Fix Battlegrounds in three easy steps (that we can argue about until the end of time)
DNA3000
Member, Guardian Guardian › Posts: 19,846 Guardian
Challenge accepted.Kabam Jax said:It is, however, worth noting that there has not been a single suggestion that everyone has unanimously agreed with.
In my opinion, the Battlegrounds game mode has three primary issues that I believe represent the majority of objections for the design of BG. Things like modding and other cheating are excluded from this, as this is a matter of managing the mode, not designing the mode. I call these three issues the three Fs (most people just call them one F): Frustration, Frivolousness, and Fairness. I'm going to tackle all three in one overarching set of suggestions. To keep things reasonably readable, I'm not going to overly justify every single suggestion, but I will say that I have discussed all three to at least some length elsewhere on the forums.
There is a TL;DR at the end, but I still think it is worth reviewing the why and the details of each suggestion in the TL;DR. Details matter.
Frustration
I think there are a lot of players who experience some degree of frustration due to a particular design decision: you can lose tropies. There are reasons why this decision exists, but the simplified version is: the intent was for players in VT to experience something similar to players who are competing in GC experience: eventually you rise to your level of competition and start winning at a 50% rate. The "win one lose one" trophy scoring system is intended to, in a sense, create a situation where a player will eventually stop climbing. This isn't strictly true but the *intent* was for players to climb quickly, winning and not losing trophies, when they were stronger than the competition, and then stop (or at least radically slow) progressing when they reach equal competition.However, in practice this doesn't happen as gracefully. For one thing, BG seasons do not last long enough for competitors to "sort" themselves out. Most players don't reach this "equal competition" point. And players do not just alternate wins and losses. Sometimes they win a few, sometimes they lose a few. But regardless, there's a psychological impact to the way this works that goes beyond the math. Even when this is working as intended, it is psychologically crushing.
No one wants to play a game mode where they win one and lose one, when what they see is they gain a trophy then lose a trophy. Losing progress is awful for most players. Now, if you are one of the top 5% of players, you're going to lose one only every so often. It sucks, but it is not hard to shake it off and move on. When you're winning and losing back and forth often, you will be losing almost as many trophies as you are gaining. You are constantly backsliding, constantly losing progress. There is nothing more disheartening than finally winning two trophies after dozens of matches, only to lose two in a row and be back to square one.
Nothing can completely eliminate this, in fact in a competition you don't ever want to eliminate the pain of losing entirely. However, I believe there's a way to greatly soften the blow. We change how we award trophies. Right now we get one for a win, and we lose one for a loss. My recommendation:
If you win a match two wins to zero, you gain two trophies
If you win a match two wins to one loss, you gain one trophy
If you lose a match one win to two losses, you stay even
If you lose a match zero wins to two losses, you lose one trophy
What does this do? Well, it does four separate things simultaneously:
1. Very strong players will advance faster than less strong players. This rewards competitively stronger players in VT.
2. By virtue of #1 above, this accelerates the sorting of players in VT. When the stronger players promote faster, the less strong players don't have to compete against them in the lower levels. There's fewer losses whose sole reason for happening is to allow the stronger players to climb over the rest to get to GC (or higher VT).
3. This reduces the pain of backsliding. Players who lose will backslide less, because 50% of their losses will not result in loss of trophies.
4. This decreases the sense of futility in matches that appear to be lost. Sometimes random chance will put you up against a player that just seems to have your number, or the draft will hand you a disadvantageous situation. It can be easy to just throw up your hands and give up, and blame the game for screwing you. But with this type of scoring, even in bad situations there's still something to play for. Instead of trying against hope to win the match and avoid losing a trophy, you could at least try to steal one win. Get one win, and at least you don't go backwards. Not only does this give you something to fight for, it also reduces the opportunity to psychologically blame all losses on the match system or random chance. It would be more likely that you had at least some chance to get one win and save the match, putting your fate at least nominally in your hands.
