Scenario Matchmaking Rework Megathread

[Allonairre]

@JohnWayne I found your suggestion and didn’t really understand half of it.

I did wonder really how it would deal with population issues in a way that made low population SCs balanced and enjoyable. When SCs are popping non-stop a bad one is normally cancelled out by a good one next… normally… but if it’s quiet and pops are once an hour a bad SC is very deflating.

[ShadowWar]

Matchmaking is brutally hard. Looking to other places for inspiration may help. I think DotA implemented a system where when you queue for x games as a less common role, and you get priority on the more commonly played roles (or similar, haven't played in two+ years).

It's not an easy translation over here, because we are locked into a class based system, but make it an account level mark, and remove any SCs a player votes yes to surrender on to help avoid abusing it.

Just spit-balling.

[Akalukz]

There is a simple solution honestly.

Put 6 man in the queue both sides, fill that queue with 6 man, if not enough 6 man groups, go to next step

take any 5 man groups and fill with a solo to create a 6 man - put in the above queue or start a new one, prioritize to fill the 5 man (or whatever) to be 2/2/2 So if they are missing a DPS< give them a dps. If they are missing a healer give them a healer, if they have 5 dps, give them a healer

Repeat.

Develop a ranked like window showing groups queued (6 man x / 5 man x / 4 man x / etc, then break down by archtype. for totals. 15 tanks, 12 healers, 500 dps, Only launch full queueus with full 6 man groups.

Queues should launch every 15 mins, if you miss out you miss out. Catch the next one. Could even weight the classes/characters. If not enough to be a valid queue, post a message, not enough .. healers to launch scenario, next queue in 15 mins.

Could incentives further by doing something similar to the PVE weekly, where you get an account bound token for completing scenarios weekly on a toon, good for 75 crests on any toon.

[JohnnyWayne]

@JohnWayne I found your suggestion and didn’t really understand half of it.

I did wonder really how it would deal with population issues in a way that made low population SCs balanced and enjoyable. When SCs are popping non-stop a bad one is normally cancelled out by a good one next… normally… but if it’s quiet and pops are once an hour a bad SC is very deflating.

This thread is about the match maker, not increasing the population of the game or redesigning scenarios as a whole. A proper optimization algorithm will try and build the best matches it can with the ressources given to it. If you lack ressources, it will still try its best but you will have limitations. Matches will still be a lot better than now.
It is not a magic tool that spawns more players when pop is low. It will try and make sure the best possible matches are being created.

Additionally, you as a developer can tune the way it works by adjusting a formula that describes players and parties and how they are matched.

Also the part where it is "balanced out" mostly only occurs on order (I play both sides, I've been observing matches and match quality for over 3 years now). Destro has lower pop in queues, which creates kind of a panic response in the current match maker implementation, that just throws people into matches as fast as possible, so that premades get pops (on order, they build more parties in lft and have more guild parties in general).

The current implementation also does not leave good matched 6 man scs as they are and instead tries to fill them up, creating worse matches.

[stronky]

I am not going into a lengthy feedback and reponse..

Overall, I think there is a heavy YTYDBYD about premades in the current design and current (last week) sc changes. The vast majority of players are not interested in grouping for sc play. And those that do, do it to farm a far inferior opponent to get rewards fast. It's lower yield for both sides.

1. Premade vs pug is not fun for either side. At best, premade get 10 kills and pug get 0. There's no entertaining gameplay atm. Even as someone who do premade, i prefer the sc where no one has prems. The pugs are heavily objecting to being matches into premades by denying kills. It's what we see clear as day.

2. The changes you currently made are toxic to the issue and doesn't find the root cause. The last here at july forcing pugs to go out of spawn to be farmed even more. IF you care about twilight tide midtier you would not reduce drop chances, but fix queues to be pug vs pug, max 1 premade, or just remove premade prioritisation in queue. All of this would help new player experience more. Neither the sc nor twillight changes would be needed if T4 was more engaging that a typical midtier sc, i.e. had avg 30-60 kills.

You are not finding the root cause of why sc isn't currently working. Partly, because you believe sc should primarily be group play. If you made discordant for weekly only, you would see premade sc die altogether. While people who like premade should be catered to, there isn't a population to support premades. It is also a static variable, that cannot be changed. Whatever changes I can read you are making - with MMR, matching mastery builds and classes - you are removing randomness and increasing the static variable of the pug, but the pug will ALWAYS have an inferior team to the premade. If premade worked, ranked would have been a success. I don't think working on class matchmaking will work at all.

