849 replies to this topic

[FRAGTAST1C]

Anomalocaris, on 30 May 2020 - 12:27 PM, said:

+1 Past experience and Paul's limited comments so far make me skeptical about any significant improvement. I won't play merge queue until all you guinea pigs demonstrate that Paul can generate balanced teams with groups. I might be waiting a long time.....

I thought I could just shift over to FW and level up my mechs there 'cause it's much easier to do there than soup queue atm but even FW is messed up now 'cause of how some phases refuse to include Siege but instead have Scouting and only 1 game mode at a time. GG PGI.

[Nameless King]

Wont matter, go ahead

[MrMadguy]

Rkshz, on 30 May 2020 - 02:13 PM, said:

yes yes yes just reset tiers...
but I repeat - don't break current calculations system, stable "Average Score" are currently the only objective indicator of skill now. So just create dynamic tier, average score dependent. For example:
tier 5 = 0-99 AS
tier 4 = 100-199 AS
tier 3 = 200-299 AS
tier 2 = 300-399 AS
tier 1 = 400+ AS
- the tier should be updated every 20 games depending on stable "Average Score"
its objective and simple, and it will work with current calculations system, and you don’t need to break anything or redo it

Again. Many players, who think, that they're pros in this game, would be disappointed by truly working MM. Why? Simple math. If we would have perfect balance, every player would kill one 'Mech per game on average. It takes around 100-150DMG to kill one 'Mech (Lights take the same, it just doesn't register). So, on average every player would earn 200-250MS. See? This amount of MS is chosen as neutral level for REASON. And it's true for EVERY Tier. Not your 400+ for Tier 1. If you earn more, then you play BELOW your real MM rating. Nothing comes for free. Real reason, why some players earn 500MS in every match - they play with/against players, who can't earn even 100MS. They play at expense of others. And this fact isn't compensated for that players. That's why there is no reason to play this game, if you aren't pro.

Yeah. We don't have enough players. So, let's say, 1% top players need to be mixed with others in order for their queues not to last for too long. But... If we have just 1% top players and 70% "average" players, then why almost every match is terrible due to pros in it for that average players? May be that's because months and years of playing with inoperable MM have done their job? I.e. that 70% of average players are washed out of this game?

Edited by MrMadguy, 30 May 2020 - 08:55 PM.

[UnkerZ]

You don't have to reset the tiers to apply the fix. Just begin to apply the new formula and it will filter out T1s who do not deserve to be there over the next few months.

As long as you are getting there we would be happy. the problem is the most important flaw was ignored until now.

MrMadguy, on 30 May 2020 - 08:32 PM, said:

Again. Many players, who think, that they're pros in this game, would be disappointed by truly working MM. Why? Simple math. If we would have perfect balance, every player would kill one 'Mech per game on average. It takes around 100-150DMG to kill one 'Mech (Lights take the same, it just doesn't register). So, on average every player would earn 200-250MS. See? This amount of MS is chosen as neutral level for REASON. And it's true for EVERY Tier. Not your 400+ for Tier 1. If you earn more, then you play BELOW your real MM rating. Nothing comes for free. Real reason, why some players earn 500MS in every match - they play with/against players, who can't earn even 100MS. They play at expense of others. And this fact isn't compensated for that players. That's why there is no reason to play this game, if you aren't pro.

Yeah. We don't have enough players. So, let's say, 1% top players need to be mixed with others in order for their queues not to last for too long. But... If we have just 1% top players and 70% "average" players, then why almost every match is terrible due to pros in it for that average players? May be that's because months and years of playing with inoperable MM have done their job? I.e. that 70% of average players are washed out of this game?

Actually the game could just hide the Tiers and you have no idea while it does its job. Maybe we can leave it to Jarls to tell you your percentile if virtual penises are that important.

[MrMadguy]

UnkerZ, on 30 May 2020 - 08:57 PM, said:

Actually the game could just hide the Tiers and you have no idea while it does its job. Maybe we can leave it to Jarls to tell you your percentile if virtual penises are that important.

