Matchmaking Phase 4 - Feedback
#141
Posted 13 April 2013 - 11:24 AM
I feeel this would make more sense, and would greatly benefit groups that have more than 4 players online, but less than 8.
It would also dramatically reduce the unfairness that some ungrouped players feel when going up against an opposing force consisting of one or more premade groups of 4.
#142
Posted 13 April 2013 - 12:12 PM
Rashhaverak, on 13 April 2013 - 09:17 AM, said:
The first queue is for players that want to play as singles, with four players ( one lance) able to join a group (this is the same as four man is set up today).
The second queue is a team queue for players that want to play with, or play as, coordinated teams. This queue would allow any number of single players or groups to form teams up to eight (and up to twelve when the game switched to 12 vs 12). Players can make a group of 1-8, and any holes in the team would be filled by single players or smaller groups until a team of 8 is created. (This is a variation on how eight man is set up today).
Why change eight man to allow single and smaller groups to join if they wish? There are several good reasons:
- Some of us players want the challenge and opportunity to play in, and play with, more coordinated teams. We understand that we might be up against a full eight man team, while our own team might be made up of a group of five, a group of two, and a group of one, and we are okay with that. (In fact, we relish the chance to trounce that eight man and knock their ELO down a peg).
- Single players who want to experience more coordinated team play can do so, while other single players who want to play as individuals against other individuals can do so (without each others choices negatively affecting the other).
- Single players who want to try and meet other teams and other players will have the ability to do so in the team queue, while teams who are trying to build their ranks can play with, and play against, new talent.
- Allowing those players who want to play in more coordinated teams but can’t muster a team of eight can do so in the team queue without having to try and “sync drop” into the single player queue. When a sync drop works is isn’t as fun for the pugs, and when it doesn’t work it just frustrates the teams who are trying to sync it.
- It will help build a stronger MWO gaming community by bring together more teams on the battlefield. Right now I have a great group of friends who I play with, but it can be rare to get a full eight of us together. We end up having to split our friends up into a four man group and a three man group, or splitting into two groups of three, or worse, four in one group and one person doesn’t get to play or has to play alone. That isn’t fun, and isn’t encouraging us to play the game. Instead, it’s encouraging us to play something else that we can play together, or the odd man out end up not playing at all, since they can’t play in the group and don’t want to play alone.
- It will improve the time, variety and ability of the matchmaker to find and match up in the eight man team groups by increasing the number of players in the team queue. Right now it is not uncommon to match up multiple times with the same team, because at times there is not all that many eight mans out there. Letting the five and three group fill an eight would provide more groups for matching.
- When 12 vs 12 comes around, it will be even harder and less likely to find a full team of twelve. Changing the team queue to the above will make it much more likely that 12 vs 12 teams can find a match up.
No, these things are yet other restrictions with automated stuff to happen.
What this game needs is a single players only queue, forming up teams only from single pilots.
And on the other side a good old school lobby where you can open a room, form up your team AND you opponent team, chat a little, launch with your two teams and after match find yourself and the 15 ( later 23) other peopple in the same room with you again.
And for CW later on, a queue for any kind of MM, PGI sees fit for CW.
Anything else will always only cater for a fraction of the players, one time this fraction, one time an other. To make people happy, give them just simple 3 choices:
- fast match up as single pilot --> PUGGING as it should be.
- socialize in a lobby and play with and against your friends or other likeminded pilots, who like teamplay ( 1v1 up to 12v12, no restrictions)
- join up a CW match and fight over PGI's inner sphere map.
THIS ! anything below, will just create an other unhappy group with every change coming down the road.
#143
Posted 13 April 2013 - 12:20 PM
Zyne, on 13 April 2013 - 12:20 PM, said:
They would just make a 4 man DDC grp and a 4 man 3L grp,
They would sync drop which would put the heavyest grp and the lightest grp to gether (the sync drop)
The other team would be mixed of all types of mechs to meet the tonnage, and this team would not beable to sit down for a week after the match.
If they tryed to match mech weight and not class for class yhe system would be open to abuse, like the sync droppers are doing now.
