The "Balance" Problem (Survey Feedback)

Discuss the campaign and all things BF.

Moderator: Executive

cancel_man
Supporting Member
Supporting Member
Posts: 274
Joined: Fri May 02, 2014 11:58 pm
Location: Portland, OR

Re: The "Balance" Problem (Survey Feedback)

Post by cancel_man »

Divine-Sneaker wrote:Unless you can find some way of sifting through armor/air players, then it seems like it might give some skewed results for those in particular. People who exclusively play infantry will generally have a pretty different set of stats than the players in specialized roles.
It's true, but members who play mostly non-infantry tend to be noted/known as such. Plus stats tend to even everything out over time anyway- especially since nearly 50% of our attacks are Infantry Only. Someone who FC's a lot may show a lower k/d or score, but FC'ing 1 in 6 rounds each battle day wouldn't affect that player's overall stats much after 4 campaigns. You can also use basic handicaps & multipliers to even out stats (like an automatic 10K points per round played as commander) if we see a distinct difference for a subset of players.

And again, numbers don't necessarily impact the draft- players will still be drafted due to reputation, relationships and usefulness beyond their stats. The numbers just help show whether teams are a close match based on battle effectiveness and attendance.
Kilo wrote:There are plugins that can do that for you instead of manually taking screenshots.
Help finding one- especially one that can be used with our private/unranked matches- would be appreciated. Since they don't get tracked in battlelog, would it need to be server side? Even without, it's only 64 players at most per round we're tracking.
Kilo
Supporting Member
Supporting Member
Posts: 418
Joined: Tue Oct 30, 2012 7:26 pm
Location: California

Re: The "Balance" Problem (Survey Feedback)

Post by Kilo »

I know both 21CW and DSG use some form of stat tracking.

Here's 21CW's one: https://stats.21cwforums.com/index.php?ServerID=1

They forgot to turn it on until the last two rounds though, but you get the idea.

Here's DSG's: http://www.gametracker.com/server_info/ ... p_players/

A lot more limited, but I guess it shows who plays the most.
Image
Battlefield Tournament History
BF3C4: STAR Infantry Private 2Image
BF3C5: LOD Infantry Private
BF3C6: UNSC Infantry Private First Class
BF4C1: USSR Infantry Conscript
BF4C2: GoCI Infantry Private First Class
BF4C3: LN7 Infantry Private
BF4C4: JANUS Armor Private
BF4C5: KTSS SAS Captain
BF4C6: SAD Infantry CorporalImage
BF4C7: GC Infantry Captain
User avatar
Divine-Sneaker
Posts: 226
Joined: Tue Oct 23, 2012 6:26 pm

Re: The "Balance" Problem (Survey Feedback)

Post by Divine-Sneaker »

Spoiler: show
cancel_man wrote:
Divine-Sneaker wrote:Unless you can find some way of sifting through armor/air players, then it seems like it might give some skewed results for those in particular. People who exclusively play infantry will generally have a pretty different set of stats than the players in specialized roles.
It's true, but members who play mostly non-infantry tend to be noted/known as such. Plus stats tend to even everything out over time anyway- especially since nearly 50% of our attacks are Infantry Only. Someone who FC's a lot may show a lower k/d or score, but FC'ing 1 in 6 rounds each battle day wouldn't affect that player's overall stats much after 4 campaigns. You can also use basic handicaps & multipliers to even out stats (like an automatic 10K points per round played as commander) if we see a distinct difference for a subset of players.

And again, numbers don't necessarily impact the draft- players will still be drafted due to reputation, relationships and usefulness beyond their stats. The numbers just help show whether teams are a close match based on battle effectiveness and attendance.
Kilo wrote:There are plugins that can do that for you instead of manually taking screenshots.
Help finding one- especially one that can be used with our private/unranked matches- would be appreciated. Since they don't get tracked in battlelog, would it need to be server side? Even without, it's only 64 players at most per round we're tracking.
Not saying it invalidates the concept, but as Kilo pointed out it definitely means there's a large amount of extra work to put in somewhere regarding this. That's simply something that has to be utterly clear from the beginning and something that has to be done.

