Stability is a foundational problem in balance

[jul.7602]

The chances that any developer is going to read this is zero, but I think it's something needs to be discussed in detail. Over the last few years I find that Anet has focused too much on fixing symptoms of improper combat balance, as opposed to tackling the root cause: combat design. Yes, there is a difference between combat design and combat balance. Balance is a narrow term describing the mathematical probability of defeating your enemy conditioned on some set of variables. When that probability approaches such that probably of defeating your opponent is the same for any set of build, you can say that the combat is perfectly balanced, conditioned on the choice of build. While that is certainly something to admire, it by no means guarantees that combat is actually engaging and fun for anyone. As an example, during POF beta you had the big red button (Scourge), and the big blue button (Firebrand). One had extreme offensive capability, the other extreme defensive capability, and while they somewhat canceled each other's kittenous abilities the overall game mode during that time was stagnant somewhat infuriating. 

All it seems that Anet is doing lately is just changing the names on the blue and red button respectively. For a time the big blue button was minstrel scrapper, and the big red button changed between power DH, power rev, back to scourge ect. Granted this meta has a bit more variety but the overall issue of approach remains. I think the biggest design issue right now is accessibility of stability, now made even worse by powercreep and CC spam in the latest expansions.

 

Issue #1: STABILITY

Stability is arguably the most important boon in the game, especially in large scale fights and even more so against comped enemies (AKA boonblobs). It is so important for players that I believe Anet has made a combat design mistake by limiting stability to boon applications from a small subset of classes. The Anet approach to this problem would be to slightly lower the CDs of some major stability skills, and I bet that hardly anybody will notice any change. Stability is just about as important as being able to dodgeroll; can you imagine if we needed a boon to be able to execute dodges? That's basically the problem we have here. Every player, regardless of their build/utility needs some baseline access to self-stability. 

Solution #1: Breakbars

Give every player a type of "Breakbar", similar to holding a Dragonbanner but with some key changes.

• Whilst the breakbar is not depleted, you are immune to all hard CC and immobilization. Other movement impairing conditions such as cripple and chill are unaffected. The only CC that will always bypass the breakbar is daze (to allow for skilled well time interrupts).
• Once the breakbar is depleted you will become 'broken' and now fully susceptible to CCs. The breakbar will slowly replenish and once it reaches 100% you will no longer be broken.
• Stability no longer grants CC-immunity. Instead it increases the recovery speed of the breakbar.

Solution #2: Give players a new ability called "Sprint"

Leave the system as is, but allow players to bind an ability to their keyboard called Sprint. Activating Sprint grants the user superspeed and CC immunity for 6 seconds. The cooldown can be about 25 seconds.

 

[SFShinigami.1572]

You know what, it could be interesting to see stab skills converted to group stunbreaks.  Make groups respond to cc instead of trying to keep up it up at all times.  That might be interesting to see, though would probably need some better identifying of party members in a large squad.

[Xenesis.6389]

I heard anet solved this problem by nerfing boon strips, nerfing boon corruption, adding more boons, adding more boon spam, adding more support classes to the system, so players could finally feel comfortable approaching other players. 😏

[Chaba.5410]

1 hour ago, jul.7602 said:

Balance is a narrow term describing the mathematical probability of defeating your enemy conditioned on some set of variables.

"A multiplayer game is balanced if a reasonably large number of options available to the player are viable--especially, but not limited to, during high-level play by expert players."

https://www.sirlin.net/articles/balancing-multiplayer-games-part-1-definitions
 

[Sereath.1428]

Yeah they are aware of how unplayable the game is without permanent Stability, and even Resistance to some extent thanks to Immobilize.

So the cheap approach right now is to make it harder to strip it away.

CC needs an actual rework, it should have hefty diminishing returns so to punish people from unloading all their CCs as soon as they are up.
A simple one would be to half the duration of each new CC that gets applied until you become immune:
50% -> 25% -> 12.5 -> until immune for several seconds.
And only after that they should do something about permanent boons.

Edited June 11, 2023 by XxsdgxX.8109

[babana.7521]

Don't know what the problem is here, some players complained about boon strips, some complains there are too much stability. if both are getting complaints, i guess they must be balanced. 

[jul.7602]

2 hours ago, Chaba.5410 said:

"A multiplayer game is balanced if a reasonably large number of options available to the player are viable--especially, but not limited to, during high-level play by expert players."

https://www.sirlin.net/articles/balancing-multiplayer-games-part-1-definitions
 

It's semantics at this point. What you sourced is really just his particular opinion. The definition I wrote was based on the mathematical definition of a fair coin. Which is, given a sample space, every event in that sample space is equally likely. I don't like that he uses "viable", "large", "reasonably", because those terms really don't mean anything concrete anymore and defeats the point of having a clear definition. I also think including "large" is incorrect. 

