Prevent match dodging for ranked games

[sharp elm]

I see this happen alot, especially in 8v8 games and I think a strong arguement could be made for why dodging is bad for balance and player experience. I would love to see all ranked modes pool players into one group and automatically split them upon game start instead of revealing teams where people can just decide to back out. Burden of balance shouldnt be on the players mind like this imo judging who is a coinflip and who isnt etc etc to exploit the system. Am I the only one tired of this?

[elfin anvil]

As long as there is a system that allows you to join, say, in pairs, I'm in.

[sharp elm]

Absolsutely, I cant think of any reasons why the party system couldnt continue to work.

[fallow depot]

You cant garantee even closely balanced games, if players start at 17ts, and premade teams are allowed. If you have on one side premade team of experianced players (of 8 players, to underline my point, or less, to be more realistic), and on other side new players that join alone, you will have totally unbalanced game. 1-2 ranked players who dont even know basics on one team, and 4-5 ranked players that totally deserve their around 17 ts, on other team. P.S. I think premade teams are not the only things used to make unbalanced games in lower ratings.

[elfin anvil]

There is a possibility of limiting premade teams to, like, 3 or 4 people.

[undone thistle]

Mixed skill lobbies are probably the biggest structural problem with creating competitive(and fun) pvp games in BAR. Throwing everyone into one blind pool of players would just make that problem worse. Additionally you did not make an argument for WHY dodging is bad for balance and player experience. If players think they know better than the balance algorithm they should be free to use that knowledge.

[fallow zinc]

If someone wants to dodge, I say let them. Stopping people from dodging would cause its own problems. If you think there is a good argument, then make the argument. It doesn't really cause me any trouble if someone decides to dodge. Why force them into a game they don't want to play? One thing I would like changed is having the 'Join' team button removed when autobalance is on, because it falsely gives the impression people can pick their team and when they try it's a nuisance as they get automatically shifted back quickly and you can't even tell who is doing it easily, but it disrupts the game from starting as someone tries repeatedly to do something that doesn't work.

If the system has a flaw in rating people people correctly, like the current one does with new players, which is causing people to dodge some games, then the flaw needs to be fixed, not the dodging behavior.

[sharp elm]

The irony is, dodging disrupts the integrity of the rating system. If you dont want to follow the rating system you should play unranked, simple as that. Saying you need to dodge because the rating system is comprimised makes no sense, it just means other issues need to be addressed no?

I understand dodging or manual balance as means to bypass certain flaws with the current system, but itsnt a proper solution, its actually part of the problem.

[gritty aspen]

i wouldn't dodge if games were balanced

[sharp elm]

Ppl need to understand dodging is actually contributing to games being inbalanced rofl thats the point im trying to make. OS system will never improve if ppl just balance it themselves all the time and call it "ranked"

[gritty aspen]

yeah because i totally want to play with tx55/zow or a 40 os strait player trying another map for the first time

i'll rather leave that fun for someone else

i think if it would be random that you get one of these some people wont play at all

[sharp elm]

Funny thing is ppl dont dodge a good matchup tho where those ppl are on the other team xD

[gritty aspen]

that argument is even funnier considering that for a lot of ppl os doesnt move at all

[gritty aspen]

yep

[sharp elm]

There are alot of ppl who would get heavy corrections if dodging was removed

And it would probably be more volitile OS, at least for a time

[gritty aspen]

how would removing dodging affect their os

[sharp elm]

Ppl who normally dodge bad matchups would be forced into statistically fair odds for their OS

[gritty aspen]

how about os balancing them properly when they dont dodge

[sharp elm]

Even when they dont dodge, if other players are still dodging it effects everyone regardless.

I dont want to be forced into a bad game because I dont care enough to dodge

[gritty aspen]

but it is what you suggest?

force everyone into bad games

[sharp elm]

Ppl are creating an environment for these bad games by trying to avoid them lol

[gritty aspen]

and how would you balance the cases i mentioned?

[sharp elm]

Really the algo needs to be good enough to do it on its own

[undone thistle]

We assume that people dodging up their winrates by doing so. We have no real evidence of that however. For my personal stats for example, I have a list of people who I lose with the most. Those are not the people I dodge(I usually dodge maps, but thats another issue). I dodge trolly people or people on tilt. What we do know effects OS visibly is playing in silo'd lobbies, such as straits or glitters. If your goal is really to make balancing better, then having a blind player pool is less important than having a random map played.

[gritty aspen]

tx/zow don't follow any os, they play from 1 to 50 depending on how they feel
a 40 os strait player would ruin a game and go back to strait

[sharp elm]

Tx is banned so u dont need to worry about him anymore lol

[gritty aspen]

thats just one player

[sharp elm]

Randomized maps for comp playlist also a good idea I agree

Generally for competitive matchups the less in the players control the better. I understand the bar player pool isnt as big as AAA games so we need to make the most of what we got in terms of players

But dodging really does seem like a chink in the armor and ways to mitigate it would be great

[gritty aspen]

dodging is the least of the problems

it at least lets people to not ruin their days because of os flaws

[undone thistle]

I'm trying to think which players I would consider overrated because of dodging. And no one is coming to mind. My knowledge of the playerbase isn't comprehensive, but its pretty fair. There are people overrated because of playing straits/glitters, or because they stomp in small team games, or because they were early adopters of whatever strong meta strategy/units were flavor of the month. But overrated because of dodging? Hard to say.

[sharp elm]

I would also say its possible some players are under rated due to being dodge also

Karolson dodging zow too much he cant get high os :/

Cmon karolson help him out abit xD

[gritty aspen]

i dont even dodge him personally

[sharp elm]

Yea im just jokin lol. It is interesting to think about tho. Who knows maybe game would flop if ppl couldnt dodge but id still be curious what would actually happen.

Im sure it wouldnt be as bad as anon teams tho. Same as anon FFA but for teamgames xD

Full ingame names everyone names blanked out maybe too much.

Could always bring back the captains draft in a way that discourages dodging too i guess

Just food for thought

Currently long queues in prolobby most effective dodge discouragement probably

[gritty aspen]

how would that work? i would win if i play against my dodge list

[sharp elm]

Nah i was just saying if u think zow is underrated or something and if thats why u dodge him (i have no idea) then u should give him some os xd

[gritty aspen]

i said it's completely random if he plays bad or good

[sharp elm]

Yeah i think its pretty common to dodge coinflip players

Hard to define what a coinflip player is tho sometimes, esp using stats

Would be nice is OS could account for it better, uncertainty probably needs reset/new season soon

[gritty aspen]

i'm pretty sure they sometimes deliberately troll to keep os low

[final dune]

Good points on either side but how on earth would you prevent dodging without increasing already long lobby times alot

[fallow zinc]

Who says it's competitive just because it's ranked? The 'ranked' is just a rating system to help balance games, that doesn't mean it has to be competitive to take all these extra measures to make it competitive

Also, player base is small, you can't do the same things other games can with much larger player bases

[gritty aspen]

in most games unranked is same as ranked but with a separate hidden rating

[fallow zinc]

there is no such distinction in this game I don't think, unranked means no rating

[sharp elm]

How would discouraging/preventing dodging increase matchmaking time? Wouldnt it make it faster since dodges wouldnt keep preventing the game from starting? Also considering BAR has a seperate leaderboard rating for players, would that also not count as a competitive metric?

