Why does the party need to be balanced?

[kyoryu]

If you wanted to go all game-theory on it, you could say that "there are no dominated choices in the main choices available".

[ezekielraiden]

I mean to more fully respond to your whole post here, but these two needed special, immediate attention because....yeah.

My contention (really, my experience turned into a potentially untrue generalization) was that 3e was fairly unique in the extent to which one could both make those characters, and allow them to exist at most any power level relative to the party.

No offense intended, though I would understand if you found this offensive: You have very limited experience with any system other than 3e D&D, yet you are arguing that 3e D&D is uniquely good at things. That's...kind of a ballsy claim, don't you think? That is, you're arguing that the only game for which you have comprehensive experience is uniquely good at something. Even if numerous people weren't telling you that in their experience it is uniquely bad at that thing, and that numerous other systems do that specific thing better, arguing "the one thing I know is uniquely amazing!" is a pretty suspicious claim. How can you know it's unique, if you have no experience with the alternatives?

Sounds like you get some big reasons why I hate it in practice.

"Can I do x?" <snip> "no, that isn't balanced with your contribution so far this session"

I have no idea how you came to the conclusion that this is what people who care about balance typically (or even ever...?) say. But I can say without equivocation that, over a gaming career spanning at least 20 years, I have never heard even a single person say that reply, whether they cared about balance or not. Because that's not how "balance" is done in...any meaningful sense, in any forum, campaign, or system I've ever participated.

Given how many people are either very confused or strongly in disagreement with you, perhaps your use of the word "balance" is rather different from...basically everyone else's? You keep referring to the "balance" of things like aesthetics and labels that...can't be balanced in any sense of the term I've ever heard. You refer to "balance" requiring "imbalance," balance being enforced as a "nope you had your turn with the gavel, you HAVE to give it up now" etc., none of which match up with what anyone else is talking about as far as I can tell.

If you wanted to go all game-theory on it, you could say that "there are no dominated choices in the main choices available".

Which, uh, I kind of did by saying that degenerate strategies are avoided, and that intelligent play causes the play-experience intended.

[Morty]

If you radically redefine the commonly used term "balance" then yes, 3e is "balanced". If I had to give a shot at defining how balance is defined in common gaming parlance I'd say something along the lines of "Every major choice in the game is roughly mechanically equal to every other major choice and it is hard/impossible for someone making a major choice to select something significantly more or less powerful than someone else making a major choice." You can quibble over what exactly constitutes a major choice, in the context of D&D I'd say it's picking a class, in the context of Starcraft I'd say it's picking a race.

You can take it further down too and say that the game becomes even more balanced if every, shall we call them minor choices, are also balanced against each other. In D&D 3e let's call them picking builds, I suppose (In Starcraft you'd also call them builds). I'd say that it isn't strictly necessary to also balance these to call the game itself balanced, this can get into a realm of player skill. How much player skill should matter is a matter of taste, and I think is a lot murkier of a question.

I don't think balance is inherently important in the context of an RPG either. But it becomes more important in the context of a game that's mostly about killing things and characters are measured in their ability to kill things. Burning Wheel (the best RPG ever made) doesn't particularly care about balance. Elves and dwarves are significantly more powerful than humans, a noble is significantly more powerful than a peasant, and so on. But balance doesn't matter as much because the game isn't about a group of people all contributing to kill things together.

This is why I prefer to go with the "does what it says on the tin" definition, which I've brought up many times. It's similar, but more difficult for people to argue about in bad faith. D&D claims that if you're a fighter or ranger and specialize in archery, you'll be able to properly contribute this way. This isn't the case. And so on.

[NichG]

So, y'know what? Fine. I'll concede--that I was not specific enough about "goals." Literally every place I have said "goals," I specifically mean game design goals. I had thought that was understood, since we're talking about balance (something that is only meaningful in the context of design), I have repeatedly and specifically spoken about the mathematical nature of these things, and you yourself have even engaged with me about it (such as the Skill Challenge discussion above). So, fine, you win that. I'm talking pure design. Not aesthetics. Not writing. Not art. Not labelling. Not anything outside of the design of the game, the stuff that admits, to any meaningful degree, mathematical testing.

Would you consider it to be incoherent to set as a game design goal the creation of a game which is intentionally and actively resistant to mathematical testing? A game designed, for example, to make any sort of zoomed out analysis or theory-crafting as hard as possible (or to disadvantage it as much as possible) compared to behaviors determined organically during play?

[ezekielraiden]

Would you consider it to be incoherent to set as a game design goal the creation of a game which is intentionally and actively resistant to mathematical testing? A game designed, for example, to make any sort of zoomed out analysis or theory-crafting as hard as possible (or to disadvantage it as much as possible) compared to behaviors determined organically during play?

It...yes, but only because "not testable" is, technically, the end result of a (theoretically infinite) set of tests? That is, "non-non-testability" is testable; if you can produce any test, any metric whatsoever, you've proven it's non-non-testable. Essentially, because of the unique nature of "non-testability," it is the exact inverse of an actually (indeed, trivially) testable goal, "is this game testable in any way?" and therefore...I guess you could call it "shadow-testable."

This is identical to the difficulty of proving a universal negative (e.g. "for all things x, no x is a black swan" aka "there is no x that is a black swan"), because in some sense it is a universal negative, identical to the claim, "No tests exist that could apply to this object." That's a super strong claim, easily defeated by even a minor counter-example. In a sense, you're sort of looking for the game-design equivalent of a Weierstrass function (a function that is continuous on all open or closed intervals of the real line, but which is differentiable over exactly zero of those intervals)--something that has rules of some kind, but where the rules cannot even in principle be subject to any form of mathematical optimization or game-theoretic choice analysis. That's a very tall order, assuming you consider degenerate solutions unacceptable (e.g. the null game has no choices and therefore trivially no analysis can be done).

I would be very surprised if such a thing really did exist, a nontrivial completely non-testable game, that was still meaningfully "a game." It somewhat implies an aggressive randomness, like a lottery where none of your behaviors matter to the outcome, or an arbitrary and capricious game where running literally identical inputs in all ways can produce inconsistent outputs. That is, so-called "games" where changing strategy either has no effect, or has inconsistent effect, so there's no way to learn to "play better"--you just get whatever result you get, totally independent (or only capriciously dependent) on your choices/strategy.

Edit:
I just realized, you've made a distinction I reject. In the stuff I'm talking about, "behaviors determined organically through play" are not distinct from "theory-crafting." Both are valid inputs for studying a game--that's why playtesting is so incredibly vital, because it DOES give you "behaviors determined organically through play." I'm not sure if it's possible to make a game that's guaranteed impossible to analyze theoretically, but given the difficulty of analyzing things like Go, it's certainly possible to make games where pure analytic solutions (in the technical sense: "exact solutions" is how this is usually phrased in physics) are entirely unfeasable in the generic, and experience is required to really "understand" the game. (Hence why AlphaGo and the like can beat humans: a computer can, in effect, reference "all games humans have ever played" as its experience pool, and encode the data from that experience-set as a large cluster of neural network nodes, to produce with high probability the likely best move for a given input.)

But real game design incorporates those practical elements as well as theoretical ones. Playtesting and actually good survey design are so vital for exactly that reason. Badly-designed surveys tell you only what you want to hear, or don't tell you anything at all; bad playtesting does the same, concealing flaws rather than revealing them.

[NichG]

It...yes, but only because "not testable" is, technically, the end result of a (theoretically infinite) set of tests? That is, "non-non-testability" is testable; if you can produce any test, any metric whatsoever, you've proven it's non-non-testable. Essentially, because of the unique nature of "non-testability," it is the exact inverse of an actually (indeed, trivially) testable goal, "is this game testable in any way?" and therefore...I guess you could call it "shadow-testable."

This is identical to the difficulty of proving a universal negative (e.g. "black swans do not exist"), because in some sense it is a universal negative, identical to the claim, "No tests exist that could apply to this object." That's a super strong claim, easily defeated by even a minor counter-example. In a sense, you're sort of looking for the game-design equivalent of a Weierstrass function (a function that is continuous on all open or closed intervals of the real line, but which is differentiable over exactly zero of those intervals)--something that has rules of some kind, but where the rules cannot even in principle be subject to any form of mathematical optimization or game-theoretic choice analysis. That's a very tall order, assuming you consider degenerate solutions unacceptable (e.g. the null game has no choices and therefore trivially no analysis can be done).