 

In a GW2 fight, the sample space has a dimension of 2. You win the fight, or you lose the fight. If the probability of winning the fight is the same as that of losing the fight, you can say that it was a mathematically fair coin. Same situation.

From a PvP perspective, the definition of a balanced fight is not necessarily that each player equally likely to win. It's something along the lines that the controlling for all other aspects besides skill, the probability of the higher skilled player winning is 100%.

 

[Chaba.5410]

2 hours ago, jul.7602 said:

It's semantics at this point. What you sourced is really just his particular opinion. The definition I wrote was based on the mathematical definition of a fair coin. Which is, given a sample space, every event in that sample space is equally likely. I don't like that he uses "viable", "large", "reasonably", because those terms really don't mean anything concrete anymore and defeats the point of having a clear definition. I also think including "large" is incorrect. 

 

In a GW2 fight, the sample space has a dimension of 2. You win the fight, or you lose the fight. If the probability of winning the fight is the same as that of losing the fight, you can say that it was a mathematically fair coin. Same situation.

From a PvP perspective, the definition of a balanced fight is not necessarily that each player equally likely to win. It's something along the lines that the controlling for all other aspects besides skill, the probability of the higher skilled player winning is 100%.

 

What I sourced was an opinion by an experienced and knowledgeable game designer.  There's also a series of blog posts by League of Legends game designers that touch upon similar ideas as Sirlin with regards to the options players have with the different champions at different skill and difficulty levels.  Then, of course, for general reference there's the Wikipedia article on the subject.

It's a little confused what you're trying to say.  Earlier you wrote: "When that probability approaches such that probably of defeating your opponent is the same for any set of build, you can say that the combat is perfectly balanced, conditioned on the choice of build."  This sounds like the same type of balance Sirlin is discussing where "a reasonably large number of options available to the player are viable".

But then you wrote: "As an example, during POF beta you had the big red button (Scourge), and the big blue button (Firebrand). One had extreme offensive capability, the other extreme defensive capability, and while they somewhat canceled each other's kittenous abilities the overall game mode during that time was stagnant somewhat infuriating."  What you're describing is imbalance and doesn't look at all like an example of "probability of defeating your opponent is the same for any set of build".  You described a meta that was boring because it didn't have a reasonably large number of viable options available to the player.

Then you go back to admitting that today's meta has a bit more variety and move on to your issue with stability?

No one would say that two classes cancelling each other out is balance if it's only those two classes as choices.  That's those two classes being too OP and in need of nerfing, which Anet did.
 

[Sheff.4851]

I wouldn't take something that David Sirlin writes and just go "ah but that's just one guy's personal opinion". Sirlin is very smart, and has pretty a pretty good track record with having opinions about how to balance MMOs. But then, maybe OP is Brad McQuaid's alt account.

[jul.7602]

8 hours ago, Chaba.5410 said:

What I sourced was an opinion by an experienced and knowledgeable game designer.  There's also a series of blog posts by League of Legends game designers that touch upon similar ideas as Sirlin with regards to the options players have with the different champions at different skill and difficulty levels.  Then, of course, for general reference there's the Wikipedia article on the subject.
 

A source can be wrong or incoherent. It's more important to understood what is being sourcing, what they are trying to convey, and why. I've stated the glaring issues in his definition. It attempts to define balance, by introducing vague/subjective terms that nobody can ever quantify or agree upon. His definition is quite literally untestable, because nobody can every falsify what my opinion on what is the 'correct' answer for any of the definitions he uses. What constitutes viability in GW2? You'll get probably a hundred different answers. Tempest and druid were viable healers, but most would also say tempest is very favored druid. What constitutes an option? Are we talking about builds or comps or gear? What is considered reasonably large? Are we talking about the percentage of builds in existence that are viable, or just the absolute number of viable builds. I could go on and on. But nothing in the way he defines balance is authoritative, or let alone even testable. Nobody could parse his definition without going through an even worse rabbit hole of trying to agree on the above terms. It's a useless definition.