[sharp elm]

Reconcile this with leaderboard rating tho plz

[fallow zinc]

Some people play casually, others play more seriously, the presence of a leaderboard doesn't define everyone into a high level of competitiveness just because some are competitive or happen to be successful at it. There are always going to be top players at any game, regardless of how seriously everyone takes it or their reasons for playing the game. There aren't a lot of players for this game, so we're kind of all lumped together by necessity just to be able to play at all.

[final dune]

since bar doesnt have matchmaking so it would have to be integrated into the existing lobby system somehow

I cant think of an elegant solution personally

[sharp elm]

Bar does have a matchmaker server side technically but nobody uses it.

[final dune]

8v8 matchmaking isnt viable for a bunch of reasons discussed in other suggestion threads

[sharp elm]

It would be easy enough to group everyone into one team probably and automatically split them on game start

Regardless you wouldnt need a matchmaker for what im proposing, just an automatic pregame autobalance which is hidden from players. So all they could see is the pool of players they will be split from basically.

[sharp elm]

My point is that as long as there are competitve rankings or achievements in a game, steps should be taken so players cant manipulate the system. Otherwise leaderboard etc can all become meaningless.

You cant maintain competitve integrity when you allow ppl to cheat, dodge games, select their opponents or maps etc. Variables need to be controlled so they arent abused by players.

Casual/unranked is a different story of course and should have a seperate preset which doesnt limit the player

[vocal lion]

How would you even enforce this? What if someone decides they want to get lunch or something; will you punish them for going to spec?

[sharp elm]

They can still go to spec before the game starts

I am simply proposing just to not show the teams until people are ingame, instead only showing the pool of players, maybe each team captain and who is partied in the lobby

[vocal lion]

What constitutes dodging? Leaving before game starts, or only after game starts?

[sharp elm]

Dodging is a motive so its not exclusive to either

Some ppl forget to dodge in lobby lol

[vocal lion]

It's hard to prove a motive of someone unless they explicitly say--which brings me back to enforcement. How are you going to enforce this and demonstrate that people intended to dodge ?

What if friends party, intending to play together, get put on different teams, then dodge because they can't play together on same team. Would that be punishable?

[sharp elm]

im not proposing punishment

i just want to make life as hard as possible for people who try to get easier matchups than their rating deserves and prevent them from playing statistically favorable outcomes when it comes to maps and players. If players can exploit, its not a matter of IF, its a matter of when

party system should be handled by the algo so friends should be on the same team regardless

[vocal lion]

Party system often breaks up parties to maintain balance. Sometimes a !rebalance fixes it.

What you're suggesting seems more conducive to a matchmaking system. Kinda the point of a lobby system is to be able to see who you'll be playing with--all voluntary.

Maybe a compromise would be a lobby option that scrambles teams at start of game.

[crisp hatch]

How would parties work? What if you want to play with a friend and can’t even see if you’ll be playing together

[sharp elm]

It should definitely be possible to reveal who is partied pregame but not reveal entire teams. You could maybe also reveal team captains pre game

Sure some things might need tweaking but its doable i think

[rich garden]

Parties get broken up if they'll mess up the balance too much. Would be hard to show whether they'll be broken up if teams aren't decided yet

[glad skiff]

Sorry but I'll just resign if I'm on a team that's 100% a loss. It's not fun to play with certain people as well. I get matched up with people I have on ignore and avoid, so I need to manually dodge that game as well as smurfs at high and low OS that aren't being moderated. We even have big clan Bois now who actively throw games because they think it's funny and there isn't a damn thing you can do about it other than ignore and avoid them manually. The only way I'd eccept this if we get a true StarCraft like matchmaking experience where you queue into matches and ratings are reset and you can only get official rating through matchmaking. If someone is found to be a Smurf then rating should be retroactively restored and ignored from any games a Smurf was in.

[sharp elm]

Why arent you reporting the smurfs for investigation? Also yeah throwing games is also punishable just report them then. As far as not wanting to play with certain ppl its a valid concern, but what im proposing you can still see the pool of players, just not the final teams so you would still be able to leave if you see someone you dont like that would either end up on your team or the enemy team.

I agree about matchmaker rating could be similar to sc2

[glad skiff]

I prefer seeing the actual team instead of some theoretical pool of players. Im not saying rating like sc2. We simply need non lobby matchmaking system like any other rts or moba game. I want to be able to solo queue with others of similar skill.

Reporting players that decide they aren't giving it 100% is pointless. We've seen it with players like Zow and nothing ever happened. Clan members can decide if they want to just play a chill whatever game so their buddies have an easy win. You can't report that. It's better to just avoid them outright.

[glad skiff]

Slightly off topic. They are currently developing anti cheats that use AI that build a profile on players. When it's finished maybe the bar team will implement it. In theory it can detect who you are even if you've stopped playing for years. It will even know if you had cheated in any other game within it's database and ban you from that. Untill we have those systems in place it's better to not take online competition serious, especially FPS, what a cancer that's become.

[vocal lion]

That sounds scary and dystopian.

"It says here in our records that you cheated in Roblox three years ago. Do you know what we do to dirty cheaters around here?"

[craggy wagon]

Predicting outcome of fight is a strategy skill too. Especially when OS system is not balanced.

[sharp elm]

as I said, OS system not gonna be balanced if players can keep dodging/inflating themselves until favorable odds are met. Its like saying I have to cheat because matchmaking is too hard

majority of competitive games dont allow dodging for good reason

[vocal lion]

If someone's OS is inflated, wouldnt that work against them in balancing? Also can't gain much playing vs newbs. Eventually they would plateau, right?

New players also really mess up matchmaking. If someone sees they have a bunch of 1 Chevron 17s vs a bunch of r3-4 17s, technically "balanced" by OS but clearly not balanced, I wouldn't blame them for wanting to dodge, lol.

As a side not, I'd warn against putting too much emotional investment in OS. It's not a huge community; people who've been around generally know who the good players are, regardless of their OS. Don't necessarily need OS for recognition.

[sharp elm]

its not about OS really, but dodging ruins games for other ppl

I think we kinda already covered just because OS isnt a perfect system, it shouldnt be a warrant to dodge. it just needs to be improved in other areas which id like to hear more about

[undone thistle]

I'm still so unclear on how dodging ruins games.

[sharp elm]

not only dodging, but statistically favorable manipulations. only playing when certain conditions are met skews results.

[final dune]

Dodging inflates the people who abuse hence less balanced games is what I took from it

[next jungle]

Tbh I think the nice middleground would be to show players their team composition along with a reshuffle vote before the loading starts, this way the dodgers could dodge without wasting our time for a stop vote and other people could actually try to 'aid the matchmaking with balancing'.

