What is balance, and does it matter in Transcendence?

General discussion about anything related to Transcendence.
bzm3r
Militia Lieutenant
Militia Lieutenant
Posts: 100
Joined: Tue Oct 23, 2012 2:38 pm

Mon Feb 02, 2015 8:22 pm

Often times when a new idea is being discussed, things can get bogged down in considering whether a situation is balanced or not.

I'd like to bring up the following questions for discussion:

1) What is balance?

2) Does it matter in Transcendence?

Here's how I'd start off the discussion (please feel free to insert as many "I think", "I believe", or "In my opinion"s as necessary in order to fully convince yourself that what is written is entirely subjective):

1) "Balance" is a concept from games where competitive encounters between human players is possible. The history of the usage of the term also justifies that definition because usually "balance discussions" come about in RTS games like Starcraft (played competitively), FPS shooters like CounterStrike (played competitively), MMORPGs with competitive elements (e.g. World of Warcraft) or more recently, MOBAs like League of Legends or Dota 2 which thrive on the competitive scene.

Generally, no formula for balance has been found, suggesting that finding a truly balanced state is at least an NP-hard problem, if one was to generously assume that balance is objectively identifiable.

Instead, what is generally observed is that competitive games develop "meta-games", which refers to the "game" invonved in optimizing available gameplay options given a fixed balance state, usually leading to "rebalancing" once optimal or near/optimal options have been found, in order to create viable options against the optimum or remove the optimum entirely. Rebalancing starts the "meta-game" cycle anew.

The term has been transplanted for use in discussion in single player games, where "balance" can refer to a wide variety of things, indicative of its subjective usage in this context. Usually it refers to the question of whether the game has enjoyable enough difficulty curve (high subjective), and the related question of whether the player has the gameplay options in a given scenario to overcome its difficulties (again, highly subjective).

Real Life (also referred to as Outside) is not balanced. Since games often attempt to reach the level of enjoyment and and involvement that Real Life offers, in order to create a sense of emotional attachment or reaction similar to some of the gameplay opportunities of Real Life, the fact that Real Life is not balanced (and doesn't attempt to be), is an important point to note. Humans and Insects, some of the the most prolific guilds in the instance called Earth are well known to be able to constantly find optimal solutions given a current state of balance (so called, "adaptation"), although whether they will continue to be able to do so in the future remains an open question.

2) Transcendence does not need balance even if it were to one day become multiplayer, because the continual process of "balancing" attempts to solve a problem which is likely not a problem in the first place, and if it is a problem, it has no known solution. Rather, creators in Transcendence only need to focus on the creation of gameplay options, as many as possible, along with improving the algorithms behind NPCs to take advantage of as many of these gameplay options as possible.

george moromisato
Developer
Developer
Posts: 2950
Joined: Thu Jul 24, 2003 9:53 pm
Contact:

Mon Feb 02, 2015 8:54 pm

I agree with you, if you stick to that definition of "balance". But I've colloquially used "balance" to refer to different things (not always consistently). I've used the word "balance" in the following ways (at minimum):

1. To compare two items (weapons, armor, etc.) against each other. For example, imagine two 3rd level weapons. If weapon A is completely superior to weapon B then the player has one fewer choice. Choosing between buying weapon A or weapon B ceases to be a choice because buying weapon A is always better. [And of course, interesting choices are good.] In these cases I often say that weapon A and B are "unbalanced" with respect to each other.

2. To compare an item against an absolute scale. For example, most of the game uses the concept of item level as a useful metric. We show the item level to the player to give him/her an idea of the item's power. Internally, we use level to make decisions about item drops. Etc. Obviously this only works if items of a given level are more or less "balanced" with respect to other levels.

3. To compare cost vs. benefit. Most actions in the game have a cost/benefit trade-off. Generally, high-cost actions and high-risk actions have greater benefit than low-cost/low-risk actions. For example, if you attack a station 2 levels above you then you have a chance of gaining treasure 2 levels above you. But of course, your chance of dying increases also. Sometimes, due to design problems, an action like farming Ferians happens to be low-cost but high-reward. In those cases, I usually think of that as "unbalanced".

