435 replies to this topic

[Trevelyas]

MechregSurn, on 29 September 2015 - 04:21 PM, said:

I concur with the above statement that tonnage is not the best measure in this game. We need a battle value for each mech based on player performaNce.

I linked this earlier but I'll quote it directly for convenience:

Quote

In its final form, a player's PSR (or Elo or whatever rating system is ultimately used) should actually be a tree structure. Your global "base" PSR would be the root of the tree, and each branch would store a "difference PSR" in the following hierarchy:
Global -> Weight Class -> Chassis -> Variant

Your PSR for MatchMaker would be calculated as the Global + the PSR difference nodes leading to your currently selected variant.

After every match your PSR adjustment gets applied 100% to your current variant, some smaller fraction (say, 75%) applied to the Chassis, and so on up to your global PSR.
When you buy a new Mech its PSR value gets seeded from the closest available node on the PSR tree.

The elegance of this hierarchy is that it automatically accounts for each player's performance in all their Mechs affected by each variant's individual strengths and weaknesses. Also, having Mech performance influence reflected directly within the PSR tree means that the MatchMaker can relax some of the hard Mech-focused restrictions it currently has to deal with (Tonnage, 3/3/3/3, etc.) when trying to create balanced matchups.

There should also be separate PSR trees for solo and group queue.

Of course, all of this hinges on PSR ACTUALLY being a measure of player skill.


Edited for formatting.

Edited by Trevelyas, 29 September 2015 - 04:47 PM.

[LordLosh]

could think of a hundred + things i would rather you spend time and resources on then this right now

[JHackworth]

why remove 3/3/3/3. I like this in theory but the thought of having to face 12xACHs because EMP decides they just want to goof around for shits and giggles gives me heartburn

[Kilo 40]

Deathlike, on 29 September 2015 - 11:10 AM, said:

So... CW Phase 3 in at least Nov.

I see.

Yup. Just in time for me to invest all my time into Fallout 4.

But he's right about one thing. 2015 has been a great year for players. So long as you are a comp or large group player that is. us solo players got the finger in 2015.

[bad arcade kitty]

pls no opt in for group queue
solo queue shouldn't bleed their players into the group queue to pump their wait times on the cost of ours

[Hann Solo]

I think Kinetix is pretty spot on.

[lashropa]

I like the direction you're taking this. Can't wait.

[Threat Doc]

I'm going to have to wait and see on the tonnage values for groups. It seems like a reasonable measure to take, and I understand the reason it's being done, but some of the numbers just don't seem right.

[Kiiyor]

I think the tonnage restrictions are nice, except for those imposed on 2 players.

If you're instructing someone new, there's a chance you'll be running in the same mech as them (I usually do, anyway). The restrictions outlined above mean that all lights and assaults are out. I think this is a mistake!

2 players: Minimum 40 tonnes, max 200!

Two players won't likely make that much of a difference to matches, even if they are cheesin' it up in Cheetahs or Dires.

That's my 2c anyway.

[Nothing Whatsoever]

Only thing I wonder about is if the rate for the larger groups is too generous? 795 is basically 3/3/3/3 for a group of twelve.

So the question is where do we want the trade-offs to be? Between Tonnage to Group Size?

I think allowing smaller groups, say 4 mans and less to have smaller Minimums, while also reducing the Maxes on larger groups could be a way to go.

So for a 12 man what if the max tonnage is 720 (which is 60 tons per player with this example) or 760?



Whatever value works best should be applied, but such a setup could allow for a bit more flexibility for players that favor certain mechs over others while still making players to make some trade-offs to drop together.

And if solo players could be allowed to opt-in to the group queue, maybe allow a max number of solos per side, say one to three solos max per side, when they select that option. So it would provide players dropping solo that opt-in more opportunities to drop into a match with the default Solo Queue still being available for that solo player, and if there is a group that needs a single player for an odd numbered group could fill that need with up to a max of three solos per side possibly.

I'd probably drop-in to the Group queue as a solo if I could.

[Saskia]

I am against the restrictions. The 3/3/3/3 is hard enough to manage in a 12 group. Taking away our freedom of choice from a 2 group is even worse.
I don't know who it is that's whining about tonnage in matches but after witnessing the appalling kiss ass maneuver of you dropping the required win score from 400 to 300 in a recent event to succumb to these whiney players, I'm not surprised you are considering this.
You have slowly been destroying what I deemed to be fun and challenging in MWO. Stop imposing all these unnecessary changes upon us!
The UI is now aweful (the only thing wrong with the first one was the friend list jumping), LRMs are an overused abomination, ECM is ineffective, terrain is ridiculously difficult to walk over in assaults (such a large mech should NOT have such difficulty walking over a damn rock or crevice), an Arctic Cheetah seems to have more armour than an Atlas for example, hit registration with SRMs is aweful, the servers are terrible (started out with 230 ping now they're all 285+)... these are some things that could use your attention, not tonnage in matches.
I lead groups all the time. It's not always about creating the best kill team to death roll in. It's mostly about having fun and selecting your favourite mech but we haven't been able to do that since 3/3/3/3 was introduced. Then people drop out of groups because they can't get their preferred tonnage.