As a sidenote I kinda like the inflated players since it's pretty fun to see them schooled / play in an unfamiliar role 😄 .

[undone thistle]

I can cite numerous examples of people playing maps/roles they're not familiar with and underperforming(imo) relative to their OS, and trolling or ruining games. I cannot think of even one example during or after a game where I thought to myself, "Oh, this game is terrible because XYZ player dodged." I'm sure we can all name players who are 40-50 OS on certain maps, or good at a certain role, and are worth 20 or more OS less outside that niche. I can't think of even one good example of someone 40+ OS who got there via dodging, much less a list of of even ten such players. So I still see this anti-dodging mindset as a solution to a problem that does not exist.

[stark lotus]

if you are gonna dodge dodge in the lobby thanks

[blazing spire]

We could improve OS and balance so that individuals have no statistically meaningful residual information on which to avoid games? (Or at least increasingly minimal)

Given that I presume the most common/powerful type of match dodging is avoiding playing with overrated new players, matches could also be analysed historically to see who is or hasn't had their statistical share of matches

Although I don't think one should take any explicit actions or infer anything intentional from the results since it's not against the rules

[crisp hatch]

In my experience dodging has not occurred to be a problem. I’d say that focussing on certain maps or team sizes skews the balance more. However these people tend to stick to their favourite mode which evens this out as well

[next jungle]

Not really, dodging is more of a strait problem because you can easily tell whether the crucial roles are going to be filled by someone competent or a dud.
For example having a weak air player is quite strong indicator of a loss and that information is easily inferred due to the relatively small pool of players.

[stark lotus]

dodging prolly can inflate you <5 os. it's similiar to partying ig, refusing to play when you predict the team matchup is gonna be poor

not a ton compared to role-based skill, but still something.

[next jungle]

tbh I’ve played a bit during the lobby stacking meta and getting 7 ts or more in it was relatively easy.

If you’d meticulously weight each match I think you could get like 10ts due to a bit chess like nature of the spot counters.

Also we all know one top player who chooses his game very carefully :]

Edit: I mean I was playing in a stack with Cymraeg and got boosted from like 26 ts -> 37ts

[main coyote]

I don’t see this as an issue
If a player does not want to play in a lineup they shouldn’t be punished for that

[undone thistle]

If you're on voice and or party with people your WR is going to go up, barring some sort of wacky lobby. Communication matters. That is an entirely separate issue however. And more significant imo, because balancing cannot take voice coms into consideration.

Players however can recognize it, and dodge if they so desire.

[next jungle]

Yes dodging is ok, though It'd be nice if gui would allow to dodge before the game loads so people wont waste 5 minutes in lobby simulator

[sharp elm]

Me with individual player stats pulled up on each map and defined roles vs little timmy playing his first game xD I will dodge little timmy on sight he will ruin my games :v

[undone thistle]

is there an app with defined roles on it? Furyhawk's and MBT's never had role included. They were also rather too slow and clunky to sort map WR's for me to ever consider using them to dodge games.

[stark lotus]

https://youtu.be/f8j5ZfkQBPs?si=EWlR0tcw0nxODkOe

[blazing spire]

I've not seen anyone put one together yet (and make it publicly available), but the start position and map is theoretically there and already “easily“ extractable, so we can see the long term trends of player start positions on maps/games they actually complete.

One of the questions I commonly get is why attempt replay parsing without running the simulation, to which this would be an answer too. Merely observing the unit types and commands in their stream would likely let us cluster them into more empirical roles

We can't explicitly capture voice communication, but it would (presumably) be a residual of who you play with in our community along with general communication and teamwork. And who you play with we can observe (and presumably detect statistical deviations from expected both in terms of frequency and win rates)

[sharp elm]

teamwork/coordination is an ingame feature and not comparable to dodging imo

[blazing spire]

I've assumed the overall effect is quite large, as we have to remember OS is always a relative number as well, and the community is somewhat mutually exclusive (in that if one player dodges, it is implied the remaining players end up on teams that are statistically worse, so your +5 clyret bunjie is balanced with a negative effect distributed out somewhere else in the community as well)

[blazing spire]

Well firstly, I just want to make explicit that without randomised matchmaking, I'm not sure we can ever really solve match dodging beyond improving our match rating and balance algorithms. And I think there's good reasons to think we won't have randomised 8v8 match making take off any time soon

And also I don't think match dodging itself should really be treated as any kind of actionable offence

[main coyote]

Players I doubt it’s resolveable without punishements as the lobby for 8v8 ussualy takes 2 mins to start the next match

[blazing spire]

I don't think we currently capture games that don't start? (I hadn't really looked at it to be honest...🤔)

[sharp owl]

If you try to stop dodging in the lobby, people will just Alt+F4/resign at the start of game, leading to a !stop, wasting everyone's time.

There are many legitimate reasons to dodge games:
a) You only want to play with your friend, but sometimes the auto-balancer can't figure it out
b) The autobalancer puts 2-3 new (1-3 chev) players on one time, whilst the other team is all experienced players (5+ chev). I still think that starting OS of 17 is wrong - maybe the balancer should use rating - (uncertainty * 5) instead of *3.
c) Wanting to avoid playing with certain players
d) Wanting to avoid playing against a 4-8 person (yes I've seen 8) party who are all coordinating in discord
e) You don't want to play a particular map (e.g. sea map)
etc.

These problems can't be solved until the playerbase is larger and matchmaking lobbies become dominant over regular lobbies.

[stark lotus]

true

just kickban people if they repeatedly dodge after game starts
wasting plp's time

[sharp owl]

In my experience that's extremely rare. Ironically it usually happens because someone else dodged at the last second and the teams got shuffled around, leading to one of a) b) c) d) above.

[slender cliff]

This isn't a solution because its a different person everytime.

IMO autobalance should be more true to its name and actually automatically apply itself at the beginning of a successful start vote and before the game loads. There have been far too many times where lobbies get vote after vote of spec this player, lets play this map or that map again and again over and over. It gets so bad until the point which people are so frustrated with voting that they just want to play the game asap. Then sometimes what happens is a few more players will join / leave possibly due to match dodging and it happens so quickly that you cant even tell who joined or left because the suddenly the start vote appears and players pass it without calling the autobalance vote again. So someone complains at the start of the game (usually myself :p) that autobalance didn't run and we have to restart the match.

So why can't we just have an autobalancer that actually runs automatically each time the game is about to begin? Wouldn't that counter act the dodgers because they can't see who's on their before the game starts and leaving after the game loads could be enforced more heavily.

[wheat lark]

Could somebody quickly tell me what dodging is?

[fallen oracle]

leaving a game before or just after it starts

[final dune]

looking at both teams and thinking that ur team is at a disadvantage due to people not being good at the map, inflated etc so you leave before it starts to save a "loss"

[wheat lark]

Do these people leave the lobby before the game starts or after?