So I would either (1) make discordant weekly and see where it pushes people (2) make weekly sc also be doable on discordant (3) make weekly sc not having the win requirement or making more objectives where not all are needed (4) make personal mmr for premades, and queue an increasinly larger population of pugs against it. Point 1-3 will make it very clear whether my statement about YTYDBYD has any truth.

YTYDBYD : https://www.youtube.com/watch?v=0Wrw3c2NjeE

[Allonairre]

@JohnnyWayne

Thank you for the reply.

I am not asking for anything to fix low population but if a match maker cannot deal with low pop well it will not help.

I think the only solution when pop is low is cross realm with archetypes split evenly and trainers.

This is important to me because I play in Aussie (NZ) time zone and pop is not healthy .

Even when the pops are very fast it’s basically the same players every time which suggests to me that there is only 1 SCs worth of population Qing in n each side… so not really optimisable.

[JohnnyWayne]

@JohnnyWayne

Thank you for the reply.

I am not asking for anything to fix low population but if a match maker cannot deal with low pop well it will not help.

I think the only solution when pop is low is cross realm with archetypes split evenly and trainers.

This is important to me because I play in Aussie (NZ) time zone and pop is not healthy .

Even when the pops are very fast it’s basically the same players every time which suggests to me that there is only 1 SCs worth of population Qing in n each side… so not really optimisable.

You can't create a perfect solution for low pop. You can just try and implement a match maker that makes the most out of it. To be fair, the most issues the match maker has, is not during off times - aside of that, its a population issue, not a match making issue.

[Ninjagon]

Barrier system and "scenario kick system" are both very bad ideas.

Suggestions:
Cancel all barriers and add some new type of guards - strong ones but doing like 7k damage, not 30k.
Add some renown kill bonus for defenders while in the starting area.
And there will be fun again, even pugs vs premades.

[Emissary]

Barrier system and "scenario kick system" are both very bad ideas.

Suggestions:
Cancel all barriers and add some new type of guards - strong ones but doing like 7k damage, not 30k.
Add some renown kill bonus for defenders while in the starting area.
And there will be fun again, even pugs vs premades.

The guards not working was a bug, I have fixed that and it should be pushed next patch.

[Brizio]

I am not going into a lengthy feedback and reponse..

Overall, I think there is a heavy YTYDBYD about premades in the current design and current (last week) sc changes. The vast majority of players are not interested in grouping for sc play. And those that do, do it to farm a far inferior opponent to get rewards fast. It's lower yield for both sides.

1. Premade vs pug is not fun for either side. At best, premade get 10 kills and pug get 0. There's no entertaining gameplay atm. Even as someone who do premade, i prefer the sc where no one has prems. The pugs are heavily objecting to being matches into premades by denying kills. It's what we see clear as day.

2. The changes you currently made are toxic to the issue and doesn't find the root cause. The last here at july forcing pugs to go out of spawn to be farmed even more. IF you care about twilight tide midtier you would not reduce drop chances, but fix queues to be pug vs pug, max 1 premade, or just remove premade prioritisation in queue. All of this would help new player experience more. Neither the sc nor twillight changes would be needed if T4 was more engaging that a typical midtier sc, i.e. had avg 30-60 kills.

You are not finding the root cause of why sc isn't currently working. Partly, because you believe sc should primarily be group play. If you made discordant for weekly only, you would see premade sc die altogether. While people who like premade should be catered to, there isn't a population to support premades. It is also a static variable, that cannot be changed. Whatever changes I can read you are making - with MMR, matching mastery builds and classes - you are removing randomness and increasing the static variable of the pug, but the pug will ALWAYS have an inferior team to the premade. If premade worked, ranked would have been a success. I don't think working on class matchmaking will work at all.

So I would either (1) make discordant weekly and see where it pushes people (2) make weekly sc also be doable on discordant (3) make weekly sc not having the win requirement or making more objectives where not all are needed (4) make personal mmr for premades, and queue an increasinly larger population of pugs against it. Point 1-3 will make it very clear whether my statement about YTYDBYD has any truth.

YTYDBYD : https://www.youtube.com/watch?v=0Wrw3c2NjeE

Wow, i like this guy. Agreed.