They would just make a 4 man DDC grp and a 4 man 3L grp,
They would sync drop which would put the heavyest grp and the lightest grp together (the sync drop)
The other team would be mixed of all types of mechs to meet the tonnage, and this team would not beable to sit down for a week after the match.
Edited by Zyne, 13 April 2013 - 12:23 PM.
#144
Posted 13 April 2013 - 12:47 PM
Yokaiko, on 13 April 2013 - 11:10 AM, said:
20 is the difference between a single class....as in one.
Basically you are talking to normalize that you are talking an astronomical number of people in the queue, 16 mechs with 12 different weight possibilities, so that means 16 to the 12th power to normalize that. Its a 12 billion to one chance that you get 16 mechs of the same weight in the queue.
Wat?
You take the mechs available in the queue, then distribute them to two teams so that the total tonnage per team is within roughly 30 tons or so. Start small, increase the variance with time.
It's not at all difficult to do that with a smaller pool of players, particularly given how granular the tonnages are.
The number of mechs is irrelevant, only the tonnages - 12 different tonnages. Pick 8 (then 12) mechs to come out to a given average tonnage, and do that twice. Attempt to pick evenly from each weight class, but don't sweat it.
So, not considering Elo:
An example that tries to keep to a roughly 2/2/2/2 + 4 at random setup:
Pick an assault, a heavy, a medium and a light for each team, the ones that have been in the queue longest. Determine total tonnages. Do this again, using the next longest of each weight class in the pool, but in the second set assign mechs to teams to bring the averages closer together. For the third set, just pick a set of 4 mechs for each team from the pool at random irrespective of weight class to bring the average tonnages as closely as possible together.
This is not terribly complicated.
Edited by Wintersdark, 13 April 2013 - 12:50 PM.
#145
Posted 13 April 2013 - 12:51 PM
Wintersdark, on 13 April 2013 - 12:47 PM, said:
Wat?
You take the mechs available in the queue, then distribute them to two teams so that the total tonnage per team is within roughly 30 tons or so. Start small, increase the variance with time.
It's not at all difficult to do that with a smaller pool of players, particularly given how granular the tonnages are.
The number of mechs is irrelevant, only the tonnages - 12 different tonnages. Pick 8 (then 12) mechs to come out to a given average tonnage, and do that twice. Attempt to pick evenly from each weight class, but don't sweat it.
So, not considering Elo:
pick an assault, a heavy, a medium and a light for each team, the ones that have been in the queue longest. Determine total tonnages. Do this again, using the next longest of each weight class in the pool, but in the second set assign mechs to teams to bring the averages closer together. For the third set, just pick a set of 4 mechs for each team from the pool at random irrespective of weight class to bring the average tonnages as closely as possible together.
This is not terribly complicated.
Ok, program that loop without a determinate number of people in the queue, a 2 minute limit and fail conditions for 4 different light mech weights, two mediums, three heavy, and three assaults.
....all without a dedicated target weight, because you don't have one that has to be generated on the fly.
That is pretty much a Sophmore level programming project.
#146
Posted 13 April 2013 - 01:18 PM
Yokaiko, on 13 April 2013 - 12:51 PM, said:
Ok, program that loop without a determinate number of people in the queue, a 2 minute limit and fail conditions for 4 different light mech weights, two mediums, three heavy, and three assaults.
....all without a dedicated target weight, because you don't have one that has to be generated on the fly.
That is pretty much a Sophmore level programming project.
It's absolutely doable. You've got a lot of helpful data in your set, too: You know how many people are in the queue right now, you can quickly calculate the total numbers of L/M/H/A mechs. You CAN have a dedicated target weight, just one that isn't set in stone - the MM attempts to acheive a specified tonnage total/average, but isn't concerned about breaking it if there's a dispropotionate number of heavier or lighter mechs in the queue.
With a preset (and thus tunable) starting value, you can help "guide" the matchmaker to a relatively balanced loadout. If you're ideal match is 3/3/3/3, and a 75 ton delta between mechs, you'd want something roughly around 725 tons as a rough starting point.
You could adjust the Rough Starting Point before even looking at particular mechs by looking at the current queue L/M/H/A distribution, and reducing/increasing as necessary to help balance out the pool thus making future MM efforts easier and leading to more balanced matches overall.