[50 50]

Given that the heavy queue is the most popular I can understand why the tonnage limits favour the upper end of the scale.
I also think it is a great move to have made these limits easy to adjust for mix and balance.
I am hoping the group and solo queues get combined so that there is a more even match up now that PSR and average group PSR is being used. It should make finding a match much quicker as well.
All up, I think this is a good direction to move in and allows teams to be a little more creative in their mech selection which will make some interesting match ups.

At present, with the 3/3/3/3 split we have the following scaled tonnage limit:




With the currently proposed limits for the 22nd October patch, you can see that the variance in tonnage is closer and a bit more stepped.


It will require a bit of tinkering as we go along, but if the goal is also to get a good mix then we probably don't want to see teams comprised of more entirely 20 ton lights or 100 ton assaults.
An alternative may be to review the limits based on a lance with a 1/1/1/1 limit as a guideline.

Edited by 50 50, 29 September 2015 - 07:41 PM.

[Kageru Ikazuchi]

50 50 ... now that I see those two sheets/graphs, the numbers initially tossed out by Russ now have some rhyme and reason to them. (Well, rhyme, anyway ... not sure about the reason.)

Now for my feedback ...

First ... while I would like to allow small groups to run pretty much whatever they want, part of the power creep / time-to-kill problem right now is small groups running near-maximum DWF and TBR. While I understand the desire for one-on-one paired training, we need to consider the balance impacts as well. This means compromise. Allowing most 0/2/0/0 and 0/0/2/0 combinations should be feasible, though.

Second ... smaller groups should have the option to have a slight tonnage-per-mech advantage over the largest groups.

Third ... we should not be able to bring 3x35 / 3x55 / 3x75 / 3x100 ... maybe 3x35 / 3x55 / 3x60 / 3x80 (for example).

What I came up with using these guidelines is:
2: 80-140, 3: 125-195, 4: 170-250, 5: 215-305, 6: 260-360, 7: 305-415, 8: 350-470, 9: 395-525, 10: 440-580, 12: 530-690

Finally ... it's going to be impossible to have a purely tonnage-based system that flat-out prevents something like a group of 12xSCR or 6xTBR 6xACH without being overly restrictive.

Edited by Kageru Ikazuchi, 29 September 2015 - 08:15 PM.

[LCCX]

--- Preface:
1) Thank you PGI/Russ for trying to improve the Group Queue match fairness/quality. I'm glad you understand that it is important. "Play with your friends!" inevitably resulting in "Get curb stomped by large groups of competitive players! [or get pitted against some poor sods who'd rather not see you on the battlefield either]" sucks all the fun out of playing the game.

2) I appreciate you trying to keep the "any group size" feature, but am pretty sure that the group size limit is going to have to drop to 6, or maybe even 4.

3) I think that player skill has a far larger impact on gameplay and match result than tonnage, at least at the higher end of skill (top third of the player base by rank?). And player skill (at any game) is distributed exponentially. A 12-man group of Tier-5 players in Dire Wolves (or, let's be honest, Atlases) probably isn't scary to anyone in Tier-3 or higher; a 4-man group of top-1% players, by rank in the player base, would probably be scary even in Locusts
---

That said, actually answering your questions about tonnage:

Russ Bullock, on 29 September 2015 - 09:57 AM, said:

- If 6 groups of 2 bumped into a single group of 12, what should the tonnage advantage be?

100% more, keeping in mind that a lot of people won't use that "extra" tonnage because they want to run their [insert not a Dire Wolf here]. 3 out of 4 mechs in a mech package are not assault mechs and people will want to play and master those mechs too. I also would not feel the least bit bad about pitting 12x Dire Wolf in pairs against a 3/3/3/3 12-man premade.

Russ Bullock, on 29 September 2015 - 09:57 AM, said:

- How restrictive should we become on the minimum tonnage while still allowing flexibility for a small group that may want to run light - keeping in mind that we're hoping to give extra tonnage to the side with more groups?

I'd err on the side of giving maximum flexibility to the smallest groups, and the least to the largest. I'd guess that group size is also of exponential performance, so I think that flexibility should be pretty low even at 6-player-size groups and a 12-man might be forced to have an exact tonnage (unless additional group size handicaps/restrictions, such as 3/3/3/3, are in place).