Of course, I agree with you that there is no known algorithmic method for balancing the game. I think it is still more art than science because of (a) the complexity of the game, and (b) the different skills (and desires) of different players.

But it's not quite entirely subjective either. Numbers (such as DPS) can help to guide the process and avoid obvious errors.

Where I think we agree is that the goal of the game is to provide a satisfying experience. There is no point in talking about balance unless it is in service of that.

Anyway, feel free to push back.

User avatar
Song
Fleet Admiral
Fleet Admiral
Posts: 2725
Joined: Mon Aug 17, 2009 4:27 am

Mon Feb 02, 2015 9:28 pm

george moromisato wrote: Of course, I agree with you that there is no known algorithmic method for balancing the game. I think it is still more art than science because of (a) the complexity of the game, and (b) the different skills (and desires) of different players.
Yesssssssssss.


*throws out several pages of trying to follow the balancing tables for particularly unusual weapons in SM&M++*



But seriously, I do agree with this. Weapons that are in service should, generally, be competitive. If they've failed, they can be inferior products. This doesn't always follow ingame: there are weapons that are statistically inferior that cost more than their better counterparts....but that's usually reasonable. The only thing that I think really isn't "balanced" ingame right now (apart from stuff that's outright broken, like ICX...or victim of power-creep, like the Lamplighter/QAC) is the way that high-ROF/low damage weapons are penalised heavily be resistances and roundoff errors. Which is....sorta justified, but does make them really annoying to use.

Making stuff feel good (especially for weapons) is always nice. I've used statistically worse weapons before because they were fun to use.
Mischievous local moderator. She/Her pronouns.

bzm3r
Militia Lieutenant
Militia Lieutenant
Posts: 100
Joined: Tue Oct 23, 2012 2:38 pm

Mon Feb 02, 2015 9:34 pm

george moromisato wrote:1. To compare two items (weapons, armor, etc.) against each other. For example, imagine two 3rd level weapons. If weapon A is completely superior to weapon B then the player has one fewer choice. Choosing between buying weapon A or weapon B ceases to be a choice because buying weapon A is always better. [And of course, interesting choices are good.] In these cases I often say that weapon A and B are "unbalanced" with respect to each other.
Maybe weapon B is not an interesting choice if you have the option to acquire both weapon A, and weapon B, but usually in a game, you don't have every option available, right? So, weapon B still has its place as being *an option*. Removing it, and putting in weapon C instead (a changed version of weapon B), reduces the number of options in the game, because now the player cannot buy weapon B anymore. Instead, weapon C should simply added to the game alongside weapon A and B.

Then you might have a story play out where weapon B was the only weapon available to the player during an emotional time in the game, causing them to be emotionally attached to weapon B not because it was the *best*, but because it was there...so maybe when its time and they have all the credits in the world to choose between A, B and C, they still stick with B because B is "old faithful" -- and even if it isn't that good, the *player's* good enough to use it reliably.

A matter of pride. Or, you might have a story play out where the player *just made the wrong decision* given all the options, well -- oh well. Wrong decisions are part of life!

