178 replies to this topic

[Nightbird]

AnAnachronismAlive, on 19 December 2020 - 04:41 AM, said:

Since some kind of matchmaker (re)work seems to be on the table anyhow and officials got pointed towards this thread yesterday, mebbe someone can place some sort of a comprehensible and goal-oriented abstract into the applicable Command Chair Thread => Modes

Pretty, pretty please: no salt, no bad blood, no accusations (not even subliminal ones, ye smarties!) - but a goal-orientied explanation why what we have now does not work and what else might work out better.

Don't get your hopes up. Almost no ideas on this forum are backed up by facts, most of it is wishful thinking, and we live in an age where these people still want the same respect as proper ideas. It's a recipe for entertainment, not working ideas.

[ThreeStooges]

All you need to know about playing solo in qp is this: first team to 3 kills wins. always.

[crazytimes]

Nightbird, on 19 December 2020 - 01:40 PM, said:

Don't get your hopes up. Almost no ideas on this forum are backed up by facts, most of it is wishful thinking, and we live in an age where these people still want the same respect as proper ideas. It's a recipe for entertainment, not working ideas.

Surely you jest kind sir!

They had a round table and everything where people talking about their opinions on ideas that may be changed at some point! Meanwhile, the people with the access to all the actual data and analysis... yeah, they're not so full of ideas.

[AnAnachronismAlive]

Nightbird, on 19 December 2020 - 01:40 PM, said:

Don't get your hopes up. Almost no ideas on this forum are backed up by facts, most of it is wishful thinking, and we live in an age where these people still want the same respect as proper ideas. It's a recipe for entertainment, not working ideas.

To keep on walking despite the kinda likely assumption of our actions not having any meaningful impact (Hello, existential nihilism! How have you been?) in the end is what makes us go on and on and on and on in life pretty often, don't you think?

Edited by AnAnachronismAlive, 19 December 2020 - 10:59 PM.

[morosis]

AnAnachronismAlive, on 19 December 2020 - 04:41 AM, said:

Since some kind of matchmaker (re)work seems to be on the table anyhow and officials got pointed towards this thread yesterday, mebbe someone can place some sort of a comprehensible and goal-oriented abstract into the applicable Command Chair Thread => Modes

Pretty, pretty please: no salt, no bad blood, no accusations (not even subliminal ones, ye smarties!) - but a goal-orientied explanation why what we have now does not work and what else might work out better.

two points here. first of all, i have brought up the issues with soup and the matchmaker roughly a dozen times on every type of forum that has been had so far. i brought it up during the initial meeting with PGI, through highly respected community members, through NGNG, through my formal feedback, and now here. so far, every time i bring up how bad soup experience often is, the response has been either of two:

1) "soup is a good solution to the low population issue, and it could potentially be revisited once there is no longer a low population issue".
2) "the experience of certain players who play at certain times does not corroborate what you are saying, so your experiences must be an outlier."

to me, both of the above are non-answers, because in case 1 the population is very unlikely to recover when the biggest aspect of playing a game (i.e. how enjoyable the experience is), is the very thing being eroded. and in case 2 the experiences of certain people who play at certain better times is not a proof point that the current system is good at all. it just means the current solution can work as long as all the game is operating in a best-case scenario. the soup solution occasionally working as intended does not mean it is working sufficiently well.

finally - i think my feedback is fairly clear here, and has been suggested by quite a few others as well. secondary lobby balancing. nothing further is needed to correct the issue i am highlighting, and in fact secondary lobby balancing actually helps many concurrent aspects of feedback re: group play distorting PSR, game balance, tonnage balance, etc.

by this i mean, if we are assuming that any proposed solution must:
- allow for both groups and solos to play in the same 24 man match
- balance pilot skill as much as possible across both teams
- balance 3-4 mans on both ends of the PSR bell curve simultaneously
- account for tonnage
- account for wait times
- not require another pass at PSR calculations
- not require another population PSR reset
- not require any additional labor intensive recoding of matchmaker functionality

then essentially the constraints of the problem dictate a solution that must be able to balance the actual 24 players and mechs selected, as opposed to attempting to pick the most balanced 24 players and mechs out of the overall population. i think the former is what would produce the better gameplay experience anyway.