Not everyone intricately knows every player in the community well enough to know their preferences or their competences. My comment isn't meant to be negative, it's simply a note of discussion to make sure people have in mind what this type of approach actually entails.

Unless people are actually willing to put in that type of effort it's unlikely to work.
"fraking game mechanics"
User avatar
Jokerle
Supporting Member
Supporting Member
Posts: 1986
Joined: Fri Jan 11, 2013 1:13 pm
Location: latest crashsite

Re: The "Balance" Problem (Survey Feedback)

Post by Jokerle »

didnt we tried to measure ''skill'' by stats in another thread? Or at least discussed it?
Wat ne Wuchtbrumme!
Gwynzer
Executive
Executive
Posts: 3098
Joined: Wed Feb 15, 2012 5:13 pm
Location: England :(

Re: The "Balance" Problem (Survey Feedback)

Post by Gwynzer »

That was in here earlier, I think.

Measuring via stats has always (usually) been part of the draft. Most draft documents link to the players battlelog link and, especially for "unknown" players, their stats are a good way of engaging their abilities.


Guy's, I just want to say thank you very much for staying on topic, civil, and engaged. It's really awesome to see us having a proper discussion as a community, without it devolving into trolling, flaming, and pointless bickering :)
Image
StarfisherEcho
Executive
Executive
Posts: 3037
Joined: Mon Feb 13, 2012 2:24 pm

Re: The "Balance" Problem (Survey Feedback)

Post by StarfisherEcho »

I second Gwynzer's motion! Great discussion guys, keep it up.

I hate to say it, but all of this has happened before (cue Cylon music). The primary determining factor in army balance is attendance. You can break yourself trying to come up with a perfectly balanced army, and it might be balanced for about five seconds before attrition sets in. Then people drift away, quit, get busy, have kids ( :silent: ) or any of the million things that happen, and your carefully crafted balance is completely ruined.

The best way to visualize it is probably an actual army. You go out and fight and people get wounded or killed. At GC, we basically don't recruit mid-campaign (aside from occasional reddit posts), which is almost like a real army refusing to replace losses until a war is over. We shouldn't be surprised when that fails. Hell, we should be surprised when it works.

I think we really need to give Kilo's ideas a shot. We draft armies, make some effort to make sure we're starting off balanced, but after that the gloves come off. Recruit, recruit, recruit.
ImageImageImage
cancel_man
Supporting Member
Supporting Member
Posts: 274
Joined: Fri May 02, 2014 11:58 pm
Location: Portland, OR

Re: The "Balance" Problem (Survey Feedback)

Post by cancel_man »

cancel_man wrote:The Skill Quotient- or more generally player stats- is fairly simple: Track score, K/D and attendance for each player over their GC career (don't use Battlelog, it's crap for us).
StarfisherEcho wrote:I hate to say it, but all of this has happened before (cue Cylon music). The primary determining factor in army balance is attendance.
So where is the attendance tracking? We've both stated it's important and you say it has been done before, but as far as I know attendance is kept loosely at best by an army and forgotten as soon as a campaign is over. What I'm proposing is tracking a member's attendance over their GC career- a number that carries over from campaign to campaign. It's a really simple statistic that's hard to dispute or deny- either you were there or you weren't. Right now we draft on gut instinct/vague memory of a player's skill but it's easy to forget that the player might have a <25% attendance despite their skill.