Easy? No. Achievable? Absolutely. I'm not a professional programmer (though I'd often like to be, but alas that's just not in the cards) this is something that I know I could accomplish.
Incorporating Elo does increase the complexity, but not by so much as you'd imagine. It'd just be a matter of weighting mech choices: You set up your weighting on an sliding scale so you can adjust your choices based on Elo, Actual Tonnage, and Time In Queue. It's quite tuneable, then.
Not a simple piece of software, but it's absolutely doable. If I had a sample dataset to experiment with, I'd try my hand at building one purely for the fun of it.
#147
Posted 13 April 2013 - 01:25 PM
Wintersdark, on 13 April 2013 - 01:18 PM, said:
With a preset (and thus tunable) starting value, you can help "guide" the matchmaker to a relatively balanced loadout. If you're ideal match is 3/3/3/3, and a 75 ton delta between mechs, you'd want something roughly around 725 tons as a rough starting point.
This from the gang that can't keep missile damage within the realm of reason (twice now they have had to hotfix a missile patch) and after working on the ELO system for FIVE months (remember it was supposed to be out in December) borked the calculations so that it was bloating the upper tiers?
Call me skeptical.
That being said I wish they would put some form of tonnage max on the game, I'm sick of 4 atlas and 2 stalker games every damn time I try to drop.
I wouldn't care if they weren't being silly about group sizes, but even in 12 man games we are going to be capped at 2 in the puggable queue. I already can't find an 8 man unless I wake up early (my early, US prime for the rest) on the weekends.
#148
Posted 13 April 2013 - 02:02 PM
As always there are a few people who know how they think it should be done, its good to the see the passion for the game is still at a high level.
It'll be good to see how phase 4 works, and im sure it will bring more data to the table that will help with phase 5.
(yes nay-sayers, I am an optimist, but in my industry I deal with thousands of people daily so any change that with a down-line vision of improvement is a good thing. let it play out before shouting the sky is falling)
#149
Posted 13 April 2013 - 02:55 PM
Omid Kiarostami, on 12 April 2013 - 11:44 AM, said:
To clarify, it'll be similar to the forced Weight Class balancing that we used to have. We're not doing any tonnage based matching at this time.
:-(
Shoot, I was really happy for a day!
Ah, well, I didn't really want to play a blackjack, dragon, awesome, or light under 35 tons anyways!
#150
Posted 13 April 2013 - 02:58 PM
Roland, on 13 April 2013 - 11:24 AM, said:
I feeel this would make more sense, and would greatly benefit groups that have more than 4 players online, but less than 8.
It would also dramatically reduce the unfairness that some ungrouped players feel when going up against an opposing force consisting of one or more premade groups of 4.
Also, please don't do this ;-).
#151
Posted 13 April 2013 - 03:39 PM
Simply put, all 'mechs aren't considered equal on tonnage. If you go by your metrics on which variants are most taken and do the most damage / vs the ones that do the least, I propose the next best thing go BV for the new phase: Adjusted Tonnage.
What does this mean? Simple: The Matchmaker could look at a Cataphract-2X as a 65 Ton 'mech, or a Cataphract-3D as a 75 ton 'mech. Obviously those are just example numbers I've pulled up that could be adjusted as necessary.
But I think it would greatly help balance matches, in particular if this information was transparent to the player (so they could plan accordingly), but even without that aspect, this would great help the typical match.
#152
Posted 13 April 2013 - 04:05 PM
Edited by Diablobo, 13 April 2013 - 04:08 PM.
#153
Posted 13 April 2013 - 04:15 PM
The implementation of these classifications could be as simple as a counter that tallies the categories. If a mech is equipped with ECM (or the other categories), add one to the counter for the team. Then, try and balance the teams so that one team does not have a greater tally count than the other.
Edited by Diablobo, 13 April 2013 - 04:54 PM.
#154
Posted 13 April 2013 - 04:31 PM
Edited by armyof1, 13 April 2013 - 04:32 PM.