Tiers are for matchmaking, not ranking. Even if Tiers would be hidden, lose streaks would show, that they don't work anyway. And again. Jarl's isn't accurate for matchmaking. Because correlation between skill and average MS exist in all vs all situation only. I.e. when Tier 5 noob has to play against Tier 1 pro, so Tier 5 noob would be able to die instantly, while earning ~0MS, and Tier 1 pro to earn his 100500MS via killing all 12 enemy players.

Edited by MrMadguy, 30 May 2020 - 09:30 PM.

[crazytimes]

UnkerZ, on 30 May 2020 - 08:57 PM, said:

Actually the game could just hide the Tiers and you have no idea while it does its job. Maybe we can leave it to Jarls to tell you your percentile if virtual penises are that important.

I can't see any change to the tier system actually providing a visible difference to anyone. Ever.

The scrubs who are <50% and tier 1 are that unaware of what's happening in a match that they won't realise anything has changed. For everyone else, we will all get lumped into the same matches anyhow because there are that few people playing.

Realistically there is maybe enough to have two distinct groups... even though there is a massive skill difference, there just isn't the player base to have a more granular end result.

[RRAMIREZ]

A reset should happen; should have happened, and more important zero sum should be the rule.
post "placement phase", WLR should be central IMHO in a team game.

But I fear "lame" groups farming match score post reset.
I prepare myself for atm veagle, ghost assassins packs, and ultrafast rotato matchs during "give my tier 1 back!" period.

my 2 cts

But I still enjoy playing MWO, so...

Edited by RRAMIREZ, 31 May 2020 - 12:31 AM.

[Horseman]

UnkerZ, on 30 May 2020 - 08:57 PM, said:

You don't have to reset the tiers to apply the fix. Just begin to apply the new formula and it will filter out T1s who do not deserve to be there over the next few months.

Except that won't work. About 60% of the game's population is in T1 and due to that bloat they would make it substantially easier for poor players to still remain in T1.

crazytimes, on 30 May 2020 - 10:59 PM, said:

I can't see any change to the tier system actually providing a visible difference to anyone. Ever.

The scrubs who are <50% and tier 1 are that unaware of what's happening in a match that they won't realise anything has changed. For everyone else, we will all get lumped into the same matches anyhow because there are that few people playing.

Realistically there is maybe enough to have two distinct groups... even though there is a massive skill difference, there just isn't the player base to have a more granular end result.

The tiers are a product of your PSR, and PGI's matchmaker builds teams by PSR (and used to have a restriction to prevent T5s from dropping with T2+ players and T4s with T1 players).
ince the current PSR calculation is trash and allows even very bad players to grind their way to the "top" rating by sheer volume of matches, any attempt to use it for matchmaking fails by default - it won't be able to tell apart someone who like myself is usually in the 80s with a top one-percenter or someone who's below the 50% mark, because we've all reached the maximum PSR value a long time ago - the only difference is in how long it took.
The hope is that with the PSR reset and a less generous PSR formula, the matchmaker will be able to do its' job better than it is now. We're not expecting miracles, just something that's more viable than the shitshow we have to deal with now.

[Caldek]

Holy the ****, it is absolutely amazing to play MWO now. WOW. I can't tell you how much I enjoy this game now that you've eliminated solo que! I get to face off against coordinated units EVERY SINGLE DROP!!!! Can you even imagine what that does to the competitive landscape??? It's jacked to 12 even! I don't even care that 9 of 10 matches ends in a 12-3 stomp. On that 10th match I'm totally alive. I can juke and jive, rolling damage from ST to ST, to get that perfect kill shot to win an amazing close game. This is the best that MWO has ever been. I don't know exactly what you've done PGI, but keep it up. This is fantastic!!!!

I mean, we need more of this PGI, you guys are AMAZING!!!!

Best Wishes,
Caldek

[Stitchedup]

Reset I might come back just to grind back up to tier 1

[Sjorpha]

Anomalocaris, on 30 May 2020 - 12:27 PM, said:

p.s. check player numbers on steam. Already down quite a bit from the &quot;experimental&quot; phase...