Or the player never uses B, but B shows up in catalogues and the player will get to read its description at the very least -- B is something that's available, even if its useless -- B is part of the little bits that make up the universe -- and the universe is more than just the sum of its little bits.
george moromisato wrote:2. To compare an item against an absolute scale. For example, most of the game uses the concept of item level as a useful metric. We show the item level to the player to give him/her an idea of the item's power. Internally, we use level to make decisions about item drops. Etc. Obviously this only works if items of a given level are more or less "balanced" with respect to other levels.
If weapon B is not as good as weapon A, and it messes with the item level system (which I agree is useful), then balancing is only a matter of a change in description: change weapon B's level to be lower than weapon A?
george moromisato wrote:3. To compare cost vs. benefit. Most actions in the game have a cost/benefit trade-off. Generally, high-cost actions and high-risk actions have greater benefit than low-cost/low-risk actions. For example, if you attack a station 2 levels above you then you have a chance of gaining treasure 2 levels above you. But of course, your chance of dying increases also. Sometimes, due to design problems, an action like farming Ferians happens to be low-cost but high-reward. In those cases, I usually think of that as "unbalanced".
On IRC once, I remember someone describing how they had this whole system setup for farming Ferians (it was a problem of logistics, they explained) efficiently. Someone else might balk at the notion of mining Ferians knowing that they are human-derived, and that there is an ongoing-conflict in the universe (whether lip service or not) about the Ringers and Teratons creating slaves in the form of Ferians (that is part of the reason of the Ares' hostility with the Ringers?). So, even the game brings up that moral question. Even if the choice is low-risk, high reward, its not uninteresting because the player has to decide what their definition of human being is! That's pretty interesting, no?

Again with the Ferians, the solution is not to remove the "unbalance", but to add new options and consequences: it could be as simple as providing more opportunities for the player to learn about Ferians. Maybe item drops from Ferians that have no real value in the game apart from providing information about what Ferians are/their history -- "a crude amulet with what seems like four Ferian stick figures scratched in" (aww, he had a family and I killed him...) or "a human skull, with Ferian teeth marks all over it" (omg maybe that guy in the UAS mines wasn't crazy! -- evil bastards, I'm going to wipe them all out)? Just a crude example, of course, but it seems again the solution is to add content.
george moromisato wrote:But it's not quite entirely subjective either. Numbers (such as DPS) can help to guide the process and avoid obvious errors.
I totally agree with you there. I think what I am trying to say is this: if content was developed with thought and care, then later if the content is bug free (bugs need to be fixed) and it seems that "rebalancing" is needed, I say we avoid the urge to change numbers until things work right again (time consuming), and instead think about how we can add options/consequences that are of interest (the time would have been spent anyway "balancing", why not just add content instead with that time), given the content that's already there.
Last edited by bzm3r on Mon Feb 02, 2015 9:50 pm, edited 2 times in total.

bzm3r
Militia Lieutenant
Militia Lieutenant
Posts: 100
Joined: Tue Oct 23, 2012 2:38 pm

Mon Feb 02, 2015 9:40 pm

Shrike wrote:
george moromisato wrote: Of course, I agree with you that there is no known algorithmic method for balancing the game. I think it is still more art than science because of (a) the complexity of the game, and (b) the different skills (and desires) of different players.
The only thing that I think really isn't "balanced" ingame right now (apart from stuff that's outright broken, like ICX...or victim of power-creep, like the Lamplighter/QAC) is the way that high-ROF/low damage weapons are penalised heavily be resistances and roundoff errors. Which is....sorta justified, but does make them really annoying to use.
That's a bug fixing issue though, not a balancing issue, agreed?

I think that when making new content for the game, its worthwhile putting in some thought as to how it meshes with the rest of the game, but I think that process shouldn't be thought of as balancing, and it shouldn't be a continuous iteration.

Rather, if we are so inclined to make it a matter of continuous iteration, we should look into how we can put in code that allows NPCs to "engineer" either improvements to their items, or new items entirely. So, they'd need a way to figure out where shortcomings are (e.g. CW could check to see which ship type has taken the greatest number of losses, and try out various things...), a way to figure out which changes to make (here there's a lot of content possibility), and then a way to check to see if there's an improvement (could be the same method by which they see if there's a shortcoming).

Let the world adapt.

george moromisato
Developer
Developer
Posts: 2950
Joined: Thu Jul 24, 2003 9:53 pm
Contact:

Mon Feb 02, 2015 11:54 pm

bzm3r wrote:Rather, if we are so inclined to make it a matter of continuous iteration, we should look into how we can put in code that allows NPCs to "engineer" either improvements to their items, or new items entirely. So, they'd need a way to figure out where shortcomings are (e.g. CW could check to see which ship type has taken the greatest number of losses, and try out various things...), a way to figure out which changes to make (here there's a lot of content possibility), and then a way to check to see if there's an improvement (could be the same method by which they see if there's a shortcoming).