[final dune]

normally before

[wheat lark]

Because making 15 people restart a game because you decided way too late that you don't want to play is punishable. But skipping out of a lobby because you don't fancy your odds? In a game that doesn't even have automatic matchmaking? How on earth is that something that should be punished?

Some people are addicted to making their little number go up. It makes them feel a little more powerful in a world where we're all really powerless, and if that's what motivates them to play then I say let them have it.

[sharp elm]

I agree, but when dodging negatively effects other ppl its a no from me.

[wheat lark]

I can't imagine a scenario in which it negatively effects other people in a way that's the "dodger's" fault.

If autobalance can't make the teams even when somebody skips town, that's just a limitation of the algorithm.

[sharp elm]

new players dont get matched with competant players when they are always getting dodged by them

[wheat lark]

Or an inevitable byproduct of a lobby including a large number of vastly differently-skilled players.

Well that kind of dodging probably isn't fear of OS loss, it's knowing that they're effectively going to be down a player. Personally I think autobalance should balance around playtime second after considering OS.

I know what it's like to play with somebody who didn't even bother to play a couple single player skirmishes to learn the basics, and it's really disheartening. This game is a fairly big time investment and people don't want to waste it on an exercise in futility.

[sharp elm]

i strongly believe limiting ppls ability to dodge will encourage improvement of the algorithm. its not about punishing players but more so limiting them so bar can have proper competitive integrity and all players get fair treatment

[wheat lark]

Punishing somebody for leaving a lobby before the game even starts is messed up, regardless of the reasoning.

[sharp elm]

like i said, no punishment, just dont show teams pregame

[wheat lark]

There needs to be an ethical solution

Well then people will just leave the lobby if there are inexperienced players regardless.

They won't look at teams, they'll just see new players and leave solely on the possibility of getting saddled with them.

[sharp elm]

but if new players are on enemy team they would stay?

[wheat lark]

Possibly. I just don't think it'll cut down on dodging

Might just encourage more minimum playtime lobbies.

[sharp elm]

dodging lobbies is better than dodging players imo

[wheat lark]

Where anybody below 3 chevrons just gets booted

[vocal lion]

Again, a big issue is how the balance system splits up parties. It's like, really really bad at keeping parties together. It would be annoying af if you couldn't even tell you will be on the same team as your friend(s) before the game starts, and starting a game just to see your friend(s) on the other team.

I would be fine with a lobby OPTION to scramble teams upon starting the game . . . But again an OPTION that players can see is on before they get themselves into something they don't want to do.

[sharp owl]

Last night the balancer wouldn't allow my friend and I (30+31 ts) to party even though there were 38+27 that could be paired against us.

[slender cliff]

Parties in a lobby should have some sort of visual indicator. Number, icon, same color..etc how else are other player supposed to know if some are in a party or not?

[blazing spire]

And of course the other issue with parties and dodging is not just the party members doing it because they're not in the same team, but the other players who dont want to play against a stacked/unbalanced/partied team and would dodge if they knew it was working

[candid saffron]

Reading some of this stuff it sounds like being able to see who is in the lobby, who is in a party together but not knowing what teams anyone will be on would be good

[cerulean burrow]

I think it would just end up in people leaving the game after it starts more. Sure that's more punishable, but effectively the system is now forcing people to start games they don't want to play. That's not fun and will likely drive those players away.

[worldly iris]

I don't really think this is that practical for most lobbies. There are alot of lobbies where there is no real way to balance them as you have an odd number of players significantly higher than the rest of the player pool. This sounds good in higher tier lobbies only imo.

[stark lotus]

re: #1205149621861617754 message
this is aside from the leaderboard, but personally I like to take OS with a grain of salt

the 95% better way of seeing how good someone is to watch their gameplay

[wooden kayak]

Maybe I'm not understanding the OS and balance system, but wouldn't dodging high OS players in an attempt to boost yourself just result in the need to carry worse and worse allies against a better and better enemy?

And don't get me started on the party system, I just want to play with 3/4 friends and we get split so often they're talking about going back to Dota

[gritty aspen]

high os = worse team, no dodging required

[warped bolt]

I think it's a bit more nuanced then that tbh, high os = higher chances on worse team, lower chances on favorable matches, but they are still very much there (some players aren't super high in OS but are definitely better by quite a margin in comparison to similar players in their OS range) thus the high os can still opt to play matches with those players and still get favorable matches, only problem for the real dodgers is they need to dodge a whole lot more the higher they go, there are some cough known players cough that dodge more then half of the matches they queue up for

[next jungle]

It makes me kinda wonder what would happen if the dodgers were tasked with balancing the teams

[stark lotus]

Still get to cherry pick games and big pp number

But yes, high os is a curse

[candid saffron]

you can see who is in lobby
game tells you if your party is going to be able to stay together
you cannot see the overall teams the game has decided on
once start vote begins if a player leaves the start vote is auto cancelled to give people a chance to leave if their party is broken etc
leaves within 5 mins result in a loss even if the game is stopped
there u go dodging is fixed

[icy sky]

Dodging in lobby is good when compared to the counterfactual of instead leaving mid-game

For me, the ‘negative’ of allowing people who want to manipulate their OS higher by only taking favorable matchups is sufficiently justified by:

1. allowing people in parties (who just want to play with their friends) the ability to dodge should the balance algorithm not place them on the same team, and
2. Giving people who don’t want to play together an out (i.e. dodging based on your blacklist). There was the odd player, who has since been perma-banned, that I’d find so frustrating to play with, I’d rather listen to the sound of nails on a chalk board

[tribal stag]

How about giving negative OS points for leaving a full match that has not just finished a game? Say give them 30 seconds after a match to leave. If they join, ready up and the match is full and ready to start, they decide they do not like the team up they dodge and take a small OS hit that is less than a regular loss, but sufficient enough to deter the practice. Granted it could be abused by players that want to lower their OS for easier matches... If they do it often enough give them a special icon that indicates a mark of shame. The mark will disappear after they play enough matches without repeating the behavior. Being banned from a match or forced to spectator by vote would not incur the penalty. Again some people might abuse the mechanic to avoid the OS loss, add another system that checks for frequency of forcing to spectate or being kick banned. Or we can hope people will grow up and just play the damned game and learn to carry bad players which the high OS should be indicative of.

[sour skiff]

One thing that I think has to be said is that choosing specific matchups and balance to play before game start doesn't manipulate the os system at all. All the os system does is assign value and probabilities and whatnot to those who are in the game. if you are in the game, you are providing rating liquidity for everyone. There is nothing statistically broken about dodging

[tribal stag]

It is a direct manipulation of the statistics. You care looking at the odds and doing a quick assessment as to whether you think the odds favor you or not and then leaving the match. A high OS should be indicative of ability to carry bad players. A high OS player should not need to be looking for stacked teams. I see a matchup that the OS says is balanced, 3 1 chevron 17s on one team and the other has 3 chevron 14's plus an extra 9 points of OS to feed into another experienced player. Statistically that is an uneven matchup. Those 1 chevron players are quite likely to be new players making their first foray into pvp. An OS seeker is going to dodge that team because they know they cannot carry. The OS system needs to take PVP playtime into consideration. AI's and humans are completely different animals.