Well I did say 'resistant to testing' rather than impossible... But anyhow, I don't think you need to get nearly that exotic. It's already been mentioned in this thread (by Cluedrew I think?) that one big principle of designing balanced sets of abilities is that you try to make the abilities work in such a way that they're not comparable to each-other in a vacuum. Is it better to gain the ability to jump in a knight's move, or to be able to read the childhood memories of anyone you defeat in any sort of contest? The answer is context dependent, playstyle dependent, etc. Since the playout - in practice - will depend on random variables associated with individual players, scenarios, etc, then any attempt to do a mathematical analysis in a vacuum will have to assume distributions for those things at best, while any given situation, player, or table will have a specific value for those variables.

Edit: Noticed that you expanded the post while I was replying, so this response may be a bit stale. I'll wait for a response before extending so we don't just go back and forth with edits.

[ezekielraiden]

Well I did say 'resistant to testing' rather than impossible... But anyhow, I don't think you need to get nearly that exotic. It's already been mentioned in this thread (by Cluedrew I think?) that one big principle of designing balanced sets of abilities is that you try to make the abilities work in such a way that they're not comparable to each-other in a vacuum. Is it better to gain the ability to jump in a knight's move, or to be able to read the childhood memories of anyone you defeat in any sort of contest? The answer is context dependent, playstyle dependent, etc. Since the playout - in practice - will depend on random variables associated with individual players, scenarios, etc, then any attempt to do a mathematical analysis in a vacuum will have to assume distributions for those things at best, while any given situation, player, or table will have a specific value for those variables.

Edit: Noticed that you expanded the post while I was replying, so this response may be a bit stale. I'll wait for a response before extending so we don't just go back and forth with edits.

There is a lot of value in making things that are...hm. How to phrase it? Something like "incommensurable but of equal import," I guess? They are "of equal value" in a certain sense, in that you expend the same amount of resources on either. But they are "not of equal value" in that that abstract cost doesn't strictly define their relevance in every application. As an example, and not using 4e since I think I may have overused that thus far, consider the Dungeon World Paladin vs. Fighter. The Fighter gets Bend Bars, Lift Gates and Signature Weapon as its primary Starting Moves. Bend Bars, Lift Gates is for non-combat things, reflecting the Fighter's mighty thews and ability to break through barriers, prisons, and obstacles; Signature Weapon is, as I think the name implies, is primarily for combat stuff, giving you a fancy weapon with extra features. By comparison, the Paladin gets Quest, Lay on Hands, and I Am the Law. Quest lets the paladin set a quest for themselves (like "free X from the iniquities that beset them" or "slay X, a great blight on the land), picking two personal boons from a list (like "invulnerability to edged weapons" or "senses that pierce lies") and then receive from the DM one or more Vows (like "you can't lie" or "comfort those in need, no matter who they are"). Lay on Hands is a potentially self-sacrificial heal. I Am the Law lets you influence others, either obeying you, fleeing, or attacking you (but if you mess up, you're at a disadvantage against them).

These things cannot be strictly compared, because they don't all interface with the same stuff. However, one can see fairly well how they do fit into a sort of higher-order categorization: IAtL is for preventing combat or influencing others, Quest has a few defined uses, both combat- and non-combat-related, BBLG is primarily for exploration-related things, etc. IATL is more narrow in scope than BBLG, while Quest is broader, etc. We can see how there's some reasonable balance between these options, particularly since Signature Weapon is a very meaty feature, and combat is a common (and pretty dangerous) occurrence in Dungeon World. Are they balanced in every possible sense, all the time, always? No. Dungeon World isn't about that kind of balance. Instead, it's about balancing contribution to the advancing story. And all of these moves do that, to an extent that it fits a reasonable pattern. Paladins have moves that may be more powerful or versatile, but are likely to come with greater costs or commitments. Likewise, Fighters are more straightforward, but also freer. (This also doesn't take into account several Fighter moves that add versatility--such as tapping into the spirits tied to your Signature Weapon or literally being able to predict who/what WILL live and WILL die in any given scenario. Fighter has some very cool moves in DW.)

So...I guess what I'd say is, a good game is one where totally abstract reasoning about it doesn't necessarily tell you much. That is, it can do things like telling you that a certain feature or option is niche vs. common, it can tell you which things are risky vs. safe, it can reveal combinations that one wouldn't necessarily have considered (like the 4e barbarian/monk I mentioned before that can rocket-punch). But none of those statements cashes out to an absolute, unyielding, universal value. And that's what good optimization guides recognize, that context really does matter. Neither context nor abstraction matter unequivocally. A balanced game is one where there are multiple choices, and all of those choices both (a) have meaningful, non-obscure contexts in which they're relevant, and (b) good abstracted benefits that make them applicable even if you cannot know the context they'll be used in.

To use a radically different example, Fate is a game where context is always part of the design, and thus abstraction takes on a rather...different form. Because Aspects get created through DM/player interaction, the abstract analysis of Fate focuses on the characteristics a "good" Aspect has, and on teaching both players and DMs how to quickly and dynamically create/modify the available Aspects (both personal ones and environmental/situational ones) so that they aren't universal "I win" buttons, nor crazy-restrictive "I can compel this aspect on prime-numbered days if the moon is full and at least one player has a birthday this month" type stuff. Which is why I said Fate is so different from the other games I've mentioned, by balancing things to a radically different metric.

Good design balances these competing demands, between abstract universals (which is, technically, what the rules themselves are; all game rules are abstractions until someone chooses to play them) and practical effect (player experience). Good design doesn't get solved like an equation, but rather in an iterative, genetic-algorithm kind of way. Equally importantly, though, by having this iterative design process, you can communicate to your players, "This is what we think is important, and this is what we have provided to you to address those things." If the players wish to go beyond those rules, awesome, no problem. But the designer's job is to get the, well, design done. Stuff going above and beyond design is not the designer's concern, and often shouldn't be.

[NichG]

There is a lot of value in making things that are...hm. How to phrase it? Something like "incommensurable but of equal import," I guess? They are "of equal value" in a certain sense, in that you expend the same amount of resources on either. But they are "not of equal value" in that that abstract cost doesn't strictly define their relevance in every application. As an example, and not using 4e since I think I may have overused that thus far, consider the Dungeon World Paladin vs. Fighter. The Fighter gets Bend Bars, Lift Gates and Signature Weapon as its primary Starting Moves. Bend Bars, Lift Gates is for non-combat things, reflecting the Fighter's mighty thews and ability to break through barriers, prisons, and obstacles; Signature Weapon is, as I think the name implies, is primarily for combat stuff, giving you a fancy weapon with extra features. By comparison, the Paladin gets Quest, Lay on Hands, and I Am the Law. Quest lets the paladin set a quest for themselves (like "free X from the iniquities that beset them" or "slay X, a great blight on the land), picking two personal boons from a list (like "invulnerability to edged weapons" or "senses that pierce lies") and then receive from the DM one or more Vows (like "you can't lie" or "comfort those in need, no matter who they are"). Lay on Hands is a potentially self-sacrificial heal. I Am the Law lets you influence others, either obeying you, fleeing, or attacking you (but if you mess up, you're at a disadvantage against them).

These things cannot be strictly compared, because they don't all interface with the same stuff. However, one can see fairly well how they do fit into a sort of higher-order categorization: IAtL is for preventing combat or influencing others, Quest has a few defined uses, both combat- and non-combat-related, BBLG is primarily for exploration-related things, etc. IATL is more narrow in scope than BBLG, while Quest is broader, etc. We can see how there's some reasonable balance between these options, particularly since Signature Weapon is a very meaty feature, and combat is a common (and pretty dangerous) occurrence in Dungeon World. Are they balanced in every possible sense, all the time, always? No. Dungeon World isn't about that kind of balance. Instead, it's about balancing contribution to the advancing story. And all of these moves do that, to an extent that it fits a reasonable pattern. Paladins have moves that may be more powerful or versatile, but are likely to come with greater costs or commitments. Likewise, Fighters are more straightforward, but also freer. (This also doesn't take into account several Fighter moves that add versatility--such as tapping into the spirits tied to your Signature Weapon or literally being able to predict who/what WILL live and WILL die in any given scenario. Fighter has some very cool moves in DW.)

These are good examples.

Since you brought up AlphaGo in the previous post, I think it gives a concrete example of one kind of context appropriateness that's hard to model in advance, even in a very mechanical game.

AlphaGo uses early 3-3 invasions compared to what was considered best practice for professional play, and can quantify the expected advantage. So, players started to try it. For pros, some found ways to work it into their style successfully. For amateurs, playing the theoretically more optimal move discovered by AlphaGo generally made their play worse. So since this is Go, it led to a proverb - AlphaGo's moves are only good if you are as good a player as AlphaGo.

A more prosaic example is that starting an early fight is a stronger strategy if your personal reading skills are good enough, but building solid shape and making territorial exchanges for points is better than starting a fight you can't win if you can't read as well as your opponent (of course you need good positional judgement).