I could cite David Kim, one of the balance devs of SC2. He looked at multiple metrics, but in general looked more closely at win rates and race-representation in tournaments because they are objective and measurable. If all the respective non-mirror match winrates are close to 50%, and the r04 bracket of a tournament has at least one of each race, he would conclude the game is relatively balanced. His definition is more or less the same as mine, and unlike Sirlin, it can be formulated as a testable hypothesis. I can compute the exact winrates and standard deviations for every premier tournaments and reject or fail to reject my hypothesis. 

9 hours ago, Chaba.5410 said:

It's a little confused what you're trying to say.  Earlier you wrote: "When that probability approaches such that probably of defeating your opponent is the same for any set of build, you can say that the combat is perfectly balanced, conditioned on the choice of build." 
 

I'm applying the definition of a fair event in mathematics to guild wars 2. When you flip a coin, it can be either heads or tails. We say the sample space has a dimension of two. If the probability of each event in the sample space occurring is equally likely, which in this case is 50%, then we say (at least mathematically) it is a fair coin. If there was some sort of environmental factor that made it so that the distribution of heads/tails was different from 0.5, it is no longer a fair coin. In general, there aren't really any obvious environmental factors that would skew the distribution of the coin-toss, so we just leave that out for simplicity.

In guild wars 2 we know that there are a long list of variables that affect the probability of defeating an opponent (builds, comps, ect.). That is why I add "conditioned", at the very end. When all environmental factors in GW2 are controlled for, our situation is exactly the same as a coin toss, and we can directly apply the definition.

9 hours ago, Chaba.5410 said:

This sounds like the same type of balance Sirlin is discussing where "a reasonably large number of options available to the player are viable".
 

Well, again his definition is scientifically untestable and ambiguous.  Sirlin's definition isn't a meaningful definition to begin with so it can be anything you want to be. 

 

9 hours ago, Chaba.5410 said:

This sounds like the same type of balance Sirlin is discussing where "a reasonably large number of options available to the player are viable equally likely to win or lose..
 

Now its probably closer. 

9 hours ago, Chaba.5410 said:

But then you wrote: "As an example, during POF beta you had the big red button (Scourge), and the big blue button (Firebrand). One had extreme offensive capability, the other extreme defensive capability, and while they somewhat canceled each other's kittenous abilities the overall game mode during that time was stagnant somewhat infuriating."  What you're describing is imbalance and doesn't look at all like an example of "probability of defeating your opponent is the same for any set of build".  You described a meta that was boring because it didn't have a reasonably large number of viable options available to the player.

Then you go back to admitting that today's meta has a bit more variety and move on to your issue with stability?

No one would say that two classes cancelling each other out is balance if it's only those two classes as choices.  That's those two classes being too OP and in need of nerfing, which Anet did.
 

Yes. Early POF was both poorly balanced, and poorly designed. The probability of defeating your opponent was not the same for any given build. It was higher based on the number of scourges you had, so everybody stacked scourges.

Yes, the meta today has more options than in other metas, but still suffers from a whole host of design issues. Nothing is contradictory here.

I didn't say that either. My definition of balance is based on winrates being equal when all environmental factors are conditioned for.

[Chaba.5410]

58 minutes ago, jul.7602 said:

I could cite David Kim, one of the balance devs of SC2. He looked at multiple metrics, but in general looked more closely at win rates and race-representation in tournaments because they are objective and measurable.

Why don't you then?  That's nothing different than what the LoL devs wrote about in their blog.  Winrate is used across different skill levels to measure champion viability and depth.  It seems like you didn't bother digesting the Sirlin link - he discusses StarCraft too.  The concepts Sirlin wrote about is tested all the time.  LoL devs wrote a statistical analysis tool to do just that.

 

58 minutes ago, jul.7602 said:

My definition of balance is based on winrates being equal when all environmental factors are conditioned for.

You can have your two big colored buttons example where two classes cancel each other out - your "fair coin" flip - and the winrate ends up being equal.  Now what?  Balance isn't winrate.  Winrate is just a measurement used to detect imbalance amongst the choices given to players, not the goal.

Edited June 12, 2023 by Chaba.5410

[Leger.3724]

10 hours ago, Chaba.5410 said:

What I sourced was an opinion by an experienced and knowledgeable game designer. 

Someone nobody has ever heard of outside of you because they have not produced or "balanced" a game worth mentioning.

Someone self-quoting themselves in 2001 with less arrogance than you because he flat out admits that it is his definition of balance and not the universal definition of game balance.

The link you gave is basic. It has zero analysis. It can be described as "here is the problem, here are some variables and here is how a few games I like roughly fit within those variables". You understand the hard part is making adjustments once that is done, right? Defining a problem is easy. Setting up variables is easing. The solutions part is the hard part.