It's easy to say it's been done before, but since I'm relatively new I'd like to know more about how it was done, the impact it had and why it isn't being done anymore.
User avatar
Jokerle
Supporting Member
Supporting Member
Posts: 1986
Joined: Fri Jan 11, 2013 1:13 pm
Location: latest crashsite

Re: The "Balance" Problem (Survey Feedback)

Post by Jokerle »

cancel_man wrote:
It's easy to say it's been done before, but since I'm relatively new I'd like to know more about how it was done, the impact it had and why it isn't being done anymore.
there is a web-based tool that can extract a player list of a battleday, which is usually used for tracking promotions due to attenting X BDs and so on. In the ABC profiles you should see (sometimes) medals given out for attendance as well.

i dont remember a proper overview with numbers at the end of the campaign though

I only took actively part in one draft (BF4C1) and attendance was ''measured/tracked'' by adding comments to the draft document (e.g. ''didnt show up most BD last campaign'', ''known low attendance, plays levelbf at least have a BD'', and similar). I.e. all based on the memory of the officer staff.
Attendance is more important for the ''1st pick'' players that (assumingly) have a notable impact in the game (good communication, gun skills, etc.)
Wat ne Wuchtbrumme!
StarfisherEcho
Executive
Executive
Posts: 3037
Joined: Mon Feb 13, 2012 2:24 pm

Re: The "Balance" Problem (Survey Feedback)

Post by StarfisherEcho »

cancel_man:
Attendance is essentially a random number in your skill formula. Each campaign it changes dramatically for people. Even the player themselves isn't a good source for this - we could write a script to compare sign ups/attendance versus what people said in their draft card, and I'm willing to bet that most people would be wrong one way or the other. And oftentimes balance problems arise with new players who no one can quantify; I recall breaking down a new player skill draft perfectly evenly, then watching as a random selection of people didn't show up. Cue balance bitching.

The problem isn't that we can't look back - we mostly can, via archived forums and asking around - it's that no one can predict the future. We could craft two perfectly equal paper armies, past attendance and all (we did this once), but by the time you hit Scrim 1 some people have already disappeared, and somebody has to work that day, and so on and so forth and you get a stomp. People change, and as a result, any process that relies on getting things right once at the start is guaranteed to fail.

Hell, you could argue that the goal of an army is to imbalance a campaign so that they win - they train and plan and get better until they're the better army (imbalanced campaign!!) and then they win.

Kilo's idea is the only one that provides a robust mechanism for armies to fix this themselves. Down on people? Recruit! Down on skill players? Recruit! The community benefits from those who stick around afterwards and armies have a way to try to fix themselves.
ImageImageImage
User avatar
Divine-Sneaker
Posts: 226
Joined: Tue Oct 23, 2012 6:26 pm

Re: The "Balance" Problem (Survey Feedback)

Post by Divine-Sneaker »

I know I'm personally a huge failure at using the battleday signups and that's always been the case. Really, I just don't fraking know if and if so, when i'll be available ahead of time.

The best advice I can give on that is that captains etc. get active on getting to know people. I was actively texting ID when he was airforce captain to say when I could be there.


Just like Kilo has mentioned how recruiting has a lot to do with making it personal, getting people to invest into a campaign/army requires similar things.
"fraking game mechanics"
User avatar
Necromancer
Supporting Member
Supporting Member
Posts: 3315
Joined: Sat Jul 28, 2012 4:20 pm

Re: The "Balance" Problem (Survey Feedback)

Post by Necromancer »

Balance

TL;DR - Skill Pre-draft, or in the BF4 case : Complete balance draft is more likely to consistently insure a better balance.

Competitive football style draft doesn't. Each army tries to win the draft, and one army often does. Props to that army officer corps for gathering more/better information and evaluating it better, but campaign is pretty much over when that happens. What should we do next? announce that army as the winner of the draft and as such the winner of the campaign, scrap the campaign and elect new generals?
The only good outcome (community wise) is when both sides "loose" and end up in a "draw". But as we have seen it rarely happens.

The pre-draft / Balance draft get its balance properties from the principle that says more people (with more time to think it over) are less likely to be wrong. Allowing more people to get more time to think about it reduces both bad decisions and opportunism .

Skill pre-draft happened at BF3:C5 and i think it worked pretty well.
It has few cons to it:
1) More time consuming
2) Giving out "free intel/info" to the opponent.
3) Officers. On this point later.