Let the world adapt.
This is a cool idea (if I understand it correctly). I've always thought about using some kind of genetic algorithm for item design based on either frequency across the entire player-base. That is, imagine that some set of items in the game are randomly generated according to some set of constraints. We upload the frequency with which the items are used and then feed that back into the frequency of them appearing in a game.

bzm3r
Militia Lieutenant
Militia Lieutenant
Posts: 100
Joined: Tue Oct 23, 2012 2:38 pm

Tue Feb 03, 2015 1:05 am

george moromisato wrote:
bzm3r wrote:Rather, if we are so inclined to make it a matter of continuous iteration, we should look into how we can put in code that allows NPCs to "engineer" either improvements to their items, or new items entirely. So, they'd need a way to figure out where shortcomings are (e.g. CW could check to see which ship type has taken the greatest number of losses, and try out various things...), a way to figure out which changes to make (here there's a lot of content possibility), and then a way to check to see if there's an improvement (could be the same method by which they see if there's a shortcoming).

Let the world adapt.
This is a cool idea (if I understand it correctly). I've always thought about using some kind of genetic algorithm for item design based on either frequency across the entire player-base. That is, imagine that some set of items in the game are randomly generated according to some set of constraints. We upload the frequency with which the items are used and then feed that back into the frequency of them appearing in a game.
Yes, you do understand correctly! Likely, a genetic algorithm would work well, but figuring out how to encode weapon/armour information in a way that makes sense without getting totally nonsense combinations at times would be challenge -- but a really lovely one!

As for data from the entire playerbase...well, I have to admit I didn't dream as ambitiously, but now that you mention it it could be possible! Every time a game ends, we could send Multiverse data about how many ships were killed, and by whom, and by what, etc. Or, we could also just make the world evolve based on local (player's computer only) information. Another option is that we could just simulate lots of "screensaver" battles (no painting of those scenes to increase performance) before a game starts up, and collect data from it, with additional evolution (albeit, more limited, due to lower number of encounters) as the player plays? Here, parallel processing would be very useful, if a player's computer has more than one processor/core for use. Then, every new game would still be different in terms of the various ship setups that exist in the world.

Hey, actually -- we could pass off the number crunching to Hexarc servers! Right?

bzm3r
Militia Lieutenant
Militia Lieutenant
Posts: 100
Joined: Tue Oct 23, 2012 2:38 pm

Tue Feb 03, 2015 1:31 am

george moromisato wrote:I've always thought about using some kind of genetic algorithm for item design based...
There is also simulated annealing that we could look into.

namer4
Miner
Miner
Posts: 34
Joined: Tue Jul 11, 2006 5:30 am

Tue Feb 03, 2015 9:10 am

george moromisato wrote:
bzm3r wrote:Rather, if we are so inclined to make it a matter of continuous iteration, we should look into how we can put in code that allows NPCs to "engineer" either improvements to their items, or new items entirely. So, they'd need a way to figure out where shortcomings are (e.g. CW could check to see which ship type has taken the greatest number of losses, and try out various things...), a way to figure out which changes to make (here there's a lot of content possibility), and then a way to check to see if there's an improvement (could be the same method by which they see if there's a shortcoming).

Let the world adapt.
This is a cool idea (if I understand it correctly). I've always thought about using some kind of genetic algorithm for item design based on either frequency across the entire player-base. That is, imagine that some set of items in the game are randomly generated according to some set of constraints. We upload the frequency with which the items are used and then feed that back into the frequency of them appearing in a game.
I love this, and I think an initial pass would be pretty easy from an algorithmic point of view. One possibility would be to use a Dirichlet process approach (en.wikipedia.org/wiki/Dirichlet_process), maybe conditional on the particular merchant (e.g., an Akan-like cluster for Rasiermesser). This would require defining a distribution over an item class's features conditional on level, and these distributions are probably all-but-implicit in George's current constraints/rules-of-thumb, and setting a couple of parameters determining how diverse weapons tend to be & how often new ones occur. As an aside, this would also, I think, support demand-based pricing; price could scale, to some extent, with a weapon's relative popularity.