So even though Go is of a family of game that has a specific 'perfect play', Go strategies as implemented by humans or even AIs don't have a strict ranking, and can even have RPS like cycles of A beats B beats C beats A.

[Quizatzhaderac]

Because, me? I love that you can make 3e work even when you don't play it as the designers intended.

To me, that seems like saying that all Turing complete languages can complete the same tasks. Yes, but the obvious question is how difficult it is. If you like programming challenges creating a spreadsheet with OpenGL or creating a ASCII graphics engine with excel can be fun, but you should expect a few people to look at you like your crazy.

I would claim that it's possible to achieve a game to you and your friend's preferences much better than D&D 3.x (after controlling for your sunk time learning the system) by following Ezekielraiden's design methodology. One would picks different aesthetics , set tangible goals that exemplify those aesthetics, learn from past mistakes, and systemically test.

My goal was to make something to make a nice hot cup of tea.

If I made a tea kettle, or a coffee pot, or a Keurig thingy, they could all brew a cup of (sometimes very) hot tea. Or make coffee. And most could make hot chocolate or hot water.

But if I made a matter replicator, I could not just ask for "Earl grey, hot", but any other food (or any other anything) it's been programmed with.

Maybe people really will remember the matter replicator for hot tea. But me? I actively appreciate and admire the variety.

Actually, that metaphor can do a lot more work.

He had found a Nutri-Matic machine which had provided him with a plastic cup filled with a liquid that was almost, but not quite, entirely unlike tea. The way it functioned was very interesting. When the Drink button was pressed it made an instant but highly detailed examination of the subject’s taste buds, a spectroscopic analysis of the subject’s metabolism and then sent tiny experimental signals down the neural pathways to the taste centers of the subject’s brain to see what was likely to go down well. However, no one knew quite why it did this because it invariable delivered a cupful of liquid that was almost, but not quite, entirely unlike tea. The Nutri-Matic was designed and manufactured by the Sirius Cybernetics Corporation whose complaint department now covers all the major landmasses of the first three planets in the Sirius Tau Star system.

Would you consider it to be incoherent to set as a game design goal the creation of a game which is intentionally and actively resistant to mathematical testing? A game designed, for example, to make any sort of zoomed out analysis or theory-crafting as hard as possible (or to disadvantage it as much as possible) compared to behaviors determined organically during play?

I think that's somewhat confusing what "testing" means.

For examples, let's look at two simple games: tic-tack-toe and rock-paper-scissors.

tic-tack-toe is solved, which is to say that I can literally never be beaten or surprised when playing it.

rock-paper-scissors is proven unsolvable, which is to say that any strategy is always inferior to another strategy. But if we are developing a rock-paper-scissors like game we can still test that it meets certain goals. If we saw that rock always wins, we'd know we failed to reach one of our goals. We could also give a skilled minmaxer our ruleset, some time to analyze the game with spreadsheets, and test how strong that person's advantage is.

[NichG]

To me, that seems like saying that all Turing complete languages can complete the same tasks. Yes, but the obvious question is how difficult it is. If you like programming challenges creating a spreadsheet with OpenGL or creating a ASCII graphics engine with excel can be fun, but you should expect a few people to look at you like your crazy.

I would claim that it's possible to achieve a game to you and your friend's preferences much better than D&D 3.x (after controlling for your sunk time learning the system) by following Ezekielraiden's design methodology. One would picks different aesthetics , set tangible goals that exemplify those aesthetics, learn from past mistakes, and systemically test.

Actually, that metaphor can do a lot more work. I think that's somewhat confusing what "testing" means.

For examples, let's look at two simple games: tic-tack-toe and rock-paper-scissors.

tic-tack-toe is solved, which is to say that I can literally never be beaten or surprised when playing it.

rock-paper-scissors is proven unsolvable, which is to say that any strategy is always inferior to another strategy. But if we are developing a rock-paper-scissors like game we can still test that it meets certain goals. If we saw that rock always wins, we'd know we failed to reach one of our goals. We could also give a skilled minmaxer our ruleset, some time to analyze the game with spreadsheets, and test how strong that person's advantage is.

I'm trying to point out how framing design a certain way can create blind spots or things which, while they can be important to the experience of play, become harder to express in that paradigm.

If we take RPS for example, specifying that the intended win rate of rock, paper, and scissors should each be 1/3 and targeting that doesn't quite capture what makes RPS have the properties it does. A version of RPS where you achieve that through the statistics of dice rolls for example would just force all play to be equivalent to the random play fixed point. But RPS dynamics in, say, an evolutionary system can give rise to things like oscillations, spatial waves, etc which not all games with options with equal win rates will produce. In fact, the actual win rates of R,P,S will depend not only on the rules, but on the learning algorithm used by the players. It may be that, between two particular players, for some extended period of time, rock wins 70% of the time.

By specifying that the game should have no evolutionarily stable strategy or Nash equilibrium, we can get closer to it. But notice that has the format of the sort of adversarially designed approach I asked about: given that we can understand how games will be played by computing their Nash equilibria, let's intentionally find a game for which that method fails (in order to see the limits of an approach built on targeting Nash equilibria). You can often find the blind spots of a method by searching for something where the method fails but the thing can still be understood as nontrivial and nondegenerate.

There's analogies to this in computer vision and AI. One thing people might want is to produce realistic images. So the first thing people tried was to specify mathematical functions to capture what makes an image realistic and then optimize. But it turns out that it's pretty hard to specify that by hand. So then came along the idea of specifying it by asking the image to fool the best discriminator of real vs fake images you can discover, and making that discriminator based on your current best generator. Which can work, or can fail in all sorts of ways.

Interestingly to the subject of blind spots, a large portion of the field objected to the method on the basis that the dynamics don't do what the mathematical theory underpinning the method claimed. It designed the approach on the basis that the Nash equilibrium between the generator and discriminator is uniquely when the distribution of generated images is identical to the distribution of natural images. But that doesn't actually happen because there's no guarantee that the dynamics go to equilibrium. So is the method bad because it fails the quantifiable part of its design? Well, despite that it can still generate photorealistic images. Which suggests that the explicitly quantified part of the method isn't actually capturing the essential things about generating realistic images. But because a portion of the field wants to discuss things in a certain mathematical framework, there's tension over the validity of the technique.

To bring this back to point, there's been a lot of circularity in the discussion. I think this is at least in part caused by the mathematical testing and targeting framework being ill suited to capture some of the things that e.g. Quertus or I might want to design a game for. I'm pushing back against the idea that the difficulty of capturing these things mathematically makes them not be valid game design goals.

For example 'I want it to be difficult to predict what playing the game will be like until you actually sit down and play'

[ezekielraiden]

For example 'I want it to be difficult to predict what playing the game will be like until you actually sit down and play'

How do you check to see if you've accomplished that?

Perhaps a better question: Does D&D have something actually equivalent to "only do X if you're a really good player; do Y because you'll win more if you aren't as good of a player yet"? Because...that's got some big holes I could shoot at for the analogy. Like how it's competitive and D&D isn't, in fact D&D is cooperative, or how D&D is open-ended and non-zero-sum whereas Go is zero-sum.

I certainly grant that some players will struggle with some options. Conjurer Wizards in 3.5e are crazy powerful...and extremely slow except in well-prepared hands. That doesn't require in-play experience to see, since running a summoned creature *and* your character should, rationally, take more time than just running your character. Similarly, spells like illusions and suggestion depend on how receptive the DM is to creative uses. Again, this requires no direct experience to understand. I've never cast an illusion or suggestion spell myself, ever, in any 3rd edition (or spinoff) game, yet I can see at a glance how illusions will depend on the DM "playing along," like how some strategy game AI is designed to experience (or fail to experience) limited knowledge about player choices.

Edit: When I talk about things like statistical measures (center, spread, skew, etc.), I'm usually referring to things like "does the rate/distribution of successfully-landed attacks match what we want for a fun but not trivial experience?" (4e goes for a ~65%-70% success rate on "normal" challenges: more than slighly lower and it starts to feel like you fail too often at what you're "good at," much higher and you make failure too rare and squeeze out the space for truly puissant characters.) High-level or more abstract strategy may require more creative testing. For example, I can say with certainty that 4e successfully delivers on "you are a TEAM, you HAVE to fight as a TEAM or you will LOSE." By comparison, 3e feels like a game where four or five people just coincidentally *happen* to adventure in identical places at the same time, and the best way to be a "team player" is always to be maximally selfish and maximally self-optimized even if it means being less supportive to your allies.