I think point 2 is the biggest problem. From a bystander point balanced draft is better for both sides, however in practice there is a difference in how the teams approach the player draft. Some armies spent a lot of time gathering information and rating players while the other side doesn't. Then during the balance discussions each side needs to explain why it is balanced/unbalanced, and that requires revealing that hard-gathered information to the opponent "for free". The more time and effort you put into gathering that info / the more valuable info one posses, the more unfair it seems to share it.

While its unfair to the officers, that information sharing is exactly what leads to a more balanced draft, and its better for the community. There is no point in winning the draft and thus 'winning the campaign before it even started'.

BF3 had a higher learning curve and more skilled players. So ~30+ people were sorted during the skill draft, which is ~50%.
In BF4 there is smaller skill gap and most of the players are about the same level. So it seems a complete balance draft is the only option.

Con #3 of this system is that it doesn't balance out the officer corps.
The officers are about 50% responsible for the outcome of the campaign. Since there are 10 officers and ~30+ grunts, the impact of each officer is significantly bigger then of grunt. If an army has 1-2 bad officers (i use simplified terms to make this article a bit shorter. "bad" doesn't mean actual negative influence, rather 'less then expected from an officer', for example if the officer is mostly inactive), So if an army has 1-2 bad officers it may be overlooked by the rest of the army, but 3+ unbalanced officers and the rest of the army will simply not trust their officers as a whole to lead them to victory even if the "player draft" is equal.
That said, no other proposed system does balance the officers either.

Only when both armies agree the draft is balanced (or mutually unbalanced) it is made official.

Additional stuff that can be applied to further increase balance:
- Swap:
If one side thinks the draft is balanced but the other doesn't agree, or the draft is being made unbalanced by one or both armies, then an Army-swap can be done. This way the side claiming the teams are not balanced ends up with the "stronger" draft, and the other side ends up with an "equal" draft and everyone suppose to agree. The possibility to force side-swap is supposed to encourage both sides to make the draft more balanced, further eliminating opportunism as it might back-fire if the armies do get swapped.
I am aware the way this draft works and that proposed army swap goes against the officers wishes to have specific players they themselves choose ("control over the army" etc..), but our past shows that when officers get to pick freely whoever they want it doesn't end well for the armies (both) and their officers either.

- Community vote:
Since the more people agree the draft is balanced, the more likely it is to be balanced, its might be a good idea to present the draft for a community vote. If 70% of the community agree that it is balanced campaign may start, otherwise the Officers go back to the drawing board.
The 70% value is not an arbitrary number, it is a result of a few assumptions that can be discussed if GC chooses to peruse thing type of draft.

In my opinion the balance % this system offers out-wights the restrictions it imposes on the officers to choose whoever they want as that doesn't work anyway. Plus GC is all about meeting new people right?

Attendance
Spoiler: show
I don't think that's the biggest problem.
Mostly attendance tends to be the same. People that show up sometimes are still only going to show sometimes regardless them signing up for 6 hours every battle.
One guy that doesn't show up doesn't change the balance much, especially in BF4.
There aren't that big fluctuations and the teams tend to be unbalanced even before the BFI, perhaps one side miscalculates the attendance of its players, but then it doesn't fall under the "unpredictable attendance", though i guess knowing when it is predictable and when it isn't isn't easy and requires very good memory, or an attendance tracker. Knowing the people might also help (for example they might tell you they go for a 3 week vocation, or play at LevelBF first).
Officers attendance is more important and does tend to have an impact. It also happens more frequently which is a bit ironic since they are supposed to be more responsible. I don't know why armies tend to leave that "ghost officer" as it is instead of adding another officer though. Lack of interest? lack of experience? or lack of candidates?
Army Recruitment
Spoiler: show
Its a good way for an army to fix itself. However i'm worried that might change the core of GC, as armies will turn into nothing more then hiring mercenaries.
Beating the other side won't mean doing better strategies and improving teamwork, rather more aggressive recruitment as it will probably tilt the balance much more. Winning simply by getting better players is less rewarding IMO. If officers get burnt out doing map strats and FCing, imagine how much more effort itt'll take to recruit day and night.
Whats the difference between that and allowing skilled groups joining a particular army?
Kinda makes the whole tournament meaningless if you get outside help.
turns into nothing more then a recruitment race.