Russ Bullock, on 29 September 2015 - 09:57 AM, said:

- Please provide your feedback on what the min/max tonnage should be for each group size.

• A pair of players should be able to bring anything they want like they can now within the 3/3/3/3 system.
  
• A trio of players should not all be able to bring Locust, Commando, or Dire Wolf.
  
• 4 or more players should have to bring a variety of mechs; 3x lights must be accompanied by at least a heavy and 3x 100-tonners cannot be accompanied by more than a light.
  
• A full or nearly-full team should have to bring a greater variety of mechs.
  
• Groups of 8-12 should not be able to bring exclusively pairs of 35+55 ton mechs, 35+75 ton mechs, or 55+75 ton mechs (an average of 45 tons per player is too low and 55 and 65 tons per player are too high, unless additional restrictions exist).
  
• Smaller groups should have a tonnage advantage; a 2-man (+100%) should have twice the advantage of a 4-man (+50%), which should have twice the advantage of an 8-man (+25%) over a 12-man (+0%) group.

.

SpiralFace, on 29 September 2015 - 11:16 AM, said:

I think 3/3/3/3 SHOULD still exist, but put it to the client side IN ADDITION to the weight restrictions for getting into group queue.


...
PLEASE consider that large group players will fully cut out assaults in order to spam more Timber wolves if you give them tonnage restrictions WITHOUT any kind of class restrictions.

Exactly. Consider that a 12-man might bring 6x Arctic Cheetah (6*35t=210t), 6x Hellbringer (6*65t=390t) for a 600 ton team drop weight.

Russ Bullock, on 29 September 2015 - 09:57 AM, said:

The point here is to allow the matchmaker to run with some simpler metrics; therefore the MM will not be adding up tonnage or looking at the composition of the groups. It is the groups responsibility to utilize the tonnage as they see fit. This is important because if the MM finds the perfect set of players to play against each other we do not want to have to reject that match for any reason.

<picking nits>If there is a 99% probability that one team or the other will win a match, then they are not "the perfect set of players". Putting blinders on the MM so that *it* cannot calculate the match result probability doesn't mean the players don't know it once they drop, and players being able to accurately guess the result of a match while the drop timer is still counting down is the problem. Everything is fine even if the MM knows match outcomes with 99% accuracy so long as players are left guessing until the end of the show.</picking nits>
Is matchmaker simplicity a technical requirement of MWO right now? If so, then IMHO that strengthens the argument that handicaps be put into place before groups enter the MM queue, and presumably those checks can be allowed to be relatively complex.
- What if a higher group PSR incurred tonnage penalties (e.g. less tonnage flexibility)?
- What if groups were not allowed to have duplicate mechs (more than 2 of the same chassis)?

[LCCX]

Laset Kazenrete, on 29 September 2015 - 12:18 PM, said:

And then there's an interesting contradiction to prior claims: anyone remember when PGI was saying groups larger than 4-man were a small percentage of the group queue? Now they're claiming that most games have "one extra group" - which would mean the majority of games have groups with at least six members.

So what's with that, PGI?

I think you'll find you did the math in your head incorrectly. "one extra group" could be 3x2+2x3 vs. 6x2 or 4x3 groups:size matches. There are still mostly smaller groups in the group queue, which is part of why they get matched against large (8+) groups - because there isn't another 8-man or 12-man of comparable skill to pit the large group against.

[VXJaeger]

12 600 795
5s later than this comes to liveservers, some team (*cough* 228th & Th3B33f) will go for 12 Yen-Lo-Wangs
Or if they don't , we will

[Fiona Marshe]

Lower tonnage limit at the bottom end of group size should be based on 3/3/3/3.

So:
2 Players = 40 (2 lights)
3 Players = 60 (3 lights)
4 Players = 100 (3 lights, 1 medium)
etc

A small wolf-pack group is not at a disadvantage when they work together regularly.

[Kageru Ikazuchi]

VXJaeger ... and after the Wang party gets old, the true Streak-nado ... 12x Streak Dogs.

If we don't do it, someone will ...

[Jonathan Paine]

Dear PGI. As you move from the clearly not so well thought out table posted on the front page, and towards the gentle suggestions of your beloved community (be nice to smaller groups, limit bigger groups), you might want to do yourselves a favor and list what the average heaviest mechs would be, as well as what the average lightest mech would be. You don't want to find yourself in a silly situation where (for example) a nine man has to run slightly less tonnage on average than a ten man. *hint* *hint* *hint*

Also, maybe send a message or two to groups like the 228 and politely ask them "Dear competitive players, how would you break our idea? Please let us know now, pretty please? With sugar on?"

[Leopardo]

yeah group of two - min 40 !!!!!! max okay 200!