I do this kind of thing fairly often, so if there's interest and someone can point me to some tables/constraints, I'd be happy to put together a proof-of-concept. It would also be fairly straightforward to make it "GA-like" -- having random new weapons be generated from especially-popular parents rather than from scratch.

Edit: To be clear, I've never touched translisp, so it'd be more "here are examples of item features and merchant inventories" and scala code that someone else could look into porting, rather than pluggable code.

User avatar
sun1404
Militia Captain
Militia Captain
Posts: 527
Joined: Thu Nov 03, 2011 10:32 am
Location: Heretic. (Finally!)

Tue Feb 03, 2015 4:00 pm

The problem is: unbalanced contents actually reduces the sense of choice. If an item is much better than another but have the same cost, rarity, etc. it's gonna be stupid to choose anything but that item. What you have suggested (making it more available than other items, that is, making it less rare) is a way to balance items. We don't usually just remove unbalanced contents. We make them balanced. Whether we do that by changing their performance, rarity, cost, side effects, or other properties is up to circumstances.

Yes, junk items (those that are so weak for their level/rarity/cost) have some benefits in the form of storytelling through description, or filling the gaps in the details of the universe, but they can serve those purposes while being balanced, and thus useful in other cases, as well.

Unbalanced contents may also ruin the fun and ruin other contents. Imagine if the Iocrym weapons were to have howitzer range. That's unbalanced. And it'd make Heretic less fun for anyone but the most hardcore players. Imagine if Kate's cruiser in the Arena mission were to be so weak a stray shot can kill her. The quest would be much more frustrating. Imagine if she's so strong and aggressive she can bomb the station and rescue Katami by herself. The mission would be boring. That's why balancing is important. The process involves making items and entities have properties that fit their place and role, not just removing bad things.
Yes, look at my avatar, I have a wyvera type ship.

bzm3r
Militia Lieutenant
Militia Lieutenant
Posts: 100
Joined: Tue Oct 23, 2012 2:38 pm

Tue Feb 03, 2015 5:02 pm

sun1404 wrote:The problem is: unbalanced contents actually reduces the sense of choice. If an item is much better than another but have the same cost, rarity, etc. it's gonna be stupid to choose anything but that item. What you have suggested (making it more available than other items, that is, making it less rare) is a way to balance items. We don't usually just remove unbalanced contents. We make them balanced. Whether we do that by changing their performance, rarity, cost, side effects, or other properties is up to circumstances.

Yes, junk items (those that are so weak for their level/rarity/cost) have some benefits in the form of storytelling through description, or filling the gaps in the details of the universe, but they can serve those purposes while being balanced, and thus useful in other cases, as well.

Unbalanced contents may also ruin the fun and ruin other contents. Imagine if the Iocrym weapons were to have howitzer range. That's unbalanced. And it'd make Heretic less fun for anyone but the most hardcore players. Imagine if Kate's cruiser in the Arena mission were to be so weak a stray shot can kill her. The quest would be much more frustrating. Imagine if she's so strong and aggressive she can bomb the station and rescue Katami by herself. The mission would be boring. That's why balancing is important. The process involves making items and entities have properties that fit their place and role, not just removing bad things.
I think that when making new content for the game, its worthwhile putting in some thought as to how it meshes with the rest of the game, but I think that process shouldn't be thought of as balancing (that's just design), and it shouldn't be a continuous iterative process throughout the lifetime of the game.

User avatar
sun1404
Militia Captain
Militia Captain
Posts: 527
Joined: Thu Nov 03, 2011 10:32 am
Location: Heretic. (Finally!)