I don't reject the idea, just offering a different angle on the matter. We can always try and see how it works out.
Image
-“Regret your helplessness…and feel despair.”
Achievement Unlocked: Battlefield 4 Uninstalled!!
User avatar
Jokerle
Supporting Member
Supporting Member
Posts: 1986
Joined: Fri Jan 11, 2013 1:13 pm
Location: latest crashsite

Re: The "Balance" Problem (Survey Feedback)

Post by Jokerle »

Kilo's idea is the only one that provides a robust mechanism for armies to fix this themselves. Down on people? Recruit! Down on skill players? Recruit! The community benefits from those who stick around afterwards and armies have a way to try to fix themselves.
On the other hand we always prided ourselves with wanting players that fit into the GC style of fairplay and friendliness.
I am not saying these are especially difficult to find, but if 30-40% of an army consists of new-blood, it is hard to instill the GC spirit. It is already difficult to enforce the ''no all-chat'' rule with new people and bug/glitch-abusing will happen more often as well. This is just natural, one needs to adapt to the community style and find a balance between being competitive and winning-at-all-cost. (playing and teaching GC-style is also an officer's duty btw)

I just wanted to note that. In principle the idea of recruiting brought up by Kilo should work well enough for us and I am generally a fan of it.

But Necro's point about recruiting high skill players as a meta-contest should be kept in mind as well. However there should be experience with this problem in 21CW! (Kilo?) It might be a theoretical problem, but function ''ok'' in practice.
Wat ne Wuchtbrumme!
StarfisherEcho
Executive
Executive
Posts: 3037
Joined: Mon Feb 13, 2012 2:24 pm

Re: The "Balance" Problem (Survey Feedback)

Post by StarfisherEcho »

Yep, that's the tradeoff. Recruitment becomes an important part of winning the campaign. I'd like to give it a shot, but you guys are right that it would be a major shift in how things work. We've deliberately avoided it until now, on the assumption that it would disrupt our carefully crafted army balance. We also haven't had a "Balanced" campaign in two years :?

Necro: I agree that one side winning the draft is a problem, and I even argued along eerily similar lines in the BF3 days. But then we utterly failed to fix it with pre-drafting :/

You cite BF3:C5 as a good pre-draft, but that was an imbalanced campaign (6 weeks, winning army at 70% win rate, first battleday was a complete shutout) with major air balance drama. The next attempt, C6, was a complete pre-draft... and while it lasted a little longer thanks to a random good battleday for the losers, it was equally one-sided. The backlash lead to us returning to a no-pre-draft.

As an exercise, I've gone through the BF3 archives and MrBlue's stats and compiled a table with the win/loss percentages plus round won differential per week, along with what I recall as the balance measures taken. From my perspective, the best way to measure balance is that round won/loss differential - that tells you how close each battleday was, on average. The closer to 0 that gets, the more balanced the campaign.

Important to note that in my time here there has never been a losing army that didn't see things as imbalanced.

BF3
C2: No data, first reddit rush
C3: 71%:29% 5 weeks , +4,6 rounds/week: HCs preselected, draft for officers and players
C4: 56%:44% 14 weeks, +1.4 rounds/week: generals pre-draft officers and skill pre-draft, normal regular draft
C5: 69%:31% 6 weeks , +4.6 rounds/week: generals pre-draft officers and skill pre-draft, normal regular draft
C6: 73%:27% 8 weeks , +5.1 rounds/week: complete pre-draft, generals, officers and TAs collaborating

BF4
C1: 85%:15% 4 weeks, +6 rounds/week: generals pre-draft officers, regular draft otherwise
C2: 71%:29% 6 weeks, +5 rounds/week: generals pre-draft officers, I think regular draft?
C3: No data yet
C4: No data yet

A final thought. While I sound like a cynical and pessimistic asshole, I'm all for trying new things and improvement. So don't let me get in the way. Keep this conversation going and when we think we've got some solid proposals (Necro's counts as a solid proposal, folks) then we can put things to a vote.
ImageImageImage
User avatar
A Docile Sloth
Executive
Executive
Posts: 2323
Joined: Sun Jun 24, 2012 4:32 pm
Location: Somewhere where you aren't.