What is statistics? Statistics is the science concerned with developing and studying methods for collecting, analyzing, interpreting and presenting empirical data.

Another definition "Statistics is a branch of applied mathematics that involves the collection, description, analysis, and inference of conclusions from quantitative data."

[sour skiff]

Statistically according to OS there are no stacked matchups that will occur in an OS balanced game. And by playing in them, people who are "abusing the OS system" are increasing ranking accuracy of those who are misranked

[tribal stag]

OS system is broken.;) Thus we are having this conversation.

[sour skiff]

You literally don't know how it works obviously lmao

[tribal stag]

https://www.beyondallreason.info/guide/rating-and-lobby-balance

It is kind of spelled out right there.;) OS system is not working to get balanced and fair matches.

[blazing spire]

Under the assumption that OS works. But we know that OS assumptions don't model the population right, it doesn't model win probabilities perfectly, it doesn't model beginners at the optimal amount, it can't take into account team or map specialisation.

Plus, we use mu-sigma for match balancing, not the internal OS values,

[sour skiff]

Yes, but that's not a statistics problem, that is a game problem

[blazing spire]

Lemme see if i can share some little side-results too...

[sour skiff]

What some people seem to think here is that dodging breaks the statistics underlying OS rating, when it objectively does not

[blazing spire]

Well, i mean "breaks" is kinda subjective...

I think it's true that players are arbitraging differencing in actual win probability compared to actual match starts in order to maximise OS gains...

OS still works pretty well though.

[sour skiff]

Certainly. But by doing so, they are similarly adjusting their opponents OS to be more accurate relative to their own, which is working as intended

[blazing spire]

Well, more accurate in the sense that other players play in worse teams, and therefore their OS is calibrated lower. Players who selectively play in better teams have their OS calibrated higher.

But I think it's fair to say that there's an issue in the sense that OS isn't an individual statistic, even though it's notionally attached to the individual account.

Whether that's a problem 🤷‍♂️

[sour skiff]

By selecting for better or worse teams, they are working to calibrate all other players in the population though.

We know that not all players are accurately described by their OS number, that's what this is all about. We know that the difference between the OS rating and known actual skill is "exploited" by people dodging. We know that OS ratings and changes are dependent on the OS number, not the actual skill level. And we know that this "exploitation" necessarily works to minimize the difference between someone's skill and and statistically based OS rating. I don't see any problems

Yeah obviously there is not perfect population mixing and game-specific things like matchups, maps, etc. But in terms of the long term total population equilibrium that OS approaches, that we are assuming can exist by using OS at all, dodging does not affect that

[dense trench]

I wonder if people can actually significantly boost themselves by dodging. I encourage you guys to play your normal matches and before the match starts you call if your team will lose or not, then put the results on a spreadsheet or whatever. I wonder if any of you will be right more than at least 60% for dodging to actually make a significant impact.

[cerulean burrow]

It's usually pretty easy to tell in the "8v8 glitters all welcome" or any other lobby with "noobs welcome" in the title. A bunch of 17 OS 1 chevrons on your team? I'ma call a loss. I play them anyways because I don't join those lobbies to sweat try hard, just to play the game. Someone paid for my first few pvp games where I definitely single handed lost the match for everyone, gotta pay it forward.

I'm guessing the dodging issue isn't prevalant in those lobbies though

[stark lotus]

if you know the players well you can tell

player who farmed all their os on different or noob lobbies will be inflated

[blazing spire]

Now, firstly I want to make sure people are aware I'm not accusing players of willfully abusing the rating system, or necessarily actually dodging games (although I do want to see if that can be calculated/observed in the future)...

Performance can be as simple as rocking up and playing at a particular time or place on your favourite maps with the group of players who happen to be online...

But as part of the tau investigations, one thing i did was independently put out the internal OS library win probabilities for the observed matches...

At the end of the year, i can look at who's where on the leaderboard, and I can see the running average of estimated internal win probabilities for their historical matches.

Put in relatively plain english: I can look to see whether players have a long term trend of playing matches they are statistically more likely to win in.

As judged by the internal mechanics of openskill.

Further, I can then take an average of an average to see if there's trends that seem to form amongst groups of players.

So this is distribution of historical 50 match running averages of win probability for players in the top 30-60 ranked players: obviously still eminently skilled players by any measure

And this is the distribution for the top 30:

Now again, none of this implies conscious manipulaton per se. And honestly I'm not even willing to say conscious manipulation is wrong, everyone should be able to play with who they want to play with as far as I'm concerned.

If players want to try to maximise OS, or just play with their friends, or just play on a handful of maps, or not play "misbalanced maps", or just log on every night and play only with the group of people who are on at that time in particular rooms, i really don't think i should care.

[eternal arrow]

it was fun waiting 5 minute every game for MBT to look up his teammates and enemies winrates per map and then dodge half the time

i think he was onto something

[blazing spire]

An of course we can go the other end of the spectrum as well, who's playing more in shitty games and matches...

And we can see who gets an over or under allocation of playing with/avoiding noobs

[eternal arrow]

isn't that likely to be affected by stuff such as "player has high rating"? I naturally get more noobs on my team by design because my rating is higher

[blazing spire]

short answer: yes 🙂

And that's also partly why I again want to stress, misbalanced odds are not necessarily a sign of malfeascance.

But we can also analyse a few other things: not only whether one has more noobs, but whether one is consciously say avoiding members around say 16.67

And again, i don't think there's anything inherently wrong in doing so.

[stark lotus]

i repeat again if you know the players you can tell

[eternal arrow]

another factor is that matches aren't balanced 50-50, for many games my team often had, say, 20 more points than the enemy team, so I would lose 3x as much rating as I gained, due to the prediction being that my team wins more often than not--so you have balance algorithm influence on these numbers, along with the fact that retroactively looking at games often ends up with the winning team appearing as if they would've won all along as the winning team gains os and the losing team loses os

[stark lotus]

but less so for all weclome lobbies unless there's a real stinker on your team

[eternal arrow]

a match that is balanced 50-50 at the time of playing will skew toward 55-45 or more after the winning team gets rating updates

[blazing spire]

Yes, and the balance algorithm is likely not neutral.

[eternal arrow]

In certain playerlists I could be 60 os or 100 os and receive the same teams due to how games are balanced, the only thing that changes is my predicted odds of winning and therefore the win/loss ratio

[blazing spire]

And there's also a host of other little escoteria I won't go into....

And that's partly why i like to use an average of averages of players around a similar rating/position...