As an entirely non-TTRPG example: Final Fantasy XIV introduced the Dancer class with the newest expansion, a class that intentionally gas very low personal DPS but really really good buffs for allies, and this has altered how the community evaluates data. Now, instead of just examining raw DPS, the fan-made tracker program (which is pretty sophisticated!) tracks "rDPS" (how much damage was dealt "because of you"--so YOU get the credit for any group/ally buffs you hand out) and "aDPS" (which is similar to raw values, but factors out certain single-target buffs). Under this new metric, while Samurai, Monk, and Black Mage continue to be the top 3 classes for aDPS, the more support-leaning Dragoon shoots up to 3rd place in rDPS for some fights, and even takes 1st in one, because it has really valuable team support on top of its solid base damage. Meaning, even though the old metrics had to be changed to account for a shift, the new metrics now capture factors that were either only inferred, or ignored, before. Seeing Dragoon shoot up 2-4 ranks because of its good utility is proof positive that Dragoon has solid design. (It also does things like pushing Ninja up significantly, since it *also* has really strong utility, and shows that the utility Bard *lost* this expac was probably excessive.)

IOW/TL;DR: Picking the right method of data-gathering is vital. You need to ask good questions to get good answers. That's part of why it's such a shame so few game designers have statistics or mathematics backgrounds. They are poorly equipped to think about how to change the questions they ask effectively.

[NichG]

How do you check to see if you've accomplished that?

High rate of change of player playstyle and rapid increase in player confidence during the period of introduction to play. Reported dissatisfaction or lack of interest in second hand play information or 'build guides'. Strong sense of alienation when a player transfers from one table running the game to another. Inability for players to accurately describe their experiences with the game, or to agree with eachother about the nature of their experiences.

Perhaps a better question: Does D&D have something actually equivalent to "only do X if you're a really good player; do Y because you'll win more if you aren't as good of a player yet"? Because...that's got some big holes I could shoot at for the analogy. Like how it's competitive and D&D isn't, in fact D&D is cooperative, or how D&D is open-ended and non-zero-sum whereas Go is zero-sum.

I don't think bringing D&D into it is a better question, because of strong preexisting feelings about D&D on all sides. If opinions aren't stable when transferring to more neutral games, that's evidence that they have more to do with prior bias than the argument at hand.

I'm asking, in the abstract, if you could consider this to be a coherent design goal. I'm not claiming D&D satisfies that design goal or was designed with it in mind. I'd say in fact that a big defining characteristic of the feel of 3ed compared to earlier editions is that theory crafting became significantly more viable and effective in 3ed. So if anything, it moved away from this particular target.

[Mechalich]

I'm asking, in the abstract, if you could consider this to be a coherent design goal. I'm not claiming D&D satisfies that design goal or was designed with it in mind. I'd say in fact that a big defining characteristic of the feel of 3ed compared to earlier editions is that theory crafting became significantly more viable and effective in 3ed. So if anything, it moved away from this particular target.

It's a coherent goal, however I don't think it's a desirable goal. In fact I very much think the opposite is true, that you want it to be easy to predict what the game will be like before sitting down because that way you'll be far more likely to join or purchase a game that you actually enjoy. Now, obviously this won't work for generic or toolkit systems like GURPS or FATE which aren't actually complete games so much as they are shared methodologies by which a GM assembles a game, but for games that make a claim - often through fluff material - about what their game is, it's important that they deliver. Historically games that don't provide the gameplay experience they claim to deliver have problems with angry fans. Now, creative tables often get around those problems through copious house-ruling and subconsciously ignoring, disregarding, or otherwise pretending portions of the rules don't actually exist (to reiterate an example I've mentioned before, in many 3e experiences grapples just never occur), but if tables are doing this that's evidence that parts of your design are actively detrimental to the gameplay experience you intended to create.

As far as balance goes, the extension of this is that if a system is not intended to be balanced mechanically and pushes all duties to manage contributions onto the GM then it needs to be upfront about this. Additionally, the game should make it clear what sort of mechanical balance methods it's using. Some games have very strong niche protection (this approach is better suited to futuristic games where character skills are presumed to be very specialized, such as Eclipse Phase), while others stressed a generalized ability to contribute to everything. The game should also very clearly outline what the core experience is expected to be so that every character is able to participate meaningfully in that experience. For example, L5R characters are extremely oriented towards a particular playstyle, and a new GM needs to understand that only certain character types are of any use in certain campaigns. More broadly, if a game is going to involve combat at all, every character needs to be able to contribute in combat in some way (they don't need to be good at it, but they need to have character options beyond 'run and hide' even if that would be a realistic response).

[ezekielraiden]

High rate of change of player playstyle and rapid increase in player confidence during the period of introduction to play. Reported dissatisfaction or lack of interest in second hand play information or 'build guides'. Strong sense of alienation when a player transfers from one table running the game to another. Inability for players to accurately describe their experiences with the game, or to agree with eachother about the nature of their experiences.

That...sounds like an extremely tall order, and probably not a healthy thing for any real game. That is, if you can't describe your experience or even agree with other people who also played it, how can you communicate to someone else that you enjoyed it? You seem to be describing something that is so radically personal, it defies any description at all, which would seem to make playing it pretty difficult too? I mean, if you can't describe your experience, how do you describe how to play it?

I don't think bringing D&D into it is a better question, because of strong preexisting feelings about D&D on all sides. If opinions aren't stable when transferring to more neutral games, that's evidence that they have more to do with prior bias than the argument at hand.

I'm asking, in the abstract, if you could consider this to be a coherent design goal. I'm not claiming D&D satisfies that design goal or was designed with it in mind. I'd say in fact that a big defining characteristic of the feel of 3ed compared to earlier editions is that theory crafting became significantly more viable and effective in 3ed. So if anything, it moved away from this particular target.

I'm genuinely hesitant to say that it's totally impossible but...particularly with the above clarification, the more I read, the more I think "no, it's not possible." What you describe sounds not only impossible to reflect upon, but impossible to describe beyond "it cannot be described." It's...sort of like asking if it's possible to make a language where it's impossible to define any word in terms of any other combination of words. How would you teach someone that language? How would you check that someone else knew what it meant? It seems incoherent, not strictly for the non-testability per se, but because you're talking about a game that cannot be discussed at all, which seems to preclude discussing it to show someone else how to play it.

It's like Calvinball, but worse. With Calvinball, the idea is that each player actively participates in creating the game each time. Though no two games will end up the same, it's still possible to discuss the process of play, the experience of crafting the rules, and strategies for creating win-promoting rules (even if that's a very difficult task). Your "totally un-analyzable game" (call it TUG) lacks that. It has no (after-action) describable process of play. It has no strategies (because if a strategy exists, it can be articulated--a strategy is a procedure, after all). There can be no meaningful sense of getting "better" at the game, either, because doing so requires that you can articulate whether your experience was successful or not.

So...yeah, the more I think about it, the more I think this "goal" is genuinely incoherent--and may even be a logically impossible thing. A game has rules, and multiple possible outcomes. Things that have no rules (which does not include games like Nomic or Mao--that is, games where the point is to make up new rules) aren't games in the first place. Things that have rules, but only one valid solution-path to a fixed objective, are puzzles rather than games. (This does admit a grey area for "solved" games like tic-tac-toe/noughts-and-crosses, but either it's a puzzle or it's a game, that much remains true.) I'm not sure how it's possible to have (a) real, defined rules; (b) multiple real/valid solutions; and (c) complete inability to discuss the procedure or experience of play. It seems as though the first two guarantee, to at least some minimal degree, that the third is false--that there is some amount of strategy that can be applied, no matter how small, no matter how narrow in impact.

The closest I could come to something even like this (which, I admit, risks a lack-of-imagination fallacy) is a game where the rules have to be communicated in a specific sensory way (such as sight, or sound, or telepathy), which can then be lost by the player after they play. But even that seems to fail to meet your very strict requirements, since the game can be communicated, it just requires a sense that the player can gain and then subsequently lose, and thus is only "impossible to communicate" only to those who lack the qualia for that sense. I'm really struggling to see how it is possible to teach/show someone how to play the game, while also being unable to tell people what it was like to actually play. And that's not even considering complications like "how do people know they're playing by the rules, since they cannot communicate any of their prior experience of the game to anyone else?"

[NichG]

That...sounds like an extremely tall order, and probably not a healthy thing for any real game. That is, if you can't describe your experience or even agree with other people who also played it, how can you communicate to someone else that you enjoyed it? You seem to be describing something that is so radically personal, it defies any description at all, which would seem to make playing it pretty difficult too? I mean, if you can't describe your experience, how do you describe how to play it?

