My solution to DB's "Shuffle" issue.


(Amerika) #21

This is a very old and tired subject. Humans don’t learn on a curve that never changes. It flattens out after a certain amount of experience and skill gains become much slower over time. Experience matters but your gains from it are incredibly fast early on and then move towards exceptionally slow and situational later on. At some point fairly early in most games a players personal skill factors way more than total played. So if you sort by level, outside of servers where it’s level 10-15 and below, then you will end up with an even worse situation. Not only that, you punish players who have put in significant time but simply aren’t that good as players. A lot of people play to have fun and to pass time…not to constantly improve. So anybody who gets up there in level would have to start a new steam/db account so that they weren’t constantly making their team lose.

The server currently puts players on both teams using a hidden MMR system that calculates your last 25 games played. So each team, technically at the start, is “balanced” because these numbers are even regardless of their account level. However, this system is flawed since it can put too much reliance on one or two really good players to carry quite a few not as good players in a lot of situations. Also, it does not factor in disconnects. It also doesn’t factor in the mercs people choose. It also doesn’t factor in whether you’ve been drinking for 12 hours straight and are blaring AC/DC. In short, a balance system is only as good as the people who connect to the server want it to be. If they understand and care about team balance, playing the objective and are genuinely interested in winning then you will have consistently more balanced games. But even those can end in stomps. If you don’t believe me go play/watch tournament matches or play in PUGS with fairly even skill on both sides…sometimes matches end as a stomp.

Casual MM will fix a lot of this since it will use everyone playing at that time to try and balance teams. But that doesn’t fix a lot of what I talked about above. Balance is a good thing to strive for but mentally understanding why it’s not always possible or why it’s so hard to achieve goes a long ways towards setting your expectations correctly.


(bontsa) #22

Can we just sticky that as a “Look at this before making a yet another goddamn balance thread” and move along without getting these things popping out like shrooms during rain.


(LifeupOmega) #23

Personal experience I’m assuming?


(PeterComputer) #24

Can we just sticky that as a “Look at this before making a yet another goddamn balance thread” and move along without getting these things popping out like shrooms during rain.[/quote]

You know what, you’re right. I should’ve thought more before posting. I was writing under the assumption that Level=Skill, which I still think it’s true to some point but it can’t be considered the norm.

I’m sorry that I created such a mess of a thread and I’ll edit the original post.

Again, sorry. If you do want to continue talking in here though, do you have any suggestions to improve the “Shuffle” system, please post them.


(Amerika) #25

Personal experience I’m assuming?[/quote]

I change the music a little more often :slight_smile:

Can we just sticky that as a “Look at this before making a yet another goddamn balance thread” and move along without getting these things popping out like shrooms during rain.[/quote]

You know what, you’re right. I should’ve thought more before posting. I was writing under the assumption that Level=Skill, which I still think it’s true to some point but it can’t be considered the norm.

I’m sorry that I created such a mess of a thread and I’ll edit the original post.

Again, sorry. If you do want to continue talking in here though, do you have any suggestions to improve the “Shuffle” system, please post them.[/quote]

No worries…it’s fine :slight_smile: It’s just a subject that is consistently brought up and routinely shot down due to the the flawed nature of the argument. That’s why I consistently take the time to articulate the facts and try to provide some perspective so that people will/might understand and possibly become less angry and more understanding of where we are at now and where DB hopes to go. We all want good matches among people of even skill brackets. That’s the end game for any system. But you have to have proper expectations. And you have to know that even teams of pros where both sides of extremely skilled talent and are, on paper, an even match you can still end up with a stomp. Look at the results through the years among pro sport teams.


(AlphaUT) #26

As many examples I see, current shuffle system does not work as what it should be since many time, it simply switch whole team to the other side, or switch the guy who is carrying the team to the other side which will just let that match switch winning and losing.

It is true that there might have some success examples, but really rare.


(PeterComputer) #27

Yeah but… how could it be fixed?


(PeterComputer) #28

Would it be possible to base a shuffle system on player scores?

Well… if the match is just beginning it wouldn’t do much harm/good and we should consider that someone may just be having a lucky match. Plus, if you’re playing as a Med, you would probably get switched no matter what.