[eternal arrow]

but I don't think looking back at games is a correct analysis, if 20-20 fights 20-20 that's 50% winrate at the time of playing, then say it adjusts to 22.5-22.5 for winners and 17.5-17.5 for losers, so 45 vs 35 when you look back on the players, clearly the winners are picking games because they have a 70% chance to win according to the 45 vs 35 number, or whatever the value ends up being

the trend you do end up finding is if players are improving their rating in a short period of time, if you're only looking at say rolling 50, relative to how much other players are improving

or you can find that guy with a bad day who went on a loss streak

people who are further from the mean likely played more games to get that far, so they spent a longer % of games being underrated and thus more of their past games will look as if they should've won all along, because someone rated 50 os "now" would've been 30 os in that prior game

but someone who plays games to arrive at their "proper" rating, then plays another 2k on top of that at 50% winrate, will look like they've had ~50% winrate forever

[blazing spire]

Ah, i don't know if i've made clear, but I've generated fully dynamic ratings and stats. So trying to put that into english...

I see every player's relative rank to all other players, OS movements pre-and-post match, running averages of my own choosing quantified down to the lowest possible unit of time. And winprobs are calculated pre-match...

So to use your example...

when 20-20 fights 20-20 match rating, I generate both internal mu, and internal sigma both pre and post match as well as match ratings. The OS model generates a win-probability based on an internal mu and sigma pre-match.

My running average 50 in this case is the running average of past 50 pre-match win probabilities

[eternal arrow]

pre match is fine, but I still don't think its a meaningful metric especially considering that it's a feature of OS to predict games using that value... so lets say for a set of games anytime I'm playing my team has a 70% winrate, so anyone who happens to be on my team as a pattern (say random chance, or maybe someone specifically dodges all games where they aren't on my team), they will have a set of 70% win probabiity prematch values. But OS accounts for this and they lose 3x the rating they would otherwise gain for example

any time you retroactively look at past win probabilities, the point is that the new game played is providing new information for the rating system that it didn't have before

[tribal stag]

So you mean to tell me that there is no way I could tell this game was going to be a win based on probability and chevrons compared to OS?https://www.beyondallreason.info/replays?gameId=876fc5656af05709119ca358699032cf Or the fact I saw myself get switched to the 1 chevron team and I decided I was out for the next round and knew what was coming.

[blazing spire]

Just to clarify: your 70% win rate is different from your observed win probability in those matches.

So win rate is how much you actually win, win probability is what the model thought would happen before the match.

[eternal arrow]

yes im 100% aware

[blazing spire]

Just checking that we're using the same language...

[eternal arrow]

yeah I think we've both been on roughly the same page here

[blazing spire]

So generally, we'd expect in team games for OS to balance somewhere around 50% win prob...

[eternal arrow]

i haven't been referencing actual win rate anywhere in this discussion

[blazing spire]

(technically it ends up at like 51% for reasons but lets just stick with 50%)

[eternal arrow]

except this isn't true in reality, not that it matters. there's no implied benefit to only playing games where you have 55% winrate as opposed to 50% relative to the end result of your OS

[blazing spire]

hmm, i think that's the area of disagreement.

[eternal arrow]

maybe in some sense, playing games where you have a higher winrate means you have a higher likelihood to hit a winrate streak that sends you over your equilibrium rating, but that is far from sustainable in the bigger picture

[blazing spire]

Well actually, let me say, there might be players who try to only pick easy matches because of some psychological belief...

but in that you're right, their OS will dynamically adjust.

But then in the idea of what I'll call "OS arbitrage", is the belief that there's actually a genuine benefit in trying to dodge matches...

[eternal arrow]

consider the basic dodging criteria of: if my team has 3 players who are 16.67, and the enemy team has 0 players that are 16.67, I may recognize this is a shitty gameplay experience for me and not play it. I don't think that dodging this one instance of "clearly unfavorable" matchup has any effect on the player's end result rating, assuming they've played enough games (lifetime) to have arrived at their equilibrium rating

if I play it, and I lose rating as I "predicted", then I'll just win a future game and get back to my equilibrium rating

[blazing spire]

presumably because the arguement is that the internal OS model/game balancer is miscalibrated

[eternal arrow]

if I dodge it, there's no new information added to the system

[eternal arrow]

honestly I think these rating models perform much better than people think

usually (and especially at higher levels), you start to develop a mental image of what, say, a "48" player plays like compared to a "42", maybe even with more granularity to be able to differentiate between your vision of a "45" from a "48" or even better. I think that's a sign that the rating system is quite accurate that you can develop that kind of intuition without much analysis

[blazing spire]

oh yeah, i actually agree there, empirically in aggregate they actually do much better than almost everyone can achieve thewmselves

[tribal stag]

Why not just hide OS pre match then and Chevrons... Put in a 5 minute match ban on leavers. Let them show up in match.

[eternal arrow]

with extremely high player counts, you'll occasionally get the random who is misaligned with the "correct" scale, but that's just a numbers chance game

[tribal stag]

The OS system is so perfect, why not act it and hide the back end until you are in match?

[blazing spire]

but, i will say for instance that we do know that OS is definitly miscalibrated on the begginner players.

[eternal arrow]

for sure

but it is extremely unlikely that the skill levels of the playerbase follows a normal distribution...

[blazing spire]

and a player who ruthlessly avoids playing with those members on their team will, all other things being equal, reach a long run equilibrium a bit higher, in terms of reported OS, than a player who doesn't.

[eternal arrow]

maybe in a scenario where it is extremely commonplace to have beginner level players in every game, and in mismatched values

[blazing spire]

now i just choose that one because I think it's an easy idea of taking advantage of miscalibration to raise OS.

[eternal arrow]

there can always be edge cases, someone who plays against their friend 200k times and wins every match... might not align with the rest of the community

[blazing spire]

So, implicit in the idea of match-dodging (aside from just the begginner issue) is the idea that there's something else exploitable in the knowledge of teams that can be used to game OS...

[eternal arrow]

according to the analysis from doushibag in the other thread, it seems that new players have ~38% winrate on their first game, which definitely isn't 50

but it's expecting much of OS to be aware of people's skill levels before a single game is played

[blazing spire]

And one of those ideas is "well, mu minus sigma doesn't really have any justification for calculating match win probabilities from a pure stats theory perspective, but we use it to balance because it generally predicts matches a little bit better because it penalises begginner players a little bit more than the internal model thinks it should"

[eternal arrow]

that one I'm interested in. I don't think it generates a positive overall effect on the system, and creates some flaws that can be exploited

[blazing spire]

damn it, i'm going to have to dive out of this conversation early, which is a pitty because I'm enjoying it and i think it's going somewhere...

[tribal stag]

I think if the system spreads the player base a bit more based on Chevron, AKA greater uncertainty range it could help with the broken matches I keep seeing in lower tiers.

[eternal arrow]

the dodging new players thing--I don't think this has a large or concerning effect on the overall curve, "exploiting" the information that a new player has a 38% chance to win their first game (they are likely overrated) means primarily playing games where you don't have new players on your team, while the enemy team does--this likely implies lower rated games, and low quality games, not something that will have a big impact on leaderboard generally speaking

[tribal stag]

But it hurts playerbase and drives players away from pvp if they lose 8 matches in a row as OS tries to correct.