Wed Feb 04, 2015 1:49 am

Balancing is a question of design, yes. Maybe you don't like the word, but that's it's meaning. And no, we don't mean to make balancing a long-term process. It is simply that when adding new contents, there's just so many factors that influence whether it's balanced or not, that it's impossible to determine it at first. Only after many people play the game, find the new thing, interact extensively with it, only then the problems will show themselves. And they will. The more complicated the new content, the more problems will find their way in with it.
Yes, look at my avatar, I have a wyvera type ship.

bzm3r
Militia Lieutenant
Militia Lieutenant
Posts: 100
Joined: Tue Oct 23, 2012 2:38 pm

Wed Feb 04, 2015 2:19 am

sun1404 wrote: It is simply that when adding new contents, there's just so many factors that influence whether it's balanced or not, that it's impossible to determine it at first. Only after many people play the game, find the new thing, interact extensively with it, only then the problems will show themselves...
Long term consequences aren't problems, they're just consequences. Sure, maybe you didn't plan for them, but it doesn't mean you don't have to like them and view them for what they are: new opportunities.

So, you see "design" is a new idea, and your best guess for how it should be implemented. "Balance" is a potentially misguided sense of "how the game should be" and attempting to keep on "fixing it" to that state (time wasting), rather than seeing what opportunities it presents and taking advantage of them by adding new content, or changing the dynamics of the existing content in interesting ways that leave the game with more changed than simply different numbers.

I don't feel strongly either way about it, its just a theory -- we can see if it has any worth. Likely, like usual, there is no black-and-white answer.

User avatar
sun1404
Militia Captain
Militia Captain
Posts: 527
Joined: Thu Nov 03, 2011 10:32 am
Location: Heretic. (Finally!)

Wed Feb 04, 2015 2:18 pm

What I have been trying to say is: It's not like we strive to make everything fall in a strict line we made up. When we add new contents in we only fix problems that need fixing. Only those that are so out of balance it's not merely an unplanned consequence, but an unseen problem.

As I see it, 'design' applies to both creating new things and making changes to old things. It's meaningless to differentiate the two. And I don't think it's meaningful to differentiate the sense of how the game should be as 'misguided' or not too. Each person have different versions of the perfect game of their own, and they can get nearer to that by using mods. What George fix in the core game is, as I see it, those contents that, by their unbalancedness, cause considerable impact on other parts of the game, or remove some aspects of the game that was important.
Yes, look at my avatar, I have a wyvera type ship.

User avatar
Atarlost
Fleet Admiral
Fleet Admiral
Posts: 2391
Joined: Tue Aug 26, 2008 12:02 am

Wed Feb 04, 2015 7:14 pm

bzm3r wrote:1) "Balance" is a concept from games where competitive encounters between human players is possible. The history of the usage of the term also justifies that definition because usually "balance discussions" come about in RTS games like Starcraft (played competitively), FPS shooters like CounterStrike (played competitively), MMORPGs with competitive elements (e.g. World of Warcraft) or more recently, MOBAs like League of Legends or Dota 2 which thrive on the competitive scene.
This is where you're very wrong. Balance is also used in cooperative games and single player games and has been for longer than most of those have been things.

Is the Magic User worth his share of the loot? How about the Thief? If you only have three friends and one obviously has to be the DM is it equally viable to go without a Fighting Man or without a Cleric? Does the difficulty curve stay reasonably flat such that the game is playable at all levels?

If the Monk is too strong or too weak it's either a waste of paper or makes the Fighting Man, Paladin, and Ranger wastes of paper. Getting it right is finding a balance. If the game in general is too easy people will get bored and if it's too difficult people will get frustrated. Getting it right is finding a balance. The game needs to be easier early on so new players can get their teeth into it. If the difficulty rises too slowly it lacks replay value. If it rises too quickly the new players don't have time to get hooked. Getting it right is finding a balance.

Even in multiplayer balance is usually more about interesting options than equality. Both players have equal opportunity to play the cheesy class that renders all others obsolete after all. The choice to play a good class rather than a bad class is no more in need of protection in a competitive environment than the choice to swing at the baseball rather than standing at bat like a statue and hoping the pitcher is incompetent enough to give two balls out of the strike zone for every good pitch.
Literally is the new Figuratively

Post Reply