So, conclusion: Nope. Damn, I really thought I was onto something.


(Amerika) #29

[quote=“AlphaUT;214939”]As many examples I see, current shuffle system does not work as what it should be since many time, it simply switch whole team to the other side, or switch the guy who is carrying the team to the other side which will just let that match switch winning and losing.

It is true that there might have some success examples, but really rare.[/quote]

The reason for that is both teams, according to each players MMR total per team, are “equal” in regards to their skill rating. That’s why you see full team swaps with maybe a player or two swapping around. So you end up with a very rigid system that uses total MMR value to balance teams but doesn’t factor anything in like merc choices, likelihood of switching mercs, experience with each merc etc.

If the current MMR system assigned an MMR value to each merc a person played and only factored in the mercs in your squad while also making sure to include an even number of healers/engineers etc. on both teams, then it would be slightly better than what we have now. It could still end up unbalanced if 3 people on your team refuse to use the medics in their squad and the other team has a proper composition. There are multiple other fail states too. Casual MM also won’t fix this issue…people play what they want to play and how they want to play many times. And sometimes you will end up with team players and sometimes you end up with 6 drunk Vassili’s. And sometimes you end up with a person who would totally switch to that needed engineer but they don’t have one in their squad :frowning:


(Jostabeere) #30

Shuffling based on a hidden Elo which counts last 10 matches fails miserably.

(As far as I know, the shuffle does take this hidden last 10 matches Elo rating system into account, doesn’t it @Amerika ?)

So maybe we go super crazy and try balancing by levels? I mean, it can’t be worse than it is now?


(PeterComputer) #31

[quote=“Jostabeere;214967”]Shuffling based on a hidden Elo which counts last 10 matches fails miserably.

(As far as I know, the shuffle does take this hidden last 10 matches Elo rating system into account, doesn’t it @Amerika ?)

So maybe we go super crazy and try balancing by levels? I mean, it can’t be worse than it is now?[/quote]

Well… It wouldn’t really fix anything. I guess it’s something we could try but I think it would do more harm than good.


(doxjq) #32

Amerika is 100% right, and it’s been said so many times. Balance in an objective based game like Dirty Bomb is virtually impossible and it is never going to be perfect - The sooner people realize this the sooner we can move on. Even if you balance people by their “skill” or “slaying ability” there’s so many other factors to consider. Solo pushing, positioning, team synergy, merc composition etc. You can pair people up by similar skill but you can’t force them to play smart and work with their team.

I think the only FPS game I’ve ever seen that has relatively good team balancing is clan arena in Quake Live where we often see 10-9 games. But this game everyone is the same character, everyone has the same health with all guns, and the only objective is killing the enemy team. Maybe Dirty Bomb could achieve better balance in a game mode where everyone had to use the same merc and there was no objective to do, but that isn’t going to happen.

I mean honestly until about a week ago we have been playing 5v5 pugs on the Australian DBN server every single night with the top 10-20 DB players in Australia and I can safely say out of the 500 odd games we have played maybe less than 5 of them have been super close - It’s just the way objective based games are. We’ve even had the odd time where we purposely pick teams based on individual skill to try and get perfect balance and it just doesn’t work.

In our experience with pugs, synergy is just too important in Dirty Bomb. Lesser skill players who work well together are going to absolutely stomp incredibly skilled solo players.


(bontsa) #33

That was very mature approach and a great way to advance in your ways of thinking and perceiving the subject @seashellInspiration , kudos! Apologies for putting my earlier comment in quite blunt form, as said the discussion is one of those “forever-subjects” around these parts and it gets tedious, frustrating and repetitive after a while. However feels nice to have conversation like this have a better direction for once!

Since there was a notable bug in how PSR was calculated, I don’t know how shuffles are exactly going to work from now on. As said it cant take into account the person’s state behind the keyboard and monitor, neither can it probably be utilized to take into account people’s mercenary selections that well.