Because this scoring system also increases the speed of advance through VT, VT tiers will almost certainly require more trophies to advance. My tentative recommendation here is to increase Bronze from two to three, Silver and Gold from three to four, Platinum and Diamond from three to five, Diamond 1 from four to six, and Vibranium from five to seven. However, these numbers may need to be adjusted for player advancement rates.
Frivolousness
Many players believe that it doesn't make sense that a player spends time climbing all the way up to GC, then starts again at Bronze 3 the following season and has to do it all over again. The counter-argument is that just because they climbed that high, doesn't mean they should automatically bypass that requirement next season. But in fact, this requirement doesn't just impact the higher tier players. It also hurts everyone else, as the only way for those players to climb back to GC is to step all over the weaker players. Every time they win, another weaker player has to lose, and potentially lose trophies. Top tier players are not winning trophies, they are taking trophies away from lesser players. So the notion that allowing them to start higher is just handing them an advantage is false. It is not just letting them get to GC sooner, it is also letting everyone else keep more trophies in the process.But if we allow players to start higher than Bronze 3, we are actually taking rewards away from them. A big chunk of rewards comes from the advancement through VT itself. If a player were to start at GC directly, for example, they'd be out a ton of rewards (something like 135900 tokens for one). They would need a way to earn those rewards.
I say earn. Some have suggested just handing them those rewards. That won't work for three reasons: first, if they are just handed those rewards, they wouldn't even need to play BG to get them. We'd be rewarding players over and over for something they did a while ago. Second, this would be exploitable. There would be ways for players to participate minimally and still get tons of rewards. And third, VT rewards only exist to promote participation. Kabam will not give participation rewards to players who don't participate, because that's just dumb.
The way we can allow players to start at higher tiers while still earning those rewards is to use the same objective system that is currently used to give VT players access to other participation rewards. We make one objective for each VT track, and if a player starts higher than that track they gain access to those objectives. So if a player starts at Silver 3, they get objectives for Bronze 2 and Bronze 1 (consecutively) that they can fulfill by actually playing BG. These objectives expire at the end of the season, so they can only earn them in the season they start higher.
Where should a player start? In my opinion, we can't start them where they left off. There should be some decay in the system. Otherwise once a player fights their way into GC, they get all the VT rewards forever without even needing to win a match in GC. That's too exploitable, and too tempting for players to try to game the system. A safer option is to start players one full tier lower. So if they end in Bronze or Silver, they start at the beginning again. If they end anywhere in Gold, they start at Silver 3 (the beginning of Silver). If they end in Platinum they start at Gold 3 (the beginning of Gold). Etc. Even GC players will have to start at Vibranium and fight their way back into GC. Also, this decay is per season. If you end at Vibranium you will start the next season in Diamond. But even if you don't play any matches, you will start the following season in Gold (because you "started" the last season in Platinum whether you played any matches or not). Idle players will eventually decay back to the start.
The net result is top tier players don't have to grind all the way through VT again, which not only helps them but it also helps everyone else who no longer has to lose to them (losing trophies in the process). And weaker players will not have to face stronger players, even in the VT, to the same degree they do now, particularly at the start of the season when everyone "resets."
Fairness
The big banana. Probably the most controversial issue in BG. Right now, observations strongly imply that when the devs eliminated deck-based matching they implement some form of roster strength matching instead. In other words, the game "measures" your roster strength and uses that to find "equal" matches. It is like prestige matching, but probably not actually prestige (it is probably something like prestige but calculated over your top 30 champs rather than top 5, but no one knows with absolute certainty). In any event, roster strength matching has basically all the same issues that prestige matching has. But to explain what those are, it is important to take a moment to explain how ELO matching works, and why it is used. In particular, it is used in Alliance War matching, and even in the GC of Battlegrounds itself.ELO matching calculates a rating score for every competitor that is based on their wins and losses. Specifically, whenever you win or lose to another opponent, the game increases your rating if you win and decreases your rating if you lose (same for your opponent). Moreover, the stronger your opponent the larger the change. If you beat an opponent with higher rating your own rating goes up more than if you beat someone with lower rating. The math is complicated but explicitly designed to push all competitors to a "natural" rating which represents how strong they actually are, based on how well they do against everyone else. If two competitors have 1700 rating, they should win against each other about half the time. If one has 1800 and the other has 1700, you'd expect the 1800 to win more often (the percentage advantage is actually calculable).