assuming that the matchmaker picks 24 players in the same fashion it picks them today, then makes decisions about how to allocate those players/mechs to team 1 and team 2 according to their PSR and their chosen mech, regardless of whether they are a group player or a solo player then we should see the following benefits:
- players who have been "carried" to high WLR/KDA/PSR through group play will be normalized
- solo players who have been at a disadvantage should see their gameplay experience improve dramatically.
- groups will get split at low population times but should still be able to play together when population allows.
- players who lose in a stomp will know that the loss was due to poor play as opposed to poor matchmaking.
- secondary lobby balance can be done w/in the 75 seconds currently held for map vote and connection time

Edited by morosis, 20 December 2020 - 07:51 AM.

[Nightbird]

AnAnachronismAlive, on 19 December 2020 - 10:55 PM, said:

To keep on walking despite the kinda likely assumption of our actions not having any meaningful impact (Hello, existential nihilism! How have you been?) in the end is what makes us go on and on and on and on in life pretty often, don't you think?

In no way am I asking for people to stop, that would be a meaningless endeavor on my part. I'm just making an observation of cause and effect.

Just recently there was a presentation with suggestions and at the end it projected it would boost pop by 4x if implemented. There was no data on who it would help and how much, no analysis on why a particular suggestion would help a certain amount, just - this is our suggestion do it and it will help this much. The forum is filled with such ideas. Unfortunately, we're not in game, but in real life, with costs and consequences associated with every decision that is made.

Because of finite resources available be it time or money, you have to pick the limited actions that you can afford to maximize the benefits to your self/company. PGI has proven to be bad at this, they often implement updates that appeal to a small portion of the population by neglecting updates that would appeal to the larger population. They seem to be able to do cost analysis, how much an update to cost, but they never seem to analyze what proportion of the pop it helps, and to what degree.

This latest round of community suggestions have the same problem, suggestions do not analyze and project benefit scope and degree. As a result, most suggestions have negligible benefits. Regardless of what random bundle of ideas PGI implements from the pool, the budget being limited as it is, the impact of those dollars will be so diluted that the conclusion is already decided.

Edited by Nightbird, 20 December 2020 - 11:04 AM.

[Nightbird]

morosis, on 20 December 2020 - 03:47 AM, said:

matchmaker

Everyone agrees on what the result should be, a better MM. The million dollar question is how. The method you've described is not directly programmable, and also lacks evidence that it would achieve what you say it will. This is not unique criticism I am leveling against your idea, we just had a MM update half a year ago and the method that was implemented had the same issue. If you're able to present your idea with programming logic statements, and provide a projection of expected benefits using our leaderboard data that shows a vast improvement, I would not hesitate to support your idea.

[Wildstreak]

So they put Solos and Groups in the same queue as games with bigger playerbases do? Well while you have been welcomed to that hell, you have yet to experience the worst.

[Monkey Lover]

Morosis's win ratio of 1.36, kd ratio1.68

"game is not playable" lol.

[RickySpanish]

ThreeStooges, on 19 December 2020 - 01:52 PM, said:

All you need to know about playing solo in qp is this: first team to 3 kills wins. always.

I'd say a three kill advantage is not insurmountable, but it's pretty bad. If it happens late game there's a possibility of recovery if the three extra dead 'Mechs pulled their weight and the enemy has some cored out players just hanging on by a thread. First to three though absolutely not. E.g. 3-2, but I think you meant first to pull ahead by three?

Edited by RickySpanish, 20 December 2020 - 08:51 PM.