This bug was fixed in last weeks hotfix, but I’m still not entirely sure was it lost in the rollback that was put into place a day after, or are we going to see the fix again with next update. Time and patchnotes will tell. It should help how shuffles work, obviously not a silver bullet solution but nontheless. From my own personal experiences I haven’t been seeing those “all pub-heroes being put on same team” type of failures matchmaker did pull out from time to time since last week, but that sample size is bit too short to be saying anything concrete yet.


(K1X455) #34

Team cohesion. I get to experience this in US West servers more often than in ASIA and AUS servers combined. A bunch of unknown players who know the role more than the average joe and has reasonable game sense brings together a very close fight.


(GatoCommodore) #35

sometimes i have to shout and cuss like drill sergeant in the full metal jacket to get my team moving.


most people will laugh because i sound like an old man.

and it actually works a few times.


(K1X455) #36

I was just in one just right now. Lost the game because we couldn’t make it in time to get back from respawn, but hell… that was some awesome teamwork by ammo support (pro passing by Arty and Skyhammer), obj specialist doing their role in moving and repairing EV, medic station placement by Aura, remote revives by Sparks, and wouldn’t you know it, Vassili running the objective container after throwing heartbeat sensor. You don’t get to see such things happen very often.

Sometimes, I keep losing faith in this game for obvious reasons. Then this happens. You can’t get this much excitement sitting down in front of display.


#37

There are some well written posts here. :slight_smile:

The average player is far too focused blaming the supposedly poop balancing system while negating any other viewpoint such as team composition. It’s definitely not my team’s fault for having three Vassilis who can’t hit a standing target while we are unable to prove atleast a single medic, right?

I have even met some people believing that the match is only balanced when they win.
They disconnect when they are losing or due to the enemy team having a high level player.
That leaves their team in the dust with one man down.

Let’s hope the majority of players will realize the mentioned points and understand what to expect from Casual Matchmaking. It won’t cast a miracle ending every single match in a perfect draw because that’s simply impossible.


(Amerika) #38

[quote=“Jostabeere;214967”]Shuffling based on a hidden Elo which counts last 10 matches fails miserably.

(As far as I know, the shuffle does take this hidden last 10 matches Elo rating system into account, doesn’t it @Amerika ?)

So maybe we go super crazy and try balancing by levels? I mean, it can’t be worse than it is now?[/quote]

It’s your last 25 matches @Jostabeere. And yes, the shuffle uses the same MMR system as the pre-game shuffle before the start of the match.

I think I went to a lot of trouble to state why it wouldn’t work. If you disagree with what I said then quote it and write out your logic as to why I am wrong and why sorting by level only would work. And no, “might as well try” isn’t something people in development work do in a live situation. Especially when there is compelling reason not to do it that hasn’t had a proper counter-argument against it presented.


(pumpkinmeerkat) #39

In a causal, pub setting players can and should be able to play the way they want. There is no way to guarantee a balanced and competitive match every time regardless of the metrics used. Plenty of players value playing in a way they consider fun or trolling over winning so any attempts the game makes to balance a match can only do so much.

No mention of casual matchmaking in recent dev-update videos…Fingers crossed for community servers and map editor soon™.


(Amerika) #40

As @Dox said, matchmaking in a game like Quake Live Clan Arena is extremely simple as everyone is completely even across the board in every regard and there are no objectives or even something as simple as a map side that has to be considered. It’s 100% balanced on your ability to kill other players consistently. This too doesn’t take into account your personal situation but it will be much easier to get a properly balanced match. DB just has too many variables much like any games/sports with objectives and sides. CS is also easier to balance due to the lack of classes…but only a bit.

And then there is OW that has more people playing than anything else on the planet outside of CS (fps-wise) and it has a working casual MM system (which also does not factor in your account level) but instead uses a private MM score assigned to every player and matches still end in stomps a lot of times. This, almost always, boils down to a lack of team synergy and class picks and sometimes just bad timing/randomness (it is OW afterall). So you take a huge pool of players and you give them all skill scores that are extremely similar and then you create matches based around that and do you end up with perfect balance? Nope.

My point is you can try to do as much as possible to make balance as good as it can be with perfect conditions. But in the end, with a game like DB, it’s going to come down to the people on each team doing what needs to be done. And if you get both teams doing that you’ll still end up in matches that aren’t very close both in your favor and against.