228 replies to this topic

[Prototelis]

No.

[dario03]

Why not both?
8v8 and 12v12. If there is a decent population on then mostly 12v12, when not mostly 8v8.

[UnofficialOperator]

dario03, on 03 January 2020 - 02:34 AM, said:

Why not both?
8v8 and 12v12. If there is a decent population on then mostly 12v12, when not mostly 8v8.

8v8 would be faster to fill and more fun imo. Would weight it towards Assaults though.

Haven't played in 6 months. Just managed to claw my way back to a 1.0 WLR after a string of losses for this month with a 375 MS & 1.75 KD for Assaults. From what I can tell, the usual bads are still bad without improvement. The usual carries are still carrying. The demise of Assaults is overstated, only once was I nascared through my own fault.

If they ever come up with 8v8, I'll probably drop by again.

Until then, adios!

Edited by UnofficialOperator, 03 January 2020 - 05:12 AM.

[MischiefSC]

Prototelis, on 03 January 2020 - 02:03 AM, said:

No.

I dunno. 8v8 would cut MM load by 25%, which is big. That could potentially be a big improvement as the population gets smaller.

Currently it appears that the MM release valve when it has no viable well balanced team options is to just say "**** it, split on tonnage, go nuts you're on your own". As population declines this does stuff like put 3 EMP on one side. While that's a huge statistical anomaly it's still bad.

However it doesn't matter how big or small the MM change you want to make is, you still have to justify the cost of cracking open the MM, making and testing changes. The difference between a big change and a small change may literally be 30 minutes more or less coding time. The real cost is testing and taking those people off current projects to come do it, which breaks whatever development chain and associated projects other people are doing related to it. Any MM change wouldn't be cheap so saying "tweak this thing that impacts less than 2/3rd of 1% of the population" isn't gonna happen.

However if you can present a change that would have significant impact for thousands of players then you can probably get your tweak for the top X players done at the same time.

However PGI only makes money from mechpack sales so you'd also need to make a compelling case that the change would drive enough sales to more than make up for the cost. Saying "we'll play longer/more often" isn't hugely compelling. I don't know how to pitch that change and hang a clear value on it to make it worth it.

PGI has said MWO is maintenance mode so any hours spent are stealing from other projects. Not sure any of this is realistic.

[Ssamout]

I doubt PGI has any interest in fixing its matchmaker. This is a known topic even to them, as it has been discussed for years, since before the ELO system was creating long search times for the top end. If they had any inclination to do so, it would have been already done. If I remember correctly the wording in the PSR patch even said so: that purpose of the MM is only to separate new accounts from veterans.

There has even been attempts by seemingly competent players to help PGI in that area like this 4 years old post in reddit.

[justcallme A S H]

MischiefSC, on 02 January 2020 - 11:31 PM, said:

How many members of EMP are there? What percentage of the player population are they?

That's the math you keep missing. The top 100 players represent 0.67% of the player population. As in 2/3rd of 1% of the population. And that's the TOP 100 players. As in approximately the 12 best teams in comp play, all together.

Any yet if there is 2-3 or 4 of them online... We are all in each others matches most games and the MM never distributes evenly, because it DOES NOT TRY.

We aren't even trying to sync drop - just a result of such low population.

So while there migh only be 0.67% or whatever number you wanna come up with... If it happens for 75% of all your matches for 4hrs...


PSR is and always will be garbage because it is meaningless. There is no defending it.

[Prototelis]

TLDR; It only takes 24 people to make one balanced match.

[Y E O N N E]

MischiefSC, on 03 January 2020 - 10:42 AM, said:

I dunno. 8v8 would cut MM load by 25%, which is big. That could potentially be a big improvement as the population gets smaller.

Currently it appears that the MM release valve when it has no viable well balanced team options is to just say "**** it, split on tonnage, go nuts you're on your own". As population declines this does stuff like put 3 EMP on one side. While that's a huge statistical anomaly it's still bad.

This is irrelevant because the match maker doesn't actually know who is good and who isn't. It's only looking at tier, a data point that is well-established as not being representative of who is good and who is not. This renders the size of the matches and the activation of release valves immaterial to the results we see.

[Paul Meyers DEST]

Prototelis, on 03 January 2020 - 05:27 PM, said:

TLDR; It only takes 24 people to make one balanced match.

No! or Yes!

I am stupid. For sure you are right.

Edited by Paul Meyers DEST, 04 January 2020 - 11:46 AM.

[MischiefSC]

justcallme A S H, on 03 January 2020 - 05:14 PM, said:

Any yet if there is 2-3 or 4 of them online... We are all in each others matches most games and the MM never distributes evenly, because it DOES NOT TRY.

We aren't even trying to sync drop - just a result of such low population.

So while there migh only be 0.67% or whatever number you wanna come up with... If it happens for 75% of all your matches for 4hrs...


PSR is and always will be garbage because it is meaningless. There is no defending it.

Not defending anything, PSR is garbage. I've said so for a long time. The matchmaker seems to have been designed with the release valve set to 'Well, these guys seem like statistical outliers, **** all and lets just throw things together vaguely by tonnage' means that the people who ARE statistical outliers are pretty much perpetually shafted. That also sucks.

The point though is that it doesn't matter how big or small the change is the costs to a MM change are about the same so if you want this issue fixed it needs to have a big enough payoff to make it worthwhile and that would need to be one that impacts a lot of people in a material way.

[MischiefSC]

Y E O N N E, on 03 January 2020 - 06:53 PM, said:

This is irrelevant because the match maker doesn't actually know who is good and who isn't. It's only looking at tier, a data point that is well-established as not being representative of who is good and who is not. This renders the size of the matches and the activation of release valves immaterial to the results we see.