The implicit assumption built into ELO matching is that all things being equal, players should have to face equally strong players. And ELO defines how strong they are. Roster strength does not define how strong a player is. A player with a large roster can be weaker than a player with a much smaller roster in actual gameplay. We don't match rosters, we match players. And ELO is self-correcting. We have no idea how strong a player is until we watch him play. But even if we assign a totally random rating to that player, their rating will move towards the correct one. If we give him a rating that is lower than his actual strength he will win more often (because he will be matching against players equal to his rating, which is lower than his actual strength), and his rating will go up. If we give him a rating that is higher than his actual strength he will lose more often and his rating will go down.
The mechanics of ELO mean that over time everyone will match closer to their actual playing strength, because their rating keeps changing to reflect their strength. Any other metric that is independent of win/loss record fails in this. If we match with any other criteria, be it prestige or roster strength or the height of the player, those metrics cannot possibly accurately measure the player's true battlegrounds strength. They will be "wrong." But for every player whose matching metric is wrong, it will basically be permanently wrong. If they are being matched against players weaker than they are, they will always match against the same weaker players, getting a free ride to GC. If they are being matched against stronger players, they will always match against those same stronger players and get screwed. And there's nothing they can do about it.
The discussion surrounding this problem has morphed into a discussion of whether it is fair for Paragons to face Uncollected players or if it is fair for Cavalier players to only fight other Cav players. But this actually misses the real problem. The real problem is not that the current roster strength system makes it easier for Cavs and harder for Paragons. Not all Paragons are hurt by the current system. Not all Uncollecteds are helped. Rather, the problem is for every roster strength from UC to Paragon, there are players who are matching against incorrect competition because we are using a metric that doesn't represent their true strength. And the game is currently ignoring everyone's win/loss record that is telling the game who is stronger and who is weaker.
Weaker players are getting matched against stronger players. Not just Paragons, not just Uncollecteds, but everyone across the entire game. Some players are getting an unfair advantage and some are getting an unfair disadvantage, and the system never self-corrects this. In Alliance War, even new alliances that start with zero rating quickly climb to higher ratings because they win. Because they win, their rating improves, until they are fighting the right competition. In Battlegrounds GC, the same thing happens. Winners climb the rating ladder and are forced to face other winners who are also climbing the rating ladder. Losers fall, but they then face other loses and have a chance to climb back up again.
(Almost) Everyone wants fair competition in general. Everyone says fair competition is where people face "equal" competition. But the problem is everyone defines "equal" differently. I believe the correct definition is: equal competition is when equally strong competitors face each other. Not when equal rosters face each other. Not when equal decks face each other. When equal competitors face each other.
This is probably the most controversial issue across all Battleground issues. I don't really expect to gain perfect consensus here. But I think even this most radioactive of disputes has at least some general consensus. The problem is that some people don't actually want perfectly fair competition, because there's another factor to consider: participation. Competitions are meant for competitors but not all players are equally interested and driven to compete. But the BG game mode needs participants. It needs to fill its turnstile. It needs players to match each other, in real time. Without that density of participants, the mode will fail.
The devs recognized this as well, which is why BG even has a VT and GC. The GC is the pure competition for the top competitors. The VT is intended to be a more balanced participation-driven and competition-driven mode. We want people to participate in VT, but we don't want to wreck the competitive elements completely. Some people feel that roster-matching is more appropriate to encourage participation. And I agree, to a point.