So I did, and the numbers are showing an increase of about 17%
over the last 30 days. Current player count is the highest since July 2019. A small increase and still very low compared to previous years, but your claim above simply doesn't check out.

I agree the mm needs improvement and is a total mess, and that matches with the strongest groups are very imbalanced, but I also think the ability to drop casually with a friend or two brings back players and brings in new players. This is the first time you have really been able to introduce a friend to the game by playing together. I see the change as a net positive even with the imbalances it caused, and I think the increase in players reflect that.

[OneTeamPlayer]

A weird thought i just had but, why are we even being allowed input in the first place?

Crazy to ask, i know but pause and think for a second:

Have you ever heard of a company asking its players for input on design decisions involving their matchmaking system?

Like I get having player input, but I just realized that in decades of playing games with matchmaking systems I don't think any one of them has ever asked "Do you want us to change x, y, or z" about said matchmaking system.

Generally companies handle such decisions in the background and may or may not even inform the population about it, because matchmaking isn't a function of democracy it's a mathematical equation based on making fast and fair matches and tweaks to that system are based on whether one finds "fast" or "fair" more of a priority under given situations.

We're not game designers here, why is our input even remotely important to creating a functional system of this level of fundamental to a functioning product?

Like, if most players had said "no keep the broken matchmaker" would PGI have just said "oh okay then, we need to fix this critical system, but our players don't want it so i guess it will just stay broken as players slowly bleed out and stop using our product?

TL;DR I just realized this thread isn't a request and if we had said no to PSR reset PGI (if they're in they are in their right mind) would have simply kept the tier xp bar visible to players and then quietly disconnected tier from matchmaking while using an invisible new PSR system in the background to fix exactly what they wanted to fix.

This "discussion" is a bit of theater, if you really think about it.

Edited by OneTeamPlayer, 31 May 2020 - 03:16 AM.

[portus magnus]

Yes reset the psr and while you´re at it also fix the spawn locations. the stompy games dont matter to me as much as droping with an slow assault in alpha lance in some maps to be insta ganked by some rushing opfor

[Vesnuh]

I feel that with all info and data you have on each and every player. There should be a way to calculate a rank or tier based on data already collected without wiping it to zero.

[Poor-Life-Choices]

Before I agree with resetting the Tiers I have a couple of questions.
1) How big will the tiers be? Will 20% of players be in each tier? or how will you decide who goes into which tier?

Once we are sorted is the plan that for each T1 player on one side, there would be a T1 player on the other side?

Why can't you just sort us by AVG match score? Its the same criteria aside from win/loss.

You said its the fact that a player with a good match score can still move up in PSR with a loss is what breaks the PSR rating system. That makes no sense to me. Why is W/L more important than contribution. If you reset, won't we be in the same boat again. People that happen to be on winning teams will go up, people that are on losing teams will go down. If I suck and I group up with Ash and his friends I will end up in T1. But If I take Kras and stick him with some horrible players, he going to end up in T5. You are rating my by my friends, not by my ability to play well. How does this make sense?

[Nearly Dead]

Where I used to work it used to be somewhat common to discuss actions that were going to be unpopular and have focus groups involved and guide them to the desired conclusion by managing the facts, limiting the options, and steering the discussion. It made it easier to justify doing what they wanted or had to do.

I kind of see that here, where they obviously wanted to eliminate a feature (group queue) when it got too small to be viable but wanted to kind of talk up a possible improvement to the solo queue (that they ruined as a result of the change) to prevent a mass exodus.

By doing an "experiment" and making some changes they have kept people playing and getting used to the new normal. I know that without that, I would have left after the first week or so when the change was fresh.

[Marshal Jim Duncan]

For the record, I am a competent player … but not a great player.

So, the other though on PSR is this: I have been playing this game long enough to pretty much know within the 8 seconds of the match if it's going to be a win or a loss just by the way that players move off of the drop zone and the drop zone pattern on a particular map. Some patterns just inherently give one side the advantage by 'auto grouping them'. And if you have more than three potatoes in a match you can't get the win unless the other side happens to have more potatoes.