Hence why it's tied to a post about what you COULD do to make a viable matchmaker with a declining population that had a material impact on the great majority of players in majority of matches.

Some sort of Elo style ranking both for players and mechs. So when Y E O N N E drops in a stock Cicada he's not valued the same as a meta MK II.

In that context 8 v 8 might be worthwhile, especially as population continues to shrink. Dear god give people a payout bonus of some sort for playing lower tier/weight mechs. Need to incentivize medium and light mechs to try and draw tonnage towards middle, need to account for value of the chassis given that for the vast majority of players the caliber of the mech they're taking is as relevant (or even more so) than their individual skill.

Better matchmaking, reason to play mechs you normally wouldn't other than just giggles.

[Prototelis]

No.

Balanced matchmaking only becomes more granular and easier with more players.

[VonBruinwald]

Prototelis, on 04 January 2020 - 02:23 PM, said:

No.

Balanced matchmaking only becomes more granular and easier with more players.

And with a functioning matchmaker.

I suppose throwing everyone in a single bucket of random could also be considered balanced though.

Not that we have either of these.

[JediPanther]

Your team has a 100% chance of loosing when I join it in an assault. Nothing can change that and none of you are worse at assassulting than me. Game balance achieved.

[MischiefSC]

Prototelis, on 04 January 2020 - 02:23 PM, said:

No.

Balanced matchmaking only becomes more granular and easier with more players.

Only with infinite time of perfect suits of candidates.

MWO has to fit both tonnage and player skill. The larger each team pool is the more difficult it is to puzzle pack tonnage and try to keep skill balanced. Also the wider the skill gap has to be in any given team because higher skilled players are less common.

For example you've got 30 people available in a 60 second window. If I can take just 16 of them I can be more selective and group them in a narrower skill bracket. Then I can sub one of the other 14 in as needed to fit tonnage and stick as close to skill bracket as possible. Plus after I drop that match I only need 2 more people to potentially fill another drop - those 14 people won't wait as long as match turnover is higher because I am slicing population into smaller groups.

If I have to fill 24 I only have 6 variables out of the 30 and I've got to make more sacrifices in skill balance to meet tonnage balance or vice versa.

The bigger your match population is relative to the size of your matchmaking pool the fewer choices you have. No question at all MWO would still benefit in 12 v 12 by having a better evaluation of each mech than just tonnage and have it balanced vs player skill so you can have a wider tonnage gap between team but better skill gap and have that be more balanced but in a low population smaller teams = better MM balance options.

For example. I give you 30 flowers and say you've got to make a beautiful bouquet of 16 flowers you can pick the 16 that look best together. If I say you've got to use 24 the odds are higher that some won't look as good as the rest because you're required to pick from the 30 flowers.

Edited by MischiefSC, 04 January 2020 - 11:50 PM.

[killzone1]

Khobai, on 02 January 2020 - 11:50 PM, said:

Theres nothing obstensibly wrong with the tier system other than the fact tier 1 has become way too saturated and they need more tiers to separate out players. So they need to add more tiers or reset the tiers or a combination of both.

I think there are several big issues with the current tier system and matchmaking.

1 - The tier system isn't taking skill into account. It assumes all kills and wins are comparable.
2 - The current matchmaking is placing tier 5 cadets in with T4-T1.
3 - They probably need to add something like x wins and x kills against an exclusively T2-T1 team in order to advance to T2/T1 etc. But again Tier is probably moot anyway since I seem to see the same people time and time again, the population size has to be incredibly small. I just don't think any tiering system will work with that small of a population, even with really long game delays.

[Capt Deadpool]

Could probably have three somewhat balanced tiers if groups were just determined by 'How often a player uses their microphone to communicate with their team':

Tier 1: frequent microphone button users
Tier 2: infrequent microphone button users
Tier 3 (chaos tier): complete silence

(Also, Tier 3 have their minimaps blacked out until players purchase a microphone and use it so they can move into Tier 2)

[Horseman]

killzone1, on 09 January 2020 - 12:01 PM, said:

1 - The tier system isn't taking skill into account. It assumes all kills and wins are comparable.

It's largely assuming all wins are comparable.

Quote

2 - The current matchmaking is placing tier 5 cadets in with T4-T1.

That was tested and debunked recently.

Quote

3 - They probably need to add something like x wins and x kills against an exclusively T2-T1 team in order to advance to T2/T1 etc. But again Tier is probably moot anyway since I seem to see the same people time and time again, the population size has to be incredibly small.

They need to adjust how the PSR gains and losses are issued and then do a global PSR reset. Anything else would at this point be garbage in-garbage out.

Quote

just don't think any tiering system will work with that small of a population, even with really long game delays.

All the system has to do is make sure the teams are roughly on par with one another in terms of performance. Not rocket science.

[VonBruinwald]

Capt Deadpool, on 09 January 2020 - 03:15 PM, said:

Could probably have three somewhat balanced tiers if groups were just determined by 'How often a player uses their microphone to communicate with their team':

Tier 1: frequent microphone button users
Tier 2: infrequent microphone button users
Tier 3 (chaos tier): complete silence

Whoa whoa whoa! Don't start discriminating against the vocally impaired:

Tier 1: Frequent uses mic / takes company/lance command / uses command wheel
Tier 2: Infrequently uses mic / takes company/lance command / uses command wheel
Tier 3 (NASCAR Tier) : Uses none of the above - (HPG - Domination only)
Tier 4: Potatoes!
Tier 5: MASHED POTATOES!

[Capt Deadpool]

Haha fair enough!

I am happy with any and all communication--I will take a team with good communication (which I hypothesize correlaties with higher levels of situation awareness), than guys with big damage numbers who spray damage all over the place but don't coordinate.