I'm genuinely hesitant to say that it's totally impossible but...particularly with the above clarification, the more I read, the more I think "no, it's not possible." What you describe sounds not only impossible to reflect upon, but impossible to describe beyond "it cannot be described." It's...sort of like asking if it's possible to make a language where it's impossible to define any word in terms of any other combination of words. How would you teach someone that language? How would you check that someone else knew what it meant? It seems incoherent, not strictly for the non-testability per se, but because you're talking about a game that cannot be discussed at all, which seems to preclude discussing it to show someone else how to play it.

It's like Calvinball, but worse. With Calvinball, the idea is that each player actively participates in creating the game each time. Though no two games will end up the same, it's still possible to discuss the process of play, the experience of crafting the rules, and strategies for creating win-promoting rules (even if that's a very difficult task). Your "totally un-analyzable game" (call it TUG) lacks that. It has no (after-action) describable process of play. It has no strategies (because if a strategy exists, it can be articulated--a strategy is a procedure, after all). There can be no meaningful sense of getting "better" at the game, either, because doing so requires that you can articulate whether your experience was successful or not.

So...yeah, the more I think about it, the more I think this "goal" is genuinely incoherent--and may even be a logically impossible thing. A game has rules, and multiple possible outcomes. Things that have no rules (which does not include games like Nomic or Mao--that is, games where the point is to make up new rules) aren't games in the first place. Things that have rules, but only one valid solution-path to a fixed objective, are puzzles rather than games. (This does admit a grey area for "solved" games like tic-tac-toe/noughts-and-crosses, but either it's a puzzle or it's a game, that much remains true.) I'm not sure how it's possible to have (a) real, defined rules; (b) multiple real/valid solutions; and (c) complete inability to discuss the procedure or experience of play. It seems as though the first two guarantee, to at least some minimal degree, that the third is false--that there is some amount of strategy that can be applied, no matter how small, no matter how narrow in impact.

The closest I could come to something even like this (which, I admit, risks a lack-of-imagination fallacy) is a game where the rules have to be communicated in a specific sensory way (such as sight, or sound, or telepathy), which can then be lost by the player after they play. But even that seems to fail to meet your very strict requirements, since the game can be communicated, it just requires a sense that the player can gain and then subsequently lose, and thus is only "impossible to communicate" only to those who lack the qualia for that sense. I'm really struggling to see how it is possible to teach/show someone how to play the game, while also being unable to tell people what it was like to actually play. And that's not even considering complications like "how do people know they're playing by the rules, since they cannot communicate any of their prior experience of the game to anyone else?"

Actually, you've given several examples here that do move in the direction of this design goal. Something like Nomic lies much closer to the extreme point than most tabletop games, and tabletop RPGs lie closer to the extreme point than, say, perfect information deterministic zero sum games that admit provable perfect play.

A practical example of this kind of goal would be in the design of the parts of IQ tests to measure learning capabilities - I want a metric that is static per individual, cannot be improved by study, and measures how well they adapt new skills. So I need to make a test that has a high probability of hitting some combination of skills that a test-taker doesn't yet have but could have. Plus it should be robust against them talking with people who have taken the test in the past. So, among other things, I will likely design the test to change over time.

Interview questions for programming positions are also somewhat subject to this dynamic - how to make it test the applicant's skills rather than the applicant's ability to study up on interview questions?

I also kind of find it funny that many of those criteria could apply to the experience of human life. Switch me into someone else's body and I'll likely feel pretty alienated. There are life experiences where hearing others describe them is no substitute for living it yourself. Life planning advice ('build guides') is often incompatible with variations from person to person, and blindly imitating someone else's career path or lifestyle can be problematic. It's not at the theoretical extreme point of 'describing your experiences is impossible' but philosophers and artists study for years to be able to do it effectively. If we take this as an axis rather than a single destination, I'd say life is further along that axis than most games we make.

[Cluedrew]

It's already been mentioned in this thread (by Cluedrew I think?) that one big principle of designing balanced sets of abilities is that you try to make the abilities work in such a way that they're not comparable to each-other in a vacuum.

Yeah I talk about that. The more variation between abilities you have the harder it is to draw comparisons between them. You have to know more and more about the situation to say which one is better. Plus it becomes more a matter of taste at a certain point. There are certain archetypes (of characters, builds, decks, armies) that I am just not as interested in and might not use even if they are "better". Which is where balanced as being a larger window comes in.

[ezekielraiden]

A practical example of this kind of goal would be in the design of the parts of IQ tests to measure learning capabilities - I want a metric that is static per individual, cannot be improved by study, and measures how well they adapt new skills. So I need to make a test that has a high probability of hitting some combination of skills that a test-taker doesn't yet have but could have. Plus it should be robust against them talking with people who have taken the test in the past. So, among other things, I will likely design the test to change over time.

But is an IQ test a game? I would argue it's not. As you yourself noted, once you know the answer to a question, it ceases to be useful as a test--because it's a puzzle, that's what gives it value as a metric of intelligence. You create a large set of puzzles, of varying difficulty, in order to see how many a typical test-taker can answer, perhaps factoring in timing as well. So--can you make a game, something with a large number of end-states, that fits this? I'm inclined to say "no"--things we recognize as robust against being "solved," like chess, go, poker, etc., don't have that "if you already did it once, you know it and it loses its utility" nature, key to both puzzles and tests (and riddles).

Interview questions for programming positions are also somewhat subject to this dynamic - how to make it test the applicant's skills rather than the applicant's ability to study up on interview questions?

Seems to suffer the same problem: you're trying to make a test, so it's an arms race between question-design to keep interviewees on their toes, and answer-tailoring to guarantee interviewer satisfaction without necessarily having the sought qualities. Such arms-race design is, again, typical of tests and pretty much unheard-of in anything people would call "games." The closest I can think of is how people can, and will, seek out exploits (particularly in video games, where they're harder to patch up), but even there you don't have that "if you played it once, you already know exactly what will happen" nature that affects puzzles and tests.

I also kind of find it funny that many of those criteria could apply to the experience of human life. Switch me into someone else's body and I'll likely feel pretty alienated. There are life experiences where hearing others describe them is no substitute for living it yourself. Life planning advice ('build guides') is often incompatible with variations from person to person, and blindly imitating someone else's career path or lifestyle can be problematic. It's not at the theoretical extreme point of 'describing your experiences is impossible' but philosophers and artists study for years to be able to do it effectively. If we take this as an axis rather than a single destination, I'd say life is further along that axis than most games we make.

Life also isn't a designed game? It certainly doesn't have a readily-to-hand set of rules before you get started (well, I believe it does have such rules, but they aren't ready-to-hand or we wouldn't have been debating moral philosophy for 2500+ years). And, more to the point, we absolutely can discuss, analyze, and strategize. That's why there are things like psychiatry, self-help books, schools, (auto)biographies, etc. Can we articulate the complete experience in absolutely all detail, just as if we all lived it ourselves? No. But the criterion wasn't that we can only incompletely discuss, only limitedly strategize. It was that we cannot strategize, cannot discuss, even in principle. Hence why I referred to things like sensory-based stuff. I'm not sure a blind person, for example, would ever really get the concept of something like Simon, the four-color thing, because it so thoroughly depends on color. Or, for perhaps a more direct example, I strongly suspect a blind person would be very, very confused by how it could possibly be hard to identify the color a word is written with, when that color isn't the same as the word you're reading.

I hear your point that you mean it more as a spectrum, but it seems to me that even getting a lot in that direction rapidly reduces the game-ness of a thing. Tests and life are, I agree, closer to that extreme anti-strategic, anti-discussion end. But I would challenge you to give any examples of something that's actually a game and yet is even as "strategy-resistant" as a test. Hence why I am slowly overcoming my reluctance to making strong negative claims. If it is possible to create such a game, I would be pretty surprised at this point. It seems to be a genuinely incoherent goal--though, and this is the very interesting part, it doesn't seem to be obviously one. It doesn't seem like "triangle with two sides," and yet it does seem like a logical contradiction...which means it should seem like "triangle with two sides." And that's fascinating!

Yeah I talk about that. The more variation between abilities you have the harder it is to draw comparisons between them. You have to know more and more about the situation to say which one is better. Plus it becomes more a matter of taste at a certain point. There are certain archetypes (of characters, builds, decks, armies) that I am just not as interested in and might not use even if they are "better". Which is where balanced as being a larger window comes in.

It's worth noting, this is why I have been careful to say that there will always be people who make use of dominant/degenerate strategies, rather than that absolutely everyone will use them. Taste does matter, and people really do choose to avoid or not avail themselves of advantages for a variety of reasons. I mean, I already pointed out how infinite money (or at least being able to squeeze out a pretty significant profit, even if not infinitely) is present in 3.5e's rules, and yet most people who play it have never made use of this exploit. Whether it's taste, propriety, desiring a challenge, personal characteristics (e.g. someone with poor multi-tasking skills would probably not play a Conjurer in 3.5e, despite it being possibly the strongest choice), or a million other things, dominant/degenerate strategies are not guaranteed to be used by absolutely everyone. They are, however, extremely likely to be used by a significant number of people.