It frustrates other players as well.

[eternal arrow]

in a way specifically curating this kind of game set is not unlike playing in a bubble where you only play against friends you know you can beat 100% of the time and your rating slowly goes up, just a tiny bit

[stone void]

nothing to do with match dodging tho

[eternal arrow]

yea, new players having a <50% winrate has nothing to do with dodging, the playerbase skill curve however you'd like to define it is likely not a normally distributed one

[tribal stag]

Except I have seen quite a few higher OS players dodge once team comp was done and they did not like the spread of 1-3 chevron players.

[eternal arrow]

does that affect how likely the new player is to win? if the initial rating overrates them, it doesn't matter who they're fighting if we're assuming OS to be generally accurate for not-new players?

[tribal stag]

The new player still has that 38% chance to lose, unless the system stacks 3 1 chevron players on the same team.

Then I am willing to wager the odds of winning decrease. The uncertainty is too high.

Erg 38% chance to win, 62% chance to lose.

[eternal arrow]

in theory that means they all get adjusted downward together and we likely skip a few extra games of calibration that we would've needed if they got split up and half of them gained rating

not sure what your point is here--it seems a known issue that new players are likely not worth their starting rating. but do you have a recommendation? and is this related to the thread topic of dodging at all?

[icy sky]

dodging = good

And I’m not even being facetious, I genuinely believe it

[signal chasm]

The punishment for dodging before the match begins is that you don't get to play the game. There's no need for extra punishment.

[tribal stag]

New players cause other players to dodge when the matches are unbalanced.;()

[signal chasm]

There was a suggestion somewhere here about being able to use min/max chevrons for lobbies. I would be in favour of it despite being a newbie. But I think some people might be against that feature.

[blazing spire]

Just jumping in quickly, giving it some thought I'm not convinced it's a huge issue on the top end (with top end balance discrepancies potentially explained by other phenomenon as well that need further investigation), but the new-player-dodge is an issue if not just for the marginally hostile experience for new players.

[signal chasm]

The new players won't know why people are dodging. They just see someone go to the spec list. Honestly, if not for this thread, I had no idea people did this.

[icy sky]

Off topic but I’ve always wondered whether the balance algorithm would have higher predictive power if, instead of minimizing OS deviation by more or less first sorting the two highest OS players and then pairing remaining players (greedy algorithm), it instead took a divide and conquer approach, ie where a 1v1 win probability is assigned for each pair of players in the lobby and balancing is done around equalizing the cumulative win probabilities

[tribal stag]

It might actually be more accurate. But I could see limitations on partied players vs un-partied players getting an unfair advantage. There is also the problem of what to do with the new players that are being overvalued. There are stupid mistakes they will make. Cough Small sliver of defense in the back field where a nuke could get in.... cough

Then again, non new players make the same mistake at 100 hours...

[icy sky]

makes sense. maybe one low-hanging fruit approach to discourage new player dodging would be for “newbie” and “all welcome” lobbies to have a message auto-generated which spells out the expectation around treating new players with kindness and respect

[tribal stag]

Or they create a special lobby for new players where other players do not suffer a loss of OS when coaching new players and playing with them, but they also get reduced ratings gains if their discrepancy is under a certain point?

This allows new players to get stabilized and coached while the coachers are not suffering losses to their OS. Players with less than a certain amount of play time still get to see major corrections to their OS.

I am doing a horrible job explaining this.

Newbie player A loses and gets a large correction of OS. Player B with 900 hours and 22 OS does not suffer any OS loss by having player A on his team. If they win player A gets 1/10th OS gains, player B gains 1/10th what they would normally gain.

The act of losing cannot be changed, but they at least wont be penalized for being decent human beings.

There does need to be a greater differentiation between Chevrons for PVE vs Chevrons for PVP.

[eternal arrow]

Right now the only conclusion I've arrived at is that injecting 1 smurf for every 4 brand new players would at least solve the new player dodge thing 😄

[tribal stag]

I would love to understand the flawed logic that led to that conclusion....

IT would need to be 4 smurfs for every 1 new player...

[eternal arrow]

Well, if new players win 38% of their first games, and smurfs win 100%, then having 1 smurf for every 4 new players would make the odds of a "new" player winning roughly 50% 😛

[signal chasm]

People dodge if there's too many one chevs on their team because they don't want to lose. But if there's a 20% chance that one chev is Raghna in disguise they might want to stick around for the free win.

[eternal arrow]

unfortunately as funny as it would be, still doesn't help the new player experience if people lose their first games when genuinely new 😛

[tribal stag]

What about the 2nd, 3rd 4th and 5th games.;) Gotta work on the new player experience when breaking them into pvp, but gotta work on fixing the players dodging new players at the same time.

[signal chasm]

If an experienced player doesn't want to play with 1 chevs then he can just avoid the noob lobbies. The problem is that it's a little awkard to set up. Best you can do is set up a min 18OS lobby. But there are people who are experienced that have less OS than that. That's why I think a min/max chevron lobby setting is useful.

[tribal stag]

Jauggy there are a lot of things that higher OS players can teach to newer players. I currently watch my matches to learn what the other players did or did not do. I am Chevroned at 100-250H, most of that was against Barbarian AI's. I started pvp at 3 chevrons and was not ready. A steep learning curve from barbarian AI's vs human players.

[finite loom]

if the people that don't want to play don't play, then you are left with people who do want to play, and having a game filled with people who do want to play is good

[tribal stag]

Except when they wait until just as the match is in the vote to start to dodge out or they dodge as the match starts and the entire match has to be restarted. I have even see matches fall apart after 3-4 different people dodged out in a row.

[icy sky]

If you leave when the match starts I’d call that leaving, not dodging.

[tribal stag]

Semantics of the situation, it is the same thing and leads to delays in playing.

[icy sky]

Well one is bannable and the other is not so it’s not limited to just semantics

[vocal lion]

You shall play, or you shall visit the gulag. The choice is yours. Releases safety, cocks trigger

[tribal stag]

And banning hurts the entire community by reducing playerbase. That and with the crashes and other issues it is hard to distinguish a crash from a leaving.

[vocal lion]

Imo the proper recourse is all social and organic--some people get a bad reputation for dodging and get made fun of for it. Moderators don't have to do anything, and those habitually dodging get denied the social prestige they are trying to create through OS cheesing.

[tender kindle]

Or fix the autobalancer stacking a team with newbies so people won't feel the need to dodge teams stacked with newbies...

[white wharf]

Burden of balance isn't really on the players, the OS system does that, its just not perfect

Especially with newer players (1 chevs) who could be veterans of other TA-likes or RTS or completely new to RTS

[vocal lion]

That bit seems easier said than done. How would you translate time played into a skill estimate for balancing?

You can't just lower their OS value; starting OS is just an arbitrary number for players to go above or below. Changing starting OS means you'd also have to change the OS of all other players relative to your new starting OS for skill estimates to be valid--and then it's just the same outcome again.