We already have two match systems: roster strength and ELO. They already exist. And I think most people would agree or at least accept that in the very early stages of VT, roster matching is not altogether wrong. In fact I would argue it makes perfect sense. ELO requires actual matches to "refine" its numbers. Until a player has played enough matches, won and lost enough fights against enough players, their ELO score is simply a guess. That guess gets better the more matches they play. So it is fair to say that there is at least some window of time where ELO is not actually as good as it nominally is. So if we know two things about a player, their roster strength and their ELO, but their ELO is shaky, there's no reason why we couldn't match on roster strength until we had enough confidence in ELO to start using ELO.
Ideally we would want a match system that started off looking like Roster matching and ended up looking like ELO matching. So what if instead of inventing a new match system we created a sliding scale of matching. We create a new metric, a "confidence metric" that tells us how much to "trust" ELO. This starts at zero. Every time the player enlists, we randomly pick Roster Match or ELO Match. The confidence score tells us how often to use ELO. When it is at zero, the player will always match Roster. But as confidence rises, the probability they will match according to their ELO also rises. When the confidence score is 50, the game will match them by Roster 50% of the time and ELO 50% of the time. Over time, the player will get "exposed" to more and more ELO-driven matches. If they were getting all weak matches and they were winning most of the time, their match ups will get stronger. Conversely if they were getting all strong matches and they were losing most of the time, their match ups will get weaker over time to a more appropriate level. And when confidence reaches 100, the player will be matching ELO all the time, and leave Roster matching behind.
We could simply set confidence to the number of matches played times ten. So after ten matches players were pure ELO. Or we could do this by tier: all Bronze tier players have confidence zero, all Silver players have confidence 25, and so on. My preference is for players to have confidence zero for their first five matches, then have confidence increase by 10 for every match lost and 15 for every match won.
Also, anyone who starts higher than Bronze due to #2 retains their ELO and confidence rating.
By the way, we have to TELL the players what's going on. The players will never trust a match system they don't understand and have no idea what's happening, and that's especially true of this one. If they are getting a Roster match or an ELO match, the game has to signal that. Otherwise the player will just be mystified why they are sometimes getting strong matches and sometimes weak ones, seemingly randomly (because it is). Players have to understand why the game is doing what it is doing so at least they can comprehend why the game is behaving the way it is. They might not agree, but it is much worse when they think the game is being randomly capricious.
Okay, so here's the TL;DR:
1. Change scoring from win = +1 trophy loss = -1 trophy to:Win 2/0 = +2 trophies
Win 2/1 = +1 trophy
Loss 1/2 = even
Loss 0/2 = -1 trophy
Increase number of tokens required to promote to compensate.
2. Start everyone one full tier lower than they ended the previous season. If the player ended in GC, start in Vibranium. Add solo objectives to allow players starting higher in VT to earn the missed VT progress rewards.
3. Start everyone matching by Roster strength, use a confidence parameter to slowly shift to ELO matching.
Yeah, I don't actually think I'm going to get universal agreement to all of this, or even any of this. But I believe this represents something most players would accept as reasonably fair for the most part, and acceptable enough where it isn't, at least compared to the current system. But, time to find out. Kabam thinks we don't all agree how to move BG forward, and they are right: we don't all agree. But can we is the real question.
Also, I did try to keep this as short as possible. I really, really tried:
Post edited by Kabam Ahab on
160
This discussion has been closed.
Comments
Dr. Zola
This does solve the problems.
Regardless, there have been many good ideas and it's pretty much a consensus (outside of obvious ignorant trolls) that the current system isn't working.
All match systems are random, in the sense that they pick an opponent randomly from a pool of valid opponents. We limit that pool initially by VT tier. We match Bronze 2 against other Bronze 2 players (or at least adjacent tiers). The current system further limits the pool to players with similar strength roster. From what I've seen, most of the players advocating "random" matching are actually advocating eliminating the roster/prestige limit and allowing the match system to find a random opponent from the wider pool of players within or nearby the player's VT tier. That's consistent with the suggestion to match by ELO (which is how GC works) by eliminating the roster limits and imposing an ELO range limit instead.
