23 replies to this topic

[FireStoat]

xe N on, on 02 April 2017 - 01:13 PM, said:

I did not state that there are no valves, but only there is no 3/3/3/3 rule anymore.

I have only been playing this game since September of last year, but a tonnage class restriction hasn't been present in the game since I started. A 3/3/3/3 rule? No way. On the Huntsman release day I am positive that I was in a match where my side had two full lances (8 mechs) of Huntsmen, and I was one of them. During the Marauder IIC release, I was one of seven (7) Marauder IIC's on my quick drop team. Anyone that played during the MAD IIC release week will recall exactly what I'm referring to.

Edited by FireStoat, 02 April 2017 - 01:34 PM.

[Mawcor]

FireStoat, on 02 April 2017 - 01:33 PM, said:

I have only been playing this game since September of last year, but a tonnage class restriction hasn't been present in the game since I started. A 3/3/3/3 rule? No way. On the Huntsman release day I am positive that I was in a match where my side had two full lances (8 mechs) of Huntsmen, and I was one of them. During the Marauder IIC release, I was one of seven (7) Marauder IIC's on my quick drop team. Anyone that played during the MAD IIC release week will recall exactly what I'm referring to.

I don't generally like to call people out for not reading, it seems to be more confrontational than productive most of the time, but Egoslayer specifically mentioned that that launch days were an exception. And of course they would be, everyone who bought it will want to try out their shiny new mech, and so flood that mech's weight class and leave the others emptier than usual. I've also found that that state of affairs only lasts for a day or two anyways, before the proportions normalize.

You are technically correct in that there is no hard 3/3/3/3 rule. I can't remember the last time I actually had a game that was exactly 3/3/3/3. But here's the thing, that's because it's not a hard rule but a soft one. Most of my solo games end up something like, 2/3/4/3, or 3/3/4/2, or 2/4/4/2. That is to say, each category is only one mech off. 3/3/3/3 is used as the baseline, to ensure some noticeable representation of each weight class (across several games, I have seen the occasional one-off game with no lights but it's not common). This also tends to curb huge weight disparities, as while you can still get one team heavier than the other, you won't get the 4/5/3/0-vs-1/2/3/6 sorts of situations I sometimes see in group queue.


As to the topic of the thread itself, I agree that the matchmaker could probably stand to be more robust than it is, but I'm not sure an 'avoid players' option would be a good idea. It's based on a problematic assumption, that the way a player happens to be playing when you encounter them is the way they always play. So, the idea that a player who gets 18 damage in a given game never breaks 100 damage. I remember a time recently where I ended a match with only two damage, because I pushed out too far ahead of the team, expecting some of my team to be right behind me and the enemy to not quite be as fast as they were. I typically get around 300-350 damage in a game though, just with the occasional outlying game where I get a lot more or less. So, the idea of giving people the option to block others based on what might be a single bad game doesn't make a lot of sense to me.

[BLOOD WOLF]

EgoSlayer, on 02 April 2017 - 11:45 AM, said:

Population is the issue, We have 5 tiers and four mech classes used in the criteria to build a match. Your simplistic view of how the match gets built is wrong. I again suggest you educate yourself on how things work before making wildly off base assumptions. The paging Karl Berg thread has details on exactly how the match maker builds the matches, and it's nothing like you described.

Simple example - say there are 2000 players in the queue(a extremely high estimate - since even at double steam count numbers there are ~3000 Playing - that's not in queue only most would be in matches) and you want to build a tier 3 match. With a rough bell curve of players (10/20/40/20/10) that means about 800 (40%) tier 3 in queue. Then of that 800 you have to apply 3/3/3/3 rule on mech class and see if you can build a match of 24. Now if you use a more realistic number like 800 total in queue (which is still 40% in queue 60% in matches) it's much harder to make a match. Now imagine the problems when your pool is on of the 10% categories and there are only 200 people that meet the tier requirement.

first of all nothing Karl berg wrote doesn't even address what I was talking about. He is talking more on match prediction outcomes. The MM is trying to make predictions on outcomes based on the given information.
I've read it over, and most of your points seem solid. It's very hard to have certainty here, that matchmaker ultimately kicks off several tens of thousands of games a day under all sorts of crazy input conditions. That's why I'd really consider Elo penalties for groups; we can use our telemetry to *prove* this will result in closer matches. For example, we can test the following, *if* groups of 2 win an average of 60% of the time, a 70 point Elo penalty will remove that benefit

Reading what he was writing about he was talking about setting parameters for the data that was coming in from the match results of teams and weighing certain variables. So first you said I Was wrong but, I said like he does, that the MM is kicking of thousands of matches, even at any given time. Second He is talking about ELO. I was just talking about the principles of Matchmaking. Nothing you said really went against what I said either, so I am not getting the part were I was wrong. I didn't make any wild assumptions. Without elo, the concept of a MM is really simple. I think you couldn't help yourself and quote me and make an *** out yourself.
I would say I didn't account for elo in my example but I was just talking about a MM and not necessarily the nuances. So stop assuming what you think I don't know. Next time make sure what you are saying actual rebuts me before you post.

Edited by BLOOD WOLF, 02 April 2017 - 06:12 PM.

[EgoSlayer]

BLOOD WOLF, on 02 April 2017 - 05:07 PM, said:

first of all nothing Karl berg wrote doesn't even address what I was talking about. He is talking more on match prediction outcomes. The MM is trying to make predictions on outcomes based on the given information.
I've read it over, and most of your points seem solid. It's very hard to have certainty here, that matchmaker ultimately kicks off several tens of thousands of games a day under all sorts of crazy input conditions. That's why I'd really consider Elo penalties for groups; we can use our telemetry to *prove* this will result in closer matches. For example, we can test the following, *if* groups of 2 win an average of 60% of the time, a 70 point Elo penalty will remove that benefit

Reading what he was writing about he was talking about setting parameters for the data that was coming in from the match results of teams and weighing certain variables. So first you said I Was wrong but, I said like he does, that the MM is kicking of thousands of matches, even at any given time. Second He is talking about ELO. I was just talking about the principles of Matchmaking. Nothing you said really went against what I said either, so I am not getting the part were I was wrong. I didn't make any wild assumptions. Without elo, the concept of a MM is really simple. I think you couldn't help yourself and quote me and make an *** out yourself.
I would say I didn't account for elo in my example but I was just talking about a MM and not necessarily the nuances. So stop assuming what you think I don't know. Next time make sure what you are saying actual rebuts me before you post.

The match maker isn't kicking off thousands of matches a minute. At 10,000 matches a day that would be ~ 7 matches a minute. Anything you see referring to elo in the match maker you can just replace with PSR, it's the same type of seeding method. They didn't rewrite the match maker they just use the PSR seed instead of elo. And frankly I'd be surprised if PSR didn't have the exact same scale behind the scenes as elo had to further simplify the transition.

Specifically the posts of note to the match maker seeding and building the match:
https://mwomercs.com...93#entry3695793
and
https://mwomercs.com...69#entry3696769

So yes, I stand by my original statement that your simplistic SC2 match making reference being equivalent to MWO is wrong. It's not even remotely similar and has zero bearing. You can call me names all you want, doesn't change anything. Or, you can keep it civil and not resort to name calling and defensive retorts.

Edited by EgoSlayer, 03 April 2017 - 03:33 PM.