[white wharf]

The real dodging problem I think is the amount of games that need to be stopped before they begin due to leavers

I think if the game launches and you dont show up there should be some sort of scaling timer, regardless of technical difficulty

Copy the dodge system from LoL, but apply it to matches that haven't started (placement phase doesn't end)

As in if the game has to be stopped because you weren't there, you get a 5 min timer preventing you from entering the join-queue of any multiplayer lobby, scaling to 15 then 30 then 60 minutes
(scaling resetting 8 hours after the first penalty)

If you dont like the team you have, dont ready up, if the other people in the lobby dont want to make changes, find a new lobby with a balance you like

[tender kindle]

You'd do it by increasing initial uncertainty, not changing the OS midpoint. We talked about this a bit in the new season thread.

Uncertainty already subtracts from OS. It just isn't large enough to truly account for most brand-new players being utterly useless.

[vocal lion]

40 os after lucky first three games? 😛

[tender kindle]

And then right back down into the gutter when they're the red player and lose. 😛

[vocal lion]

Would also need more SmurfGuard.

[tender kindle]

I dunno if you'd want to just change how uncertainty affects the autobalancer, how much uncertainty affects OS, or upend the whole system by massively increasing uncertainty. Maybe some combination. But the core of the issue is the system is ranking newbies too highly.

[vocal lion]

Creates more incentives for smurfs, and smurfs dodging games.

[tender kindle]

Maybe, but I bet newbies getting kicked from lobbies because the autobalancer is stacking them probably isn't a great first impression either.

[vocal lion]

Agreed, it's a bit awkward.

[signal chasm]

I created a noob lobby recently and made it unranked. That’s a simple solution. You don’t care as much if you have one chevs on your team.

[tribal stag]

But the one chevs need a drop in OS to balance the game later. The answer is to make newplayer lobbies the default and not penalize or reduce penalties on players willing to play on low OS teams and just have fun, but still give the new players the OS correction they should get. Say if they have less than a certain number of PVP matches they still get OS correction in unranked lobbies.

[white wharf]

Even then OS is just a balancing number, it feels good to go up in skill rating but there is no reward for having a high number other than reputation

[signal chasm]

They gain chevs and experience when playing unranked. Yes they will have to play ranked eventually. But it’s quite quick to move to two chevs. Effectively they start playing ranked when not total noobs. The purpose of this thread was about match dodging not about perfect balance.

[tribal stag]

Then it needs a better indicator that it is an unraked match before people join the lobby.

[signal chasm]

Yeah that would be useful. I think a filter would work because at the moment the lobbies just keep moving when I hover over it.

[vocal lion]

It's related because unbalance is a major motivator for dodging.

But agreed that there doesn't seem to be a good way to create perfect balance.

[wheat lark]

High OS people can usually force lower-OS people off the start location they want even if they get there later. There's a lot of bullshit that high OS can let people get away with.

[white wharf]

there's only peer pressure, once you lock a spot its yours until you unlock and click off

I mean, they could get the lobby together to kick you, but OS doesn't affect the vote

[wheat lark]

Today I learned votekicks are a form of peer pressure 😛

OS absolutely affects the vote. Not mechanically, but people always side with the higher OS.

[fallow zinc]

Another method of dodging is to block people and then get the lobby to boss you and all the people you don't want to play with get kicked out

And people voting for the boss don't even realize what they're doing

Got blocked out of another lobby for that reason and don't even know what I'm doing to get on people's block list

If people just want to leave and dodge that doesn't bother me at all, but people who get others kicked out to 'dodge' but still play are more nefarious I think

[signal chasm]

I've been kicked out of a lobby on entry, but I thought it's because the people who created it wanted a private server and didn't know how to do it properly.

[signal chasm]

If you leave after the match starts, does that count as a resign?

[candid saffron]

You will gain 0 if your team wins and lose full amount if they lose

However if you leave before like 90% of the game is done or something like that the normal os gain / loss applies

[stable harbor]

Did not read the whole thread (too much sorry), but imo a good solution to this is a matchmaking system like in other RTS, where you get matched automatically and where you can veto a few maps.

This also leads to the result that your OS can not be totally biased by only playing strait or maps with eco spots.

This matchmaking system could also have two queues - one for solo gamers and one mixed queue with premades. Though the letter one could be difficult to fill up if there would be a strict solo queue (see similar issues in DotA).

Also in general there should be either a limit of premade people (especially if the other team has none -> they should get splitted among both teams) and/or there should be a little "handicap" for premade teams (regarded as slightly higher OS for instance) since they have the advantage of having certain strategies together, communicating in discord etc.

At the moment the elo system is definetely not working properly because of all the dodging, premade teams etc. But a matchmaking system might be only feasible with a huger playerbase (but you should give it already a try before release).

[icy sky]

thanks! didn't read your message (cbf sorry), but it's cool that you like stuff

[cerulean burrow]

Not super on topic for this, but kind of related to the matchmaking suggestion:

Something my group has been liking playing BAR os the fact that we do get to see repeat names and there's some chatting and banter on games.

We've been burnt out a bit on the anti-socialness of other games that are matchmaking only. You just get thrown into a match with a random person, but you never really notice who it is or care that the opponent is actually a human on the other end.

I think BAR could do something really unique with a hybrid approach.

[stable harbor]

you expect me to read 370+ messages? I just mentioned it in case this was already debated. We have here enough people in discord, who do not even read the last 10+ messages of debates and just keep posting their bullshit / derailing threads.

[icy sky]

kinda - matchmaking is not a new suggestion, maybe a ctrl + f or something

[ashen wasp]

Would be a an interesting game mode. Hide OS in game and lobby. Everyone goes into a pot and the game starts. Balancing would still happen but you would get the team you get. It would be an interesting thought experiment to see how players play when they don't know the OS of anyone in the match 😂

[granite rock]

The day I don't get teamed with 3 completely new players is the day I will agree with this and I don't care about os I just want a fun game and having to play with 3 completely noobs takes my enjoyment away

[candid saffron]

don’t play all welcome rhen

if you are good enough to play min rating then play it, if not you can’t complain about other people being noobs

[granite rock]

Again I want enjoyment and don't want to sweat it out in the min lobbies so my solution was make a mid level lobby but no luck getting it to take off

[candid saffron]

lets pls get a way to stop dodging? dont show the teams before the game starts just show who is in lobby feels like an easy fix to me

u ready up and then find out who is on whos team once the game starts very op lets do it

[stark lotus]

might lead to leavers after the game starts which is worse for playability

specially parties

[candid saffron]

yup just make leaving at any point during the game guaranteed OS loss

[stark lotus]

i mean they might just accept the os loss and leave anyways

[candid saffron]

well people leave / dodge to avoid the OS loss

and for the parties u could show the parties and whether they will be put together or not ig

[stark lotus]

they can do that thing where dodging lowers your os by a penalty which is seperate from os. like -0.2, but you regain 1/10th of that for every match you finish

[candid saffron]

idk why not just remove it entirely