The only way an Uncollected player would match against a Paragon player with ELO matching is when the UC players is incredibly good and the Paragon player is blind. In which case, they should match against each other due to their relative ability to win. But no small account stumbles randomly into super strong accounts with ELO matching. The small account would have to win a lot of matches to find themselves running into strong Cav accounts, and they have to beat those accounts to find themselves matching against strong TB accounts.
At least, if the "trust metric" for ELO was implemented as discussed above (a weak account can randomly run into a very strong account with ELO matching if both players haven't played enough matches for ELO to be reasonably solid).
Or does the 2-0 stand as is and end at 2-0, and any of the breakdowns you suggest for a 3-fight match is conditional only for cases of an initial 1-1, where the 3rd fight is now acting as like a tie breaker, and like in the NHL how an overtime loss still gets “some” points even though you lost, because you at least tied during regular time.
*2* Good (and I’ve seen this suggested quite a lot by people). But my thoughts would be for maybe dropping everyone down 2 medal colors for new seasons, not just 1 color.
*3* however (without getting into roster vs ELO randomness). And, I guess maybe the definition of what ELO stands for maybe got edited out due to character limit of post ?
What would prevent the ABUSE that people would SANDBAG quite a lot on purpose to intentionally drop their ELO big-time (say, when they are at 0 tokens, like currently with “Santa-Bagging”, but now with much bigger benefits, not just for being gracious and using up some excess Energy). Much like what used to be rampant in AW before some changes a long time ago.
If indeed, your meaning of ELO is as you suggested, a history of your fights. Not taking into account if those fights were “meaningful” or not.
Without getting too deep in the weeds, the idea behind ELO rating systems is that we presume that when any two players go head to head, one of them has some probability of winning. If they are exactly even in strength, then the odds of both sides winning is 50%. But if they aren't equal, one side would win more often. A strong player might beat a weak player 75% of the time, or 90% of the time. There is a mathematical function that could describe this probability: for any two players who have ratings X and Y, this function F(x,y) would spit out the probability of the first player winning. If, whenever two players meet, we increase the rating of the winner and the loser in a specific mathematical way, we can ensure that over time the players' ratings will gravitate towards an equilibrium. That theoretical equilibrium state would be the "true" strength of the player, such that when they fight against all other players with the same rating their own rating stays the same over time (moving up and down of course).
Ultimately, ELO is a bunch of math driven by wins and losses. It would not be far from the truth to say that with ELO matching, winners face winners and losers face losers.
As to ELO sandbagging, there was not enough space to squeeze this into the post, but there are ways to limit the advantages of sandbagging by deliberately losing. For example, we can freeze ELO with forfeits so you can't just easily drop matches. We can also factor into the math the notion that losses at zero trophies are less "meaningful" than losses that actually have the potential to costs trophies. But at the end of the day, it is difficult to tell the difference between someone that is sandbagging and someone that is genuinely losing and ought to face lower competition. That's an inherent weakness in the VT itself: because it is not a pure competition, it is theoretically possible to grind it with pure effort even if you aren't particularly good at it.
The idea is to limit this, with the understanding there's no way to completely eliminate it.
Also, starting stronger players at higher tiers has only beneficial impacts on lower tier players. Those players are eliminated from their bracket. It only "hinders" progress in the jealousy sense: those players might think it was unfair for those players to be allowed to start at a higher starting tier than they do. In which case I would give those players a choice: allow them to start higher, or face them.
The third option, forcing them to start at the same tier but protecting lower players from ever facing them, is not an option I would allow them to keep. They would have to collectively decide for themselves if the net benefits from the system outweigh this hang up or not. If it doesn't, it doesn't, but I am banking on the number of players willing to die on that hill against their own self-interest to be a low enough number that any other suggestion would increase the number of unhappy people. Someone is going to be unhappy, and it should be the smallest number reasonably achievable.