[morosis]

Monkey Lover, on 20 December 2020 - 08:22 PM, said:

Morosis's win ratio of 1.36, kd ratio1.68

"game is not playable" lol.

the point of this thread is that the game experience is often appalling as a solo. with all the stat fluffing, grouping, imbalanced matchmaking aspects of the current game, individual PSR, WLR, and KDA metrics are close to meaningless. mine included.

anyone who puts stock in their tracked statistics with the current implementation of soup is deluding themselves. both positively and negatively.

[morosis]

Nightbird, on 20 December 2020 - 10:59 AM, said:

Everyone agrees on what the result should be, a better MM. The million dollar question is how. The method you've described is not directly programmable, and also lacks evidence that it would achieve what you say it will. This is not unique criticism I am leveling against your idea, we just had a MM update half a year ago and the method that was implemented had the same issue. If you're able to present your idea with programming logic statements, and provide a projection of expected benefits using our leaderboard data that shows a vast improvement, I would not hesitate to support your idea.

i dont think we know enough about how the current matchmaker logic functions in order to do that. I don't know what metrics the matchmaker uses to select the pilots it does from the available pool, and why it would choose one over another. if you have been exposed to that logic, i'd be happy to hear about it though.

what i can say is that I and many others have proposed a secondary lobby balancer as the path forward because it could operate using only the output of the current matchmaker as its input material, allowing that system to remain intact. it would just be a second step that should be computationally simple enough (because it only runs for the 24 pilots already selected) to operate within the existing dead time during lobby construction, and its only job would be to evaluate the current matchmaker's desired solution for balance (with the constraint of keeping premade groups intact) against the possibility of arriving at a better and more balanced solution if premade groups are treated as individuals and the 24 players redistributed instead.

the major assumption that I am making here, that may break the whole thing if not true, is that somewhere PGI tracks an individual's PSR as an integer, and that both teams could be evaluated as a sum of their individual PSR's. this would need to be true for both the case where premade groups are left intact (and evaluated with whatever the group PSR value is) and also in the case where premade groups are split and treated as individuals (using each pilot's individual PSR value instead).

i dont think it takes any elaborate proof to understand how this would lead to better solutions to lobby balance in most cases.

[Nightbird]

morosis, on 21 December 2020 - 08:23 AM, said:

i dont think it takes any elaborate proof to understand how this would lead to better solutions to lobby balance in most cases.

Well, try reading the community proposal for the current PSR+MM implemented.

https://mwomercs.com...66#entry6337266

They said without proof that their idea leads to better lobby balance as well. How are you sure your proposal is different from their proposal?

Compare to that, before the last PSR+MM update was implemented, I use a statistical model and accurately predicted the current state of Matchmaker performance, almost half a year in the future.

Original prediction:
https://mwomercs.com...cy-with-graphs/

The post comparing the prediction to today's MM:
https://mwomercs.com...ost__p__6361878


Making this prediction required understanding how the past and present MM worked, don't you think?

Edited by Nightbird, 21 December 2020 - 08:53 AM.

[MechWarrior254947]

Nightbird, on 21 December 2020 - 08:35 AM, said:

Well, try reading the community proposal for the current PSR+MM implemented.
https://mwomercs.com...66#entry6337266

MM?

I don't see any mention of matchmaker changes in the community proposal.
All I see is PSR calculation, but I can't find any references how the actual Matchmaker is going to use the new PSR along with mech tonnage and groups to form a lobby.

How groups PSR is calculated? Is PSR or mech tonnage take precedence when matchmaker is balancing players in the lobby? I see no such things discussed in the threads you provided.

[Monkey Lover]

morosis, on 21 December 2020 - 07:44 AM, said:

the point of this thread is that the game experience is often appalling as a solo. with all the stat fluffing, grouping, imbalanced matchmaking aspects of the current game, individual PSR, WLR, and KDA metrics are close to meaningless. mine included.

anyone who puts stock in their tracked statistics with the current implementation of soup is deluding themselves. both positively and negatively.

What I see is people crying about getting killed by premade groups but then they show win ratios larger than premade groups lol

Edited by Monkey Lover, 21 December 2020 - 02:34 PM.

[Nightbird]

OwO Kewensky, on 21 December 2020 - 01:40 PM, said:

MM?