[NichG]

But is an IQ test a game? I would argue it's not. As you yourself noted, once you know the answer to a question, it ceases to be useful as a test--because it's a puzzle, that's what gives it value as a metric of intelligence. You create a large set of puzzles, of varying difficulty, in order to see how many a typical test-taker can answer, perhaps factoring in timing as well. So--can you make a game, something with a large number of end-states, that fits this? I'm inclined to say "no"--things we recognize as robust against being "solved," like chess, go, poker, etc., don't have that "if you already did it once, you know it and it loses its utility" nature, key to both puzzles and tests (and riddles).


Seems to suffer the same problem: you're trying to make a test, so it's an arms race between question-design to keep interviewees on their toes, and answer-tailoring to guarantee interviewer satisfaction without necessarily having the sought qualities. Such arms-race design is, again, typical of tests and pretty much unheard-of in anything people would call "games." The closest I can think of is how people can, and will, seek out exploits (particularly in video games, where they're harder to patch up), but even there you don't have that "if you played it once, you already know exactly what will happen" nature that affects puzzles and tests.


Life also isn't a designed game? It certainly doesn't have a readily-to-hand set of rules before you get started (well, I believe it does have such rules, but they aren't ready-to-hand or we wouldn't have been debating moral philosophy for 2500+ years). And, more to the point, we absolutely can discuss, analyze, and strategize. That's why there are things like psychiatry, self-help books, schools, (auto)biographies, etc. Can we articulate the complete experience in absolutely all detail, just as if we all lived it ourselves? No. But the criterion wasn't that we can only incompletely discuss, only limitedly strategize. It was that we cannot strategize, cannot discuss, even in principle. Hence why I referred to things like sensory-based stuff. I'm not sure a blind person, for example, would ever really get the concept of something like Simon, the four-color thing, because it so thoroughly depends on color. Or, for perhaps a more direct example, I strongly suspect a blind person would be very, very confused by how it could possibly be hard to identify the color a word is written with, when that color isn't the same as the word you're reading.

I hear your point that you mean it more as a spectrum, but it seems to me that even getting a lot in that direction rapidly reduces the game-ness of a thing. Tests and life are, I agree, closer to that extreme anti-strategic, anti-discussion end. But I would challenge you to give any examples of something that's actually a game and yet is even as "strategy-resistant" as a test. Hence why I am slowly overcoming my reluctance to making strong negative claims. If it is possible to create such a game, I would be pretty surprised at this point. It seems to be a genuinely incoherent goal--though, and this is the very interesting part, it doesn't seem to be obviously one. It doesn't seem like "triangle with two sides," and yet it does seem like a logical contradiction...which means it should seem like "triangle with two sides." And that's fascinating!

It's kind of the point of the exercise. Game-like or not, if there are things which people do design towards (tests) or might want to design towards (making the experience of a game like the experience of life) which can't be discussed coherently within a particular design framework, then that helps identify the limitations of that framework (and hopefully helps explain why one might choose to reject it).

I'd say you've already identified several design strategies for landing in this space - precisely some of the things you list that makes things feel un-gamelike to you. I'll focus on one briefly - rules uncertainty.

For my tabletop campaigns over the last 10 years or so, both in games I've played in and games I've run, a common thread is that there have been unstated rules which must be either uncovered or created during play. The primary examples in ones I've played: combinatoric powers alchemy where the new power you get by fusing others results from a kind of portmanteau word game; a lair with unlockable 'buildings' which each introduced new game mechanics or forms of advancement that the players didn't get to know in advance; a system where all elements of the natural world had unlockable magical potential that had to be discovered by experimentation in character; a mechanic that let players alter the cosmology of the setting and could result in the wholesale creation of new kinds of supernatural creatures with their own mechanics. In my own games: a civilization-building campaign where the players choose what technologies to invent (and which aren't in the rules ahead of time), a game about the afterlife where characters can install arbitrary significant memories or themes into a set of slots in their soul and gain unique abilities as they invest in the memories; a game about superheroes where characters start with almost literally infinite power and the primary form of advancement is the fact that most players will not immediately feel comfortable acting at that scale; currently in design, a game about invention and discovery, where the main mechanic is that each player gets an income of points that can be spent to lock adhoc rulings into the rules.

So, since everything I play is basically some form of nomic, design frameworks that conclude that e.g. these are not games or that they're not coherently designable are of limited utility to me, because it's hard to even discuss my goals in those frameworks without creating the equivalent of imaginary numbers or Lovecraftian geometry.

[ezekielraiden]

It's kind of the point of the exercise. Game-like or not, if there are things which people do design towards (tests) or might want to design towards (making the experience of a game like the experience of life) which can't be discussed coherently within a particular design framework, then that helps identify the limitations of that framework (and hopefully helps explain why one might choose to reject it).

I'd say you've already identified several design strategies for landing in this space - precisely some of the things you list that makes things feel un-gamelike to you. I'll focus on one briefly - rules uncertainty.

I don't see rules uncertainty as un-game-like, per se. Uncovering the rules is a gameable experience, one I have personally enjoyed as well. (The flash "God Game Thing is? It's also an experience that admits description and strategy. Because the rules either already exist to be discovered (and thus admit strategies to find them), or they are determined by some consistent formula, such that the same combinations will consistently produce the same outcomes (and thus admit strategies that apply to any combinatoric exercise).

What I see as un-game-like is there not being any rules at all, unless arbitrarily projected there, hence why I mentioned life doesn't have a rulebook you receive at birth. It's not that rules cannot be projected onto it, they absolutely can. It's that those rules are purely elective and projective in nature: the physical world itself contains no justice, no mercy, not one atom of kindness. Life just is not a game--you can't "win" it or even "play" it except by meeting conditions you arbitrarily applied, and why anyone should value one condition over another is, as you said, a complex subject about which there is dramatic disagreement. You can certainly make games intending to capture certain aspects of that, but that gets us into the map/territory distinction: a game like life is not life, it's just an intended approximation.

As a clarification: a game with rules for changing its rules (including changing how the rules are changed) is not "arbitrarily projected." They were chosen and/or agreed upon at the outset, and then play itself changed them after. With life, there is no "chosen/agreed upon at the outset." We are thrust into life without our permission and removed from it without our consent.

For my tabletop campaigns over the last 10 years or so, both in games I've played in and games I've run, a common thread is that there have been unstated rules which must be either uncovered or created during play. The primary examples in ones I've played: combinatoric powers alchemy where the new power you get by fusing others results from a kind of portmanteau word game; a lair with unlockable 'buildings' which each introduced new game mechanics or forms of advancement that the players didn't get to know in advance; a system where all elements of the natural world had unlockable magical potential that had to be discovered by experimentation in character; a mechanic that let players alter the cosmology of the setting and could result in the wholesale creation of new kinds of supernatural creatures with their own mechanics. In my own games: a civilization-building campaign where the players choose what technologies to invent (and which aren't in the rules ahead of time), a game about the afterlife where characters can install arbitrary significant memories or themes into a set of slots in their soul and gain unique abilities as they invest in the memories; a game about superheroes where characters start with almost literally infinite power and the primary form of advancement is the fact that most players will not immediately feel comfortable acting at that scale; currently in design, a game about invention and discovery, where the main mechanic is that each player gets an income of points that can be spent to lock adhoc rulings into the rules.

So, since everything I play is basically some form of nomic, design frameworks that conclude that e.g. these are not games or that they're not coherently designable are of limited utility to me, because it's hard to even discuss my goals in those frameworks without creating the equivalent of imaginary numbers or Lovecraftian geometry.

Again, I'm not saying that a nomic isn't a game. I'm saying that the part of them that captures what you spoke of--being actively inimical to strategy--is the part of them that becomes non-game-like. It's sort of like how the Prisoner's Dilemma isn't a game, but the Iterated Prisoner's Dilemma IS a game that can be subject to strategy and can be analyzed such. A game where rules are unknown but concrete, or not-currently-existent but following a concrete predefined procedure to become existent, is still a game, and still admits testing (albeit in ways radically different from D&D, since there are very few grounding assumptions).