Edited June 12, 2023 by Leger.3724

[Chaba.5410]

1 minute ago, Leger.3724 said:

Someone nobody has ever heard of outside of you because they have not produced or "balanced" a game worth mentioning.

Someone self-quoting themselves in 2001 with less arrogance than you because he flat out admits that it is his definition of balance and not the universal definition of game balance.

So you never heard of Sirlin.  So what?  Your loss.  Here's some reference:  https://en.wikipedia.org/wiki/David_Sirlin

Julian can use his own definition to talk about stability.  No one else is required to accept his definition though, especially when his examples are confusing and seem contradictory.

[Leger.3724]

Just now, Chaba.5410 said:

So you never heard of Sirlin.  So what?  Your loss.  Here's some reference:  https://en.wikipedia.org/wiki/David_Sirlin

Julian can use his own definition to talk about stability.  No one else is required to accept his definition though, especially when his examples are confusing and seem contradictory.

So an unsourced claim he "balanced" a fighting game released back in 1996 and then "balanced" the remake in 2008. 

Two mildly successful fighting games - the last one released over 15 years ago.

This is the guy you want to present as the authority on successful game balancing? You do you. 

 

[Chaba.5410]

8 minutes ago, Leger.3724 said:

So an unsourced claim he "balanced" a fighting game released back in 1996 and then "balanced" the remake in 2008. 

Two mildly successful fighting games - the last one released over 15 years ago.

This is the guy you want to present as the authority on successful game balancing? You do you. 

 

Julian is capable of defending his ideas himself, you know.  What kind of games has Julian released over the last 15 years?  Has Julian spoken at Game Dev's Conference?  Does Julian write a blog for the layman on game design?  It's really odd of you to question Sirlin's record while letting Julian's slide.  It's disingenuous.

Edited June 12, 2023 by Chaba.5410

[jul.7602]

42 minutes ago, Chaba.5410 said:

Why don't you then?  That's nothing different than what the LoL devs wrote about in their blog.  Winrate is used across different skill levels to measure champion viability and depth.  It seems like you didn't bother digesting the Sirlin link - he discusses StarCraft too.  The concepts Sirlin wrote about is tested all the time.  LoL devs wrote a statistical analysis tool to do just that.

 

You can have your two big colored buttons example where two classes cancel each other out - your "fair coin" flip - and the winrate ends up being equal.  Now what?  Balance isn't winrate.  Winrate is just a measurement used to detect imbalance amongst the choices given to players, not the goal.

Because mathematics is better than anybody else's opinion. i can assure you nobody is using the definition you provided for any statistical analysis, because the definition is you provided is not scientifically testable. It's literally impossible to test the following statement :"A multiplayer game is balanced if a reasonably large number of options available to the player are viable". In order for that definition to be testable you have to be able rigorously define and quantify each parameter. 

For the 2nd paragraph you are obfuscating the difference between mathematically fair, and something that is balanced. If the sample space has a dimension of two, and the the probability of each event is 50%, then it is mathematically fair. When it comes to balance, you have to bake in some additional assumptions.

 

In combat, we're not necessarily looking for each fight to be mathematically fair, because no such fair fights really exist. There is presumably always going to be a difference in skill level, and we assume that those that are trying to balance the game want the victor to be the person with the highest skill. In mathematical terms I would express that as: The probability that person 1 defeats person 2, given person 1 has higher skill than person 2, is 100%.

[Chaba.5410]

Just now, jul.7602 said:

In order for that definition to be testable you have to be able rigorously define and quantify each parameter. 

Literally what the LoL tool does.

[jul.7602]

1 minute ago, Chaba.5410 said:

Literally what the LoL tool does.

You only quoted his definition, and the link you provided doesn't give any detail about statistical tool. Not that it matters because, as I said his first definition is untestable. The statisticians who are probably designing such tools are using much more detailed, and technical language because as I said before, its impossible to test his definition.

[Chaba.5410]

3 minutes ago, jul.7602 said:

You only quoted his definition, and the link you provided doesn't give any detail about statistical tool. Not that it matters because, as I said his first definition is untestable. The statisticians who are probably designing such tools are using much more detailed, and technical language because as I said before, its impossible to test his definition.

The link I provided is a blog written for laymen.  Your only issue is that he's not using technical language in a conceptual article?  Sirlin has a degree in mathematics, btw.  He's not omitting technical language because he's unable to provide it.