Re: The "Balance" Problem (Survey Feedback)

Post by A Docile Sloth »

Which campaign was it where we used some automated thing to do the draft of reddit rushers back in BF3 (I think C5?)? Was mentioned here already I think. If I'm remembering right, it balanced skill pretty well but then attendance was an issue.

We could potentially do the draft entirely automated like this. As said somewhere we'd have to account for attendance somehow (or hope the law of averages works). For people who have been around for a campaign or two, I think the law of averages would probably cover the attendance, just fresh faces it may not. Could run twice, once for vet GC and once for new faces. Once the two Generals/HCs/Officers are unhappy, coin flip for who gets which player set.
Image
User avatar
Necromancer
Supporting Member
Supporting Member
Posts: 3315
Joined: Sat Jul 28, 2012 4:20 pm

Re: The "Balance" Problem (Survey Feedback)

Post by Necromancer »

Sorry, i was referring to BF3:C6.
Pre-drafting doesn't guarantee a balanced campaign. It doesn't guarantee day 1 balance, and not even day 0 balance.
How come? because as i said it doesn't balance the officers. I think the pre-draft made in C6 was balanced, and i don't think the outcome would of been any different if we switched the armies around.
Though i do agree the campaign ended up pretty unbalanced.

Why? well, we had an interesting situation with one HQ placed just one tile away from the opponents HQ, and our side saw it as a major threat so e decided to deal with it as soon as possible. While the first week of a campaign is usually pretty slow and no crazy attacks are done, i think this time we had the most activity preparing to that first battle. We had a thread with ~5-6 pages discussing strats and tactics how to successfully attack and capture the enemy HQ that was placed on a map that up until then was considered THE MOST biased map. I think all this discussion and preparation even created a small hype, and our officers made sure to get everyone, specially the skilled players on that attack.
While the other side.... i don't even know if they considered their HQ might be attacked on the first battle.

So no, a pre-draft doesn't make balanced sides because it doesn't balance out officers. It doesn't guarantee day 1 (battle/scrim 1) balance because its already involves tactics, strategies and teamwork (which can be a result of officer imbalance as well), and it sure doesn't guarantee anything in the long run.

I think it makes a more balanced player draft, but that's all it does.

That said i only know how the draft looked like in C6, and one campaign is not enough to conclude whether or not the results match the theory or not. It could be that theory simply doesn't work in practice, could be it needs to be tweaked a little , or could be just a statistical deviation.

* The Risk system winning condition pretty much forces every campaign to end in a stomp. Demanding control of 70%, not only that but also defending those 70% next battle. If one side can do that, that in itself is an indication of a pretty huge imbalance.
So we can tweak our starting point, and we can discuss how to balance things during the campaign, but the requirement for it to end in a massive imbalance is set in the rules.
That means all campaigns coming to an end under this system will not end balanced. And didn't in the past either. So measuring a draft system by its end result is not a very good idea.

Although RISK and army balance are connected to each other, lets leave the RISK part to another discussion as the army balance is complicated enough without it.

Computerized balance
Statistics works on large numbers. Thats might be one of the reasons our bigger BF3 campaigns were more balanced then the recent ones. I doubt that a program would make a better draft with our current numbers, but i guess that can be proven one way or the other rather easily.
Last edited by Necromancer on Sun Dec 14, 2014 1:10 pm, edited 3 times in total.
Image
-“Regret your helplessness…and feel despair.”
Achievement Unlocked: Battlefield 4 Uninstalled!!
Post Reply