#155
Posted 13 April 2013 - 05:24 PM
armyof1, on 13 April 2013 - 04:31 PM, said:
Agreed. New players coming in should have their Elo artificially lowered. It can be restored to its real value once they're out of the cadet phase. This would have the added benefit of helping to ensure that players in trial mechs end up dropping against each other instead of getting mixed in with the more experienced guys in real mechs. It makes their matches more fun and our matches more competitive.
EDIT: Actually, an even better fix would be to make dropping in a trial mech automatically lower your Elo regardless of matches played. That way new players who may have one good mech but are still potentially relying on trials to keep some variety going won't get hosed.
Edited by TOGSolid, 13 April 2013 - 05:27 PM.
#156
Posted 13 April 2013 - 06:11 PM
Diablobo, on 13 April 2013 - 04:05 PM, said:
It's all about balance ... the best thing about the blind match maker (not knowing the enemy's composition) is that every choice a team makes is a risk. With random PUG games, the first risk you take as a player is choosing your mech ... do you bring a slow but hard hitting Assault, or a fast but relatively weak Light, or something in between?
The most significant risk decision, though, is as a 80-120 kph mech (especially on Alpine and Tourma), do you stick with the heavyweights and bring your firepower (but weak armor) to the almost inevitable battle for Epsilon (on Alpine) or Theta (on Tourma), or do you try to help out the Speedsters with the outlying capture points? For the Lights, do you wait for your slower reinforcements, or do you go for the early capture advantage?
These risk decisions are one of the things that make the game interesting and exciting to me.
#157
Posted 13 April 2013 - 06:20 PM
Kageru Ikazuchi, on 13 April 2013 - 06:11 PM, said:
The most significant risk decision, though, is as a 80-120 kph mech (especially on Alpine and Tourma), do you stick with the heavyweights and bring your firepower (but weak armor) to the almost inevitable battle for Epsilon (on Alpine) or Theta (on Tourma), or do you try to help out the Speedsters with the outlying capture points? For the Lights, do you wait for your slower reinforcements, or do you go for the early capture advantage?
These risk decisions are one of the things that make the game interesting and exciting to me.
Likewise for me.
I personally love my 100kph DRG-FLAME on Alpine and Tourmaline for this. It's a blast to rip off to grab caps, knowing if just a single light headed there first, while he'll start capping before I get there, I'll either chase him off or flat out kill him, but either way, I will get that cap point.
Then I can head to another cap point, hold mine, or rejoin the battle, depending on the situation.
#158
Posted 13 April 2013 - 09:12 PM
Jenner,Hunchback,Awesome,and Atlas. These mechs can not be modified and players will be randomly assigned a mech at the start of the match so each team has the same mechs. No premades.
Matches are now balanced.
Edited by Fabe, 13 April 2013 - 09:37 PM.
#159
Posted 13 April 2013 - 09:21 PM
Asakara, on 12 April 2013 - 12:54 PM, said:
Bell curves on top of bell curves actually look like bell curves most of the time (visually). If they actually spotted it via a statistical analysis, that's encouraging because it means they might actually have a statistician on the team to try and get this stuff right.
armyof1, on 13 April 2013 - 04:31 PM, said:
The way ELO works in general, the entry point is the average player point, since it is a zero sum system. The only exception to this is if there's inflation due to people with poor ELO's dropping the game at a faster rate than those with higher ELO's.
#160
Posted 13 April 2013 - 10:09 PM
ohtochooseaname, on 13 April 2013 - 09:21 PM, said:
The way ELO works in general, the entry point is the average player point, since it is a zero sum system. The only exception to this is if there's inflation due to people with poor ELO's dropping the game at a faster rate than those with higher ELO's.
My point is that is not a good way to do it. How many totally new players do you think would qualify to be considered an average player in his first 100 or so matches? If I say 1 out of 10 I think I'd be optimistic. But while the 9 other guys are trying to just walk around without bumping into things, they are being put in teams with players that can both maneuver and shoot fairly well and frankly dragging down that team because of their inaccurate Elo rank. Just give them a generally more accurate Elo ranking and let them start from there instead. If they're better than below average they'll do well in matches and climb up quite fast in ranking anyway.
21 user(s) are reading this topic
0 members, 21 guests, 0 anonymous users