I don't know how to suggest you filter out the potato factor.


So if you have to reset the PSR to do it, then please do it with an event where I can get a bunch of cookies taking salvage from a bunch of low-skill players.

[Anomalocaris]

Sjorpha, on 31 May 2020 - 03:07 AM, said:

So I did, and the numbers are showing an increase of about 17%
over the last 30 days. Current player count is the highest since July 2019. A small increase and still very low compared to previous years, but your claim above simply doesn't check out.

Sure it does. Peak player numbers right after the queue merge were trending 100 higher than they are now a few weeks later. Absent the memorial day weekend spike (which was also a holiday weekend in the UK), the trend line since the first couple weeks of the merge was even worse. Saturday's peak numbers are only 38 players higher than the last event weekend peak pre-merge, and they are lower than any Saturday since the merge. Minimum player numbers are also dropping which is really critical if you're thinking about matchmaker issues.

Furthermore, the average player numbers are still following the same general trend they've been on for the last 3 months since all the Wuhan coronavirus lockdowns started. Absent the memorial day spike last weekend the average was also trending about 30-40 lower than now and has been dropping since then.

I've already made the prediction, but I'll repeat it. As lockdowns are removed, particularly in the US, you will see player numbers drop noticeably. I think that the trend will be quite long tailed as it will be till at least end of July when bonus unemployment payments run out. Barring some other major event to spike interest (such as really fixing the matchmaker and accounting for groups), avg player numbers by end of July will be back below April numbers. PGI has already shot their wad on increasing event frequency. I don't think they're going to be able to keep that going like they did during the merge queue "experimental" phase. And notice how the weekly events stopped as soon as the experiment did....

[Nelio]

Hi everyone!

Yes please reset the psr and make it equally gain/lose!

But I don’t like the idea about only the members of the winning team get up and the members of the losing team get down.

I don’t know if this idea has already been discussed (or discarded ;-) )
How about ranking all 24 players in one table after a game (and show it!). The first 8 with the highest match score get +1 and the last 8 with the lowest match score get a -1 to their personal score the rest stays the same.
So if you lose but perform very well you have a chance to advance or stay at the same level (by skill). If you are in the winning team but don't contribute anything to the win you will lose (for example afk).

Thanks Paul and looking forward to the change!

[DAEDALOS513]

Anomalocaris, on 31 May 2020 - 07:50 AM, said:

Sure it does. Peak player numbers right after the queue merge were trending 100 higher than they are now a few weeks later. Absent the memorial day weekend spike (which was also a holiday weekend in the UK), the trend line since the first couple weeks of the merge was even worse. Saturday's peak numbers are only 38 players higher than the last event weekend peak pre-merge, and they are lower than any Saturday since the merge. Minimum player numbers are also dropping which is really critical if you're thinking about matchmaker issues.

Furthermore, the average player numbers are still following the same general trend they've been on for the last 3 months since all the Wuhan coronavirus lockdowns started. Absent the memorial day spike last weekend the average was also trending about 30-40 lower than now and has been dropping since then.

I've already made the prediction, but I'll repeat it. As lockdowns are removed, particularly in the US, you will see player numbers drop noticeably. I think that the trend will be quite long tailed as it will be till at least end of July when bonus unemployment payments run out. Barring some other major event to spike interest (such as really fixing the matchmaker and accounting for groups), avg player numbers by end of July will be back below April numbers. PGI has already shot their wad on increasing event frequency. I don't think they're going to be able to keep that going like they did during the merge queue "experimental" phase. And notice how the weekly events stopped as soon as the experiment did....

You can speculate all you want.. doesn't mean much. Steam charts have been trending up the last three months (that's what the green numbers below the chart mean).
https://steamcharts.com/app/342200#3m

I already shot your event theory down when I showed that event frequency has not increased and is following a steady pattern all the way back to January averaging 2 or 3 events per 30 day period.

You're just grasping at straws because by now you must be realizing this change has been a net positive for the game..

Edited by DAEDALOS513, 31 May 2020 - 11:53 AM.