Nomic, for example, could have been a really fun game that just happened to have rules for modifying its own rules. But the designer had a concrete goal: make the act of rule-making the primary focus of play. That's a testable goal, as you can look at the behaviors players choose to engage in and modify the rules to increase (or decrease) the rate of engaging in one or more behaviors. In this case, they very intentionally made the "default" rules of Nomic simplistic and boring, so that players would be tempted to alter the rules to make something more entertaining. That's a testable objective derived from a concrete goal, about the game I allegedly cannot analyze under this model. Of course I cannot analyze all possible games of Nomic, since "all possible games of Nomic" is at least no smaller than the set of all possible games period, and that's way beyond "too big to analyze." But that doesn't mean one cannot be "good at Nomic," that there can't be strategies for Nomic (even if they are highly unlike most game strategy), and above all else, it doesn't mean you cannot communicate to me the experience of playing the game.

I mean...with every single example you've given, I have gotten what I consider a very concise, yet also highly informative, description of several gaming environments where strategic thinking primarily takes the form of creativity and analysis is primarily open-mindedness. You have meaningfully communicated to me, rather than how one plays, at least starting insights into how one might play effectively or well, despite having specifically stated zero rules. Does that not mean we are discussing the game abstractly? Are we not discussing your experiences in a way that is meaningful to both you and me?

These games don't just sound like they fall short of "completely impossible to discuss the experience of play," they sound rather easy to discuss so, given the facility with which you have done so. Unless you mean to say that everything in your examples is really bad at actually telling me anything about the games in question?

[NichG]

I don't see rules uncertainty as un-game-like, per se. Uncovering the rules is a gameable experience, one I have personally enjoyed as well. (The flash "God Game Thing is? It's also an experience that admits description and strategy. Because the rules either already exist to be discovered (and thus admit strategies to find them), or they are determined by some consistent formula, such that the same combinations will consistently produce the same outcomes (and thus admit strategies that apply to any combinatoric exercise).

What I see as un-game-like is there not being any rules at all, unless arbitrarily projected there, hence why I mentioned life doesn't have a rulebook you receive at birth. It's not that rules cannot be projected onto it, they absolutely can. It's that those rules are purely elective and projective in nature: the physical world itself contains no justice, no mercy, not one atom of kindness. Life just is not a game--you can't "win" it or even "play" it except by meeting conditions you arbitrarily applied, and why anyone should value one condition over another is, as you said, a complex subject about which there is dramatic disagreement. You can certainly make games intending to capture certain aspects of that, but that gets us into the map/territory distinction: a game like life is not life, it's just an intended approximation.

As a clarification: a game with rules for changing its rules (including changing how the rules are changed) is not "arbitrarily projected." They were chosen and/or agreed upon at the outset, and then play itself changed them after. With life, there is no "chosen/agreed upon at the outset." We are thrust into life without our permission and removed from it without our consent.

I'm not sure whether something is a game under a particular formal sense is all that relevant, at least to my goals. If my design goal is better satisfied by a non-game, that's not actually a reason to abandon either that direction of design, or the goal.

Puzzle games may not be game theoretic games, but that doesn't prevent people from playing them...

I can definitely see 'I want to design a group activity, the experience of which tries to be as rich and varied as real life while yet being distinct from it' as an interesting and worthy goal for a tabletop RPG. Even if the result ends up not technically being a game.

Again, I'm not saying that a nomic isn't a game. I'm saying that the part of them that captures what you spoke of--being actively inimical to strategy--is the part of them that becomes non-game-like. It's sort of like how the Prisoner's Dilemma isn't a game, but the Iterated Prisoner's Dilemma IS a game that can be subject to strategy and can be analyzed such. A game where rules are unknown but concrete, or not-currently-existent but following a concrete predefined procedure to become existent, is still a game, and still admits testing (albeit in ways radically different from D&D, since there are very few grounding assumptions).

Nomic, for example, could have been a really fun game that just happened to have rules for modifying its own rules. But the designer had a concrete goal: make the act of rule-making the primary focus of play. That's a testable goal, as you can look at the behaviors players choose to engage in and modify the rules to increase (or decrease) the rate of engaging in one or more behaviors. In this case, they very intentionally made the "default" rules of Nomic simplistic and boring, so that players would be tempted to alter the rules to make something more entertaining. That's a testable objective derived from a concrete goal, about the game I allegedly cannot analyze under this model. Of course I cannot analyze all possible games of Nomic, since "all possible games of Nomic" is at least no smaller than the set of all possible games period, and that's way beyond "too big to analyze." But that doesn't mean one cannot be "good at Nomic," that there can't be strategies for Nomic (even if they are highly unlike most game strategy), and above all else, it doesn't mean you cannot communicate to me the experience of playing the game.

I mean...with every single example you've given, I have gotten what I consider a very concise, yet also highly informative, description of several gaming environments where strategic thinking primarily takes the form of creativity and analysis is primarily open-mindedness. You have meaningfully communicated to me, rather than how one plays, at least starting insights into how one might play effectively or well, despite having specifically stated zero rules. Does that not mean we are discussing the game abstractly? Are we not discussing your experiences in a way that is meaningful to both you and me?

These games don't just sound like they fall short of "completely impossible to discuss the experience of play," they sound rather easy to discuss so, given the facility with which you have done so. Unless you mean to say that everything in your examples is really bad at actually telling me anything about the games in question?

In the context of your Prisoner's Dilemma vs IPD example, I would say that my description of Limit Break for example (the overpowered superhero game) is like me describing a single round of PD. We can agree on the events that happened and maybe even what it was like, but the next round - another campaign in that system with different players - is highly likely to be extremely different. That's because the system is intentionally incomplete in a way that completes itself using the psychology and interests of the players. One could strategize in the abstract, but without knowing how the other players are going to play those strategies would have a very short half life.

In the game I ran, every player chose highly abstract power themes with (conflicting) cosmological implications. Gameplay often revolved around literally editing the meanings of concepts in order to undermine crowd behaviors that had gained sentience.

Next game, someone could decide they want time travel powers, and whatever conclusions we drew might go up in flames.

The concept of the system itself doesn't even really work if the same player plays twice, because advancement is at least in large part player-side. It's like taking the same IQ test twice.

[ezekielraiden]

I'm not sure whether something is a game under a particular formal sense is all that relevant, at least to my goals. If my design goal is better satisfied by a non-game, that's not actually a reason to abandon either that direction of design, or the goal.

Okay. I appreciate that thread drift and stuff happens, but I guess I'm coming at this from, "How is this relevant to the thread topic?" We already answered the original question--"need" is too strong a word without any context--and had, as I understood it, moved on to the more useful, "What does 'balance' mean for the D&D context, and how necessary is it there?" Talking about abstract goals that may go entirely outside of gaming and game design isn't bad, but it seems a spinoff topic, at least to me.

Puzzle games may not be game theoretic games, but that doesn't prevent people from playing them...

Well, I'd argue that a "puzzle game" always has one of two features: either it is a puzzle generator, and thus open-ended rather than closed-ended like a proper singular puzzle is, or it is a significantly long puzzle sequence, so that it comes more closely to resemble the IPD (that is, play is about learning to solve puzzles quickly, or about building up a skillset to identify key information, etc.) In other words, anything that a typical person would call a "puzzle game" needs that iterative component; otherwise it would just be a stand-alone puzzle.

I can definitely see 'I want to design a group activity, the experience of which tries to be as rich and varied as real life while yet being distinct from it' as an interesting and worthy goal for a tabletop RPG. Even if the result ends up not technically being a game.

Well, I mean, then it's not really a goal for a tabletop RPG, which was sort of my point. I'm interested in talking about games, since this is the Roleplaying Games forum, and our focus topic is balance (with the more-refined "what is balance, and how much is worthwhile?" question).

In the context of your Prisoner's Dilemma vs IPD example, I would say that my description of Limit Break for example (the overpowered superhero game) is like me describing a single round of PD. We can agree on the events that happened and maybe even what it was like, but the next round - another campaign in that system with different players - is highly likely to be extremely different. That's because the system is intentionally incomplete in a way that completes itself using the psychology and interests of the players. One could strategize in the abstract, but without knowing how the other players are going to play those strategies would have a very short half life.

In the game I ran, every player chose highly abstract power themes with (conflicting) cosmological implications. Gameplay often revolved around literally editing the meanings of concepts in order to undermine crowd behaviors that had gained sentience.

Next game, someone could decide they want time travel powers, and whatever conclusions we drew might go up in flames.

The concept of the system itself doesn't even really work if the same player plays twice, because advancement is at least in large part player-side. It's like taking the same IQ test twice.

Then...yeah I guess I just don't consider that a game. If you play it twice and necessarily generate effectively identical outcomes....it's not a game. More like a really entertaining personality test or self-introspection questionnaire? Regardless, though, while it is a worthy subject of discussion, I just...don't really have anything to say about it in the context of "Roleplaying Games," balance, and TTRPGs in general.

I'd also argue that that's really not very life-like--a second go-round at life should at least in theory look pretty different, right? That's why belief systems with reincarnation don't expect you to be a carbon-copy of the person you were before, just similar. (Frex, the Dalai Lama.)

