhoenixFire55, on 17 April 2017 - 12:58 PM, said:

That has everything to do with everything. Also, basic school education is required to understand principles of a matchmaker or any principles at all.

FYI ... computers and algorithms do not operate in random terms like "skill" and other fancy words you keep saying, they operate with values, numbers as you put it. That goes for that little thingy you call MWO as well. Surprise surprise.

dude, the only person who used fancy words was you, and I never said the MM operates in any sort of way.

"multi dimensional array"(your word), sure does sound fancy.

I just said that it's possible to build an MM, with such and such properties. You can make it as complex to account for as much as you want, or make it account for as little as you want.

PhoenixFire55, on 17 April 2017 - 12:58 PM, said:

FYI ... computers and algorithms do not operate in random terms like "skill" and other fancy words you keep saying, they operate with values, numbers as you put it.

I understand algorithms. Computer logic, as some would put it. in other words, when we say "skill", there is a line of code that follows processes.

Edited by BLOOD WOLF, 17 April 2017 - 01:11 PM.

[MischiefSC]

No, population depth is critical. It can only match who is available. It tries to account for skill and tonnage. Often there is no good balance involving those factors.

[Deathlike]

PhoenixFire55, on 17 April 2017 - 12:55 PM, said:

I can't. PGI can't either. But that is their job, not mine. And yes, everything that is being discussed here is discussed for solo queue and 12v12 matches.

You still already have flaws in your implementation. I'll try to break it down in other sections.

Quote

What is MM trying to do? ... Match players according to skill. Skill is one thing, one value. How many other values you use in order to figure out this one is irrelevant. Player individual skill, his ability to work in a team, his mechs and loadout, it all translates into one thing ... the impact this player makes in a match. That impact must be evaluated and translated into a single value MM uses. Anything else is not a skill-based MM to begin with.

You're making a few assertions though. It's not like it's one player in like CS:GO that has access to whatever. You have to have AT LEAST picked a mech. Someone picking a Mist Lynx is far different from someone else picking a Kodiak-3. It's like trying to put Proton in the worst mech possible vs some T5 unicorn in a Kodiak-3. They are drastically different aspects that you're just holding onto just 1 number for. It doesn't work.

So in your MM example, what if the result is having 12 Lights on one side and 12 Assaults on the other? If we go STRICTLY by the numbers, such a result can happen. While obviously my example is an extreme one, but that already happens in the solo queue as is. What you forget or mistake is that you assume that the distribution of mechs/tonnage is relatively even... which still requires more people/population and you can't magically lump them all in the same boat. It just makes the MM result much worse and broken.

Quote

Once again, no. Not as long as I have a proper skill-evaluating criteria. If I do need significant population it simply means I don't have any sort of an even close to proper criteria.

You can't really excuse it either. It's like saying, everything is broken and we should keep it broken because it's fine. Yea, that'll go far.... unless you want to be taken seriously.

Quote

I need 24 people, since there is no match unless there are at least 24 people it means that population at the moment in question (or any moment) is irrelevant for balancing.

It's not just 24 people, it's 24 mechs and not everyone is necessarily as effective in every mech (remember, there will be people grinding in mechs they are unfamiliar with or in a class they are not as good at). Simplifying and stripping relevant criteria is only going to make the MM result worse... unless you don't care about the result (which is already what PGI is doing anyways).

I feel like I have to describe math and it's not like PGI is good at it. I'm telling you that trying to do what you are doing is going to result in literally more whining even if "the numbers are right". I talk mockingly about telemetry because you literally have to understand it before you really dive deep into the numbers. Otherwise, a Scat gets nerfed "for reasons" other than it's the last one to die (it doesn't have firepower to do much and hides naturally due to ECM, but PGI doesn't seem to know this).

There's a reason why filters exist (like even the region filters) affects how effective an MM is. If only a 1/3 of the population actively runs Oceanic (which doesn't ever happen for me, since I totally deselect it), it affects the quality of your results much more significantly... and it would still be the case for tonnage/weight classes in these examples. It's not even that hard to find 6 Assaults on one side vs 0 to 2 Assaults on the other side in a normal group queue match. These are things that will get people whining about stuff that much more readily.

Edited by Deathlike, 17 April 2017 - 01:12 PM.

[PhoenixFire55]

BLOOD WOLF, on 17 April 2017 - 01:00 PM, said:

dude, the only person who used fancy words was you, and I never said the MM operates in any sort of way.

"multi dimensional array"(your word), sure does sound fancy.

Yeah, right ...

BLOOD WOLF, on 17 April 2017 - 12:10 PM, said:

what?.....no dude. its you who doesn't understand

MrJeffers, on 17 April 2017 - 12:09 PM, said:

Actually when you put the "match making criteria" into a multi-dimentional array it does completely change everything.

yep.

I mean, sounds like you agree with the guy. Or was it just clueless "yep" in the middle of nowhere?

MischiefSC, on 17 April 2017 - 01:09 PM, said:

No, population depth is critical. It can only match who is available. It tries to account for skill and tonnage. Often there is no good balance involving those factors.

There is always a good balance as long as you properly account for all these factors. And when you do, you need exactly 24 people for a balanced match. The fact that devs can't figure out how to properly account for such factors have nothing to do with player population.

[BLOOD WOLF]

PhoenixFire55, on 17 April 2017 - 01:12 PM, said:

Yeah, right ...



I mean, sounds like you agree with the guy. Or was it just clueless "yep" in the middle of nowhere?

I agreed with some of what he said but the other half I did not. Same with the other guy. nice to know people aren't fully reading what is typed if longer than a few senteces.

It also looks like you took my quote out of context. I was saying "you don't understand" to you, not jeffers. You made it look like I was saying that to mr jeffers. Or am I reading that wrong?

Edited by BLOOD WOLF, 17 April 2017 - 01:19 PM.

[MischiefSC]

To clarify - suppose you're on an assembly line getting random fruit. You're trying to put them in packages with everything in pairs - 2 apples, 2 oranges, etc. Chute opens and..... you get 7 potatoes, 1 kiwi, 3 oranges, 5 assorted apples (different types), a banana, a cherry, 1 small and 1 large wedges of cheese, and 3 flavored condoms.

Now put that into 2 equal, balanced sets.

If you got a bigger pool so you always had even numbers you could do so but when you've got to make "sets" out of odd numbers of odd assortments you're going to end up with issues.