If you want to read more about how game devs define balance, which I would think you'd want to do if you're trying to appeal to game devs, I pointed you in that direction by mentioning the blog written by the LoL devs too.  Up to you to utilize those resources.
 

[Leger.3724]

29 minutes ago, Chaba.5410 said:

Julian is capable of defending his ideas himself, you know.  What kind of games has Julian released over the last 15 years?  Has Julian spoken at Game Dev's Conference?  Does Julian write a blog for the layman on game design?  It's really odd of you to question Sirlin's record while letting Julian's slide.  It's disingenuous.

You are the one trying to shut down a discussion with a dubious source. If you have someone with relevant and sourced experience by all means.

[Chaba.5410]

6 minutes ago, Leger.3724 said:

You are the one trying to shut down a discussion with a dubious source. If you have someone with relevant and sourced experience by all means.

Asking Julian to further explain his idea of balance, especially since his examples looked contradictory, isn't shutting down discussion.  It's generating more discussion.  Did you have anything else to add about balance that isn't disingenuous?

Edited June 12, 2023 by Chaba.5410

[Zyreva.1078]

The problem isn't stability, the problem is that the combat system isn't designed for large scale combat and trying to "fix" that would just ruin the system as a whole. It is simply not possible to balance arround situations where players can be exposed to skill spam from 50+ othes.

Edited June 12, 2023 by Zyreva.1078

[jul.7602]

Just now, Chaba.5410 said:

The link I provided is a blog written for laymen.  Your only issue is that he's not using technical language in a conceptual article?  Sirlin has a degree in mathematics, btw.  He's not omitting technical language because he's unable to provide it.

If you want to read more about how game devs define balance, which I would think you'd want to do if you're trying to appeal to game devs, I pointed you in that direction by mentioning the blog written by the LoL devs too.  Up to you to utilize those resources.
 

Your original quote claimed to be an authoritative definition of balance (which there isn't one btw, because balance is not a mathematical term, it is a subjective term based on how we think a victor should be determined). I clearly identified which parts of his definition were lacking using mathematics. I'm sure Sirlin probably has thought of a more rigorous definition, but you haven't cited it. That's not his fault, but rather your fault for not understanding and comprehending what he is trying to convey.

For instance, I get the feeling that you still dont' quite grasp why its necessary to use technical language when your trying to set up a statistical experiment. The term "reasonably large options" can arguably be quantified in a game like street fighter. We can  say that the options, or in other words sample space (notice how I use the mathematically precise definition) is the set of possible character match ups. (Lets also note that we changed our definition. We just went from reasonably large, to "all possible character match ups/entire sample space". You can easily represent this with a square-matrix like figure.

In a Guild Wars 2 fight, what on earth is an option in this context? You have to define the sample space before you can even talk about probability, let alone do some sort of statistical analysis. Let's be generous Chaba, and say that the we can at least define an option as the build that you bring to a fight. Okay well just how many builds are there in GW2? Almost an uncountable number of them. When we have multiple players in a fight, maybe an even better definition is the comps that you can bring to the fight. Okay so what a comp and how can we define the sample space of comps? A comp is basically a permutation of builds, which we already established is basically uncountable. So now our "options" are an uncountable permutation of uncountable builds. 

 

See this is why you need to understand your source as well when you cite it. You can't just copy and paste his layman definition of balance and expect it to make any sense in GW2. There are a discrete number of match ups in street fighter. There are nigh infinite number of builds in SC2 and GW2 so incorporating "reasonably large option" into your hypothetical definition of balance is a nightmare. I could go on and on about why your definition and source doesn't make any sense here.

[Mabi black.1824]

45 minutes ago, jul.7602 said:

A multiplayer game is balanced if a reasonably large number of options available to the player are viable"

I am the last layman who knows how the work of a programmer works. But the sentence I quoted seems to me to be absolutely perfect. just for the concept it's providing. It merely expresses a concept of what you can mean by ''balance'' in a video game. He doesn't presume to explain to you how he gets there mathematically. Probably if he had to explain it with a mathematical model, we would be very few to understand it.

[jul.7602]

1 minute ago, Mabi black.1824 said:

I am the last layman who knows how the work of a programmer works. But the sentence I quoted seems to me to be absolutely perfect. just for the concept it's providing. It merely expresses a concept of what you can mean by ''balance'' in a video game. He doesn't presume to explain to you how he gets there mathematically. Probably if he had to explain it with a mathematical model, we would be very few to understand it.

It's a good laymans definition of what balance could be. It's not a definition or any type of workable hypothesis as Chaba incorrectly suggests.