Also, even though they are starting higher, they aren't actually gaining any rewards for starting higher. They still have to earn those VT rewards via objectives. If lower progress players want to start higher along with the higher players and fight them for their VT rewards then I would be fine with that as well. Under this system, any attempt to argue that the lower progress players are getting the shaft is counterable by saying anything they think the higher progress players are getting that they think is unfair, I'm more than happy to hand to them.
In fact, I would pay real money for the devs to implement a system whereby anyone who complained the higher progress players were getting something they weren't getting could be placed into the same position as the higher progress players. Irrevocably.
As for the starting suggestion, you're still skewing the results of the Season before it even begins. That's my hang up. You're essentially starting a new Season with standings already built-in. That's not a competition. That's a carry-over system. I suspect we will have to agree to disagree on that front.
I am 62% for wins losses, gold 3 is more than half the losses for the entire VT so far.
I personally would go random from Bronze but you gotta gift low players something I guess.
Thanks @DNA3000 for your insights in this thread, I hope the appropriate people are reading it.
You want everyone to start at 0 AND allow players to not face the top accounts. That is garbage and MUCH more unfair than the top players starting higher in the ranks (like literally EVERY OTHER competitive pvp game).
Winning the match either earns you one trophy or two. Losing either costs you a trophy or allows you to stay even. No player would be worse off under this scoring system under any circumstances, at least in a per fight sense.
The one type of person that might not benefit would be if someone out there is always winning but always winning 2-1. Such a person is currently advancing as fast as possible but under the proposed system they would not gain any of the benefits of winning 2-0 but would have to progress through a longer trophy track. This is not impossible, but I think such people would be very rare.
Starting out at some level based on previous season does NOT mean it is so locked in like AW is, where AW Tiers heavily dictate how many points you get (Tier Multiplier) compared to others, which yes is purely based on War Rating (and so in a sense, prior seasons play).
Everyone who was in GC in previous season would *ALL* start out at a relative 0 compared to each other.
None of them start in GC (and definitely not higher than any others who also finished in GC too), they still have to all start out in the bracket below GC, and fight their way back into GC. Where they also do not have an advantage versus others based on previous year's GC rank.
They have to re-earn their GC Rank from scratch every season.
Except, it isn't unfair in my opinion for a bunch of reasons. The most important of which is the notion that VT is an impediment to GC that higher tier players get to bypass is flawed. We can illustrate the flaw by simply suggesting we do away with VT altogether. Poof, gone. Everyone automatically starts in GC. Is that better?
Not for low progress players it isn't. They would get instantly destroyed. Their prize for no longer having a VT is they no longer get *anything* from BG. The "chore" they wanted to bypass was the only thing giving most of them anything at all. So anyone who says it is unfair for top tier competitors to bypass some of VT because they should be allowed to do so as well, would need to explain why they think it wouldn't be fair if we just went all the way and deleted VT for everyone, high and low. I think you'll find this will expose the real foundation of this complaint And it isn't "fairness."
VT is not there for the top players. It is there for the bottom and medium players. In effect, VT is a chore placed upon everyone that the lower players need, but the upper players don't need. So what's the justification for forcing them to do it? Structurally, we *have* to make them do at least some of it, because the reward structure pretty much demands it, and because we still need a way to decide who belongs in GC in the first place. But the idea that it is unfair if some players get to skip some of it ignores the fact that it was unfair to begin with to force those players to do VT at all. VT is not conducive to the actual competitive core of BG. VT is a concession. And I say this as someone that will likely not be visiting GC any time soon. I'm one of those players for whom VT is targeted at. But I recognize VT for what it is. A compromise between the competitive aspects of BG and the more inclusive participatory elements. Before we ever ask whether the top competitors are getting an unfair advantage, we have to remember what they were asked to give up to accommodate less competitive players while getting nothing in return.