I don't see any mention of matchmaker changes in the community proposal.
All I see is PSR calculation, but I can't find any references how the actual Matchmaker is going to use the new PSR along with mech tonnage and groups to form a lobby.

How groups PSR is calculated? Is PSR or mech tonnage take precedence when matchmaker is balancing players in the lobby? I see no such things discussed in the threads you provided.

The current Matchmaker creates teams that minimizes the PSR difference between the teams. Before, it also tried to limit the tonnage difference but this has been turned off. The only part that's unknown is exactly how group PSR is handled but it should be adding up the PSR of the group members otherwise my prediction would be off.

In other words, the reason the MM isn't working is because PSR (Player Skill Rating) doesn't represent player skill. As I explained in another thread, it is a function of skill*games_played. A mediocre skilled player that play 1000 games can have the same rating as a good player with 100 games. The MM is blind to that so it will slot mediocre and good players randomly and that gets us the current quality of games. (If an example helps, picture randomly slotting 12 black and 12 white chess pieces onto two teams. The odds of getting exactly 6 black and 6 white pieces on each team is almost 0 unless you can see their colors which the MM can't)

Edited by Nightbird, 21 December 2020 - 02:54 PM.

[BackShot]

morosis, on 21 December 2020 - 08:23 AM, said:

i dont think we know enough about how the current matchmaker logic functions in order to do that. I don't know what metrics the matchmaker uses to select the pilots it does from the available pool, and why it would choose one over another. if you have been exposed to that logic, i'd be happy to hear about it though.

i saw posts of dudes who jarl listed hundreds of games to try to understand how people were placed in one team or another, and the only logic answer you can see is : it is totally randomised.

What you (we all) need is not an improvement to existing matchmaker, what you (we all) need is A matchmaker, to begin with.

i have no faith at all in PGI willingness or even capacity to create one. They didnt do it in 8 years, what miracle would allow them to do it now ?

[BackShot]

it is not even only that nightbird, a mediocre player can even have a same or better rating than a good player in the same amount of game, as long as the mediocre player put bigger raw damage number by using spread big damage weapon, while the good player kill effectively using pinpoint low damage weapon.

aka lrms, atms, streaks, mrms lbx etc... vs gauss, ppcs, acs..

the only meaningfull difference you will see if you check their stats, is the win loss ratio, wich is :

- not checked by " matchmaker"
- flawed by the grouping.

Edited by BackShot, 21 December 2020 - 04:19 PM.

[MechWarrior254947]

Nightbird, on 21 December 2020 - 02:50 PM, said:

The current Matchmaker creates teams that minimizes the PSR difference between the teams.

Is it your assumption or it was documented somewhere?

Nightbird, on 21 December 2020 - 02:50 PM, said:

Before, it also tried to limit the tonnage difference but this has been turned off.

Is it your another assumption? Right now I see no significant tonnage disparity but there were some extreme cases before.

Do you know the exact logic how the current matchmaker create and balance 2 teams and how much weight PSR has in that formula? Was it changed in any way after PSR reset, comparing to the MM we had before?
Knowing PGI designing/programming skills, I wouldn't assume anything about the MM.


When we had old PSR, matches were more tolerable comparing to what we have today. New PSR may be not good but it's better than the old one, so the only other significant contributing factor is groups.
High skill groups or low skill groups make soup queue equally unfun for solo players.

[Egg Fu]

morosis, on 21 December 2020 - 07:44 AM, said:

the point of this thread is that the game experience is often appalling as a solo. with all the stat fluffing, grouping, imbalanced matchmaking aspects of the current game, individual PSR, WLR, and KDA metrics are close to meaningless. mine included.

anyone who puts stock in their tracked statistics with the current implementation of soup is deluding themselves. both positively and negatively.

You say "The game has reached unplayable status as a solo player" yet your stats over this last month's December leaderboard 178 games are sitting well above any month of your entire MWO playing career. You're doing better than you ever have!

Rich!!!