Edit:
To give a different example of something I consider extremely worthwhile, that is similar to all this, but that I am not interested in discussing here and now: Writing exercises. Have you ever used the "Exquisite Corpse" writing prompt method? Some might call it a "game," but it would be rather pointless to try to come up with "strategy" for it--the whole point is to break anticipated patterns and force unconventional, unexpected thinking through random association or "telephone"-like lack of symmetry. Either the word-sequence is in a fixed pattern (like "The <adjective> <noun> <adverb> <present-tense verb> a/n <adjective> <noun>," so that a guaranteed-grammatical but possibly very surreal sentence results), OR you can only see the previous word (thus connotations/expectations can change sharply from one word to the next). Either way, the whole intent is to generate surprising things, and the only "victory" condition is being inspired to write more formally as a result.

I think it's wonderful that we have this method, and that it's always fruitful to discuss this and other koan-like "break your chains" methods...just not in a discussion about balance in RPG design.

[NichG]

Okay. I appreciate that thread drift and stuff happens, but I guess I'm coming at this from, "How is this relevant to the thread topic?" We already answered the original question--"need" is too strong a word without any context--and had, as I understood it, moved on to the more useful, "What does 'balance' mean for the D&D context, and how necessary is it there?" Talking about abstract goals that may go entirely outside of gaming and game design isn't bad, but it seems a spinoff topic, at least to me.

It's relevant to understanding the motivations of people who engage with tabletop RPGs, D&D included. I think a lot of the circular conflict in this thread between e.g. you and Quertus is that his motivations for engaging in this activity have little to do with the question of whether it's formally a game. So when e.g. you define balance as a game being converged in its mathematically testable design goals, if it turns out that that framework of discussion totally misses out on a set of goals that players or DMs might have when engaging with the activity, it's going to present a blind spot that prevents any kind of mutual understanding or convergence.

The point of this spinoff is that I'm trying to make the boundaries of that blind spot clearer by way of examples that more clearly belong to it, but otherwise have aspects in common with the kinds of design problems that one might go about solving in order to ensure that one's group has a good time (whether they're technically playing a game or not).

If there's a set of activities that people engage in in the context of tabletop RPGs, which tabletop RPGs can be designed to enhance or bring about, and which comprise a portion of the goals of at least a subset of players, then a framework which is forced to exclude them on the basis of definitional boundaries isn't going to be the end-all be-all of discussion. While 'balance' can be defined within that framework, if you can recognize groups who want things but simultaneously recognize that those things are incoherent within the framework, the hope is that seeing that contrast might make it clearer why some groups or designers could be better served by rejecting (at least that sense of) balance as a design goal.

So as a meta-goal for the conversation, my success conditions are: A lesser success would be that the claims which previously seemed incoherent to you now seem consistent with at least some set of assumptions, even if they're different than the ones you'd bring to a design problem. A greater success would be that you'd find a way to extend how you're thinking about games and balance such that it becomes possible to consider this sort of experientially-focused design, such that you could e.g. in the future suggest subsystems and alternatives to people who have those things as their goal without the suggestions falling flat or being rejected.

As an example point, you've described the 4ed skill challenges system as - to your design goals - a good example of balanced design. However, to this sort of goal of providing transformative experiences that need to be lived to really have their full effect, the 4ed skill challenge system has a problem - it takes a set of disparate contexts or situations and places them under the same mechanics, effectively making subsequent engagements with the skill system more similar to each-other than they would be otherwise. I think if I said that at the start of the conversation, it'd be hard to even get across why one would want such a thing. After this spinoff conversation, I would hope that it would at least be raised to the level of 'I wouldn't personally want that, but if you want that I am able to understand the conditions that would have to be satisfied to provide it'. And, again, at a greater level of success, for you to be able to conclude 'I see why X approach to balance conflicts with that goal, but here's Y approach to balance that would still be compatible' for example.

(And, as a meta-meta thing, I think this line of conversation has made more forward progress to some kind of mutual understanding and even potentially actionable outcomes than the previous debate about 'is D&D 3.5ed a bad game?' which has been done to death on these forums and won't actually change what anyone runs or plays, so I'm personally having more fun discussing this).

Well, I mean, then it's not really a goal for a tabletop RPG, which was sort of my point. I'm interested in talking about games, since this is the Roleplaying Games forum, and our focus topic is balance (with the more-refined "what is balance, and how much is worthwhile?" question).

I think my players would probably be fairly put off if they were told that they didn't spend the last 12 months playing a tabletop RPG and could not discuss it here because there were aspects of the activity that put stress on the formal game theoretic sense of what a 'game' is... If one were to sit in on a session, it would be hard to say that its not some kind of tabletop gaming activity, even with internal strategies and consistent outcomes and stuff like that; its just that if you were to take the rules and run it yourself, it'd almost certainly end up being a different experience in the end, and copying over those internal strategies wouldn't work very well because they'd be highly tuned to the particular context of those players.

Then...yeah I guess I just don't consider that a game. If you play it twice and necessarily generate effectively identical outcomes....it's not a game. More like a really entertaining personality test or self-introspection questionnaire? Regardless, though, while it is a worthy subject of discussion, I just...don't really have anything to say about it in the context of "Roleplaying Games," balance, and TTRPGs in general.

I'd also argue that that's really not very life-like--a second go-round at life should at least in theory look pretty different, right? That's why belief systems with reincarnation don't expect you to be a carbon-copy of the person you were before, just similar. (Frex, the Dalai Lama.)

I think that's in agreement with what I'm saying? That basically, if you live in a world where everyone plays 'life' once, then life works a certain way. Take that same world, but let everyone play life N times (and remember the past experiences), and the world as a whole would look wildly different to the extent that at least some of the strategies or insights which would apply to the non-reincarnation world wouldn't apply very well to the reincarnation world.

Play Limit Break once and you get a game (or whatever we want to call it) about progressing past your own psychological boundaries. Play it a second time and your (and all other players') psychological boundaries have already been progressed past. I could go into details, but maybe a simple approximation would be easier to talk about concretely. Imagine the game: 'two players each pick a finite real number of their choice - higher wins'. A good strategy for that game might be to dive for named numbers in math journals, but once you've done it once, that strategy is burned and can't be used with the same opponent again. A decent meta-strategy might be that you want to escalate as slowly as possible while still maintaining a lead, but even that will only take you so far.

Edit:
To give a different example of something I consider extremely worthwhile, that is similar to all this, but that I am not interested in discussing here and now: Writing exercises. Have you ever used the "Exquisite Corpse" writing prompt method? Some might call it a "game," but it would be rather pointless to try to come up with "strategy" for it--the whole point is to break anticipated patterns and force unconventional, unexpected thinking through random association or "telephone"-like lack of symmetry. Either the word-sequence is in a fixed pattern (like "The <adjective> <noun> <adverb> <present-tense verb> a/n <adjective> <noun>," so that a guaranteed-grammatical but possibly very surreal sentence results), OR you can only see the previous word (thus connotations/expectations can change sharply from one word to the next). Either way, the whole intent is to generate surprising things, and the only "victory" condition is being inspired to write more formally as a result.

I think it's wonderful that we have this method, and that it's always fruitful to discuss this and other koan-like "break your chains" methods...just not in a discussion about balance in RPG design.

I haven't used that, but it sounds interesting. I guess I'd say that a large part of my reason to engage in tabletop RPGs at all is to find those kinds of 'break your chains' moments in things that can only be understood through experience (because having other elements that are not-yourself involved means that you can get shoved in directions you weren't able to see just by pure thought).

[ezekielraiden]

It's relevant to understanding the motivations of people who engage with tabletop RPGs, D&D included. I think a lot of the circular conflict in this thread between e.g. you and Quertus is that his motivations for engaging in this activity have little to do with the question of whether it's formally a game.

But...okay. Let me be as direct as possible.

How can we discuss things that, as you yourself said, cannot be discussed?

That's my whole point here. I'm not constantly circling back around this because the play experience has to be mathematical. I've given examples where the test is mathematical, but the game is not--your very own Nomic example, where there is a clearly mathematically-testable aspect to it, based on survey data about what people want to do when exposed to the rules. The designers of Nomic very clearly wanted a particular, and communicatable, discussable, experience when playing the Nomic game, and designed it in such a way to naturally trigger that experience. Thus, the most fully "un-game game" example you could give besides Limit Break (which...I honestly can't even find anywhere, so do you have a link?) was still perfectly well described by what I'm talking about.

That's why I'm so deeply confused about this idea that there is this enormous swathe of gaming/design-goals for which ANY form of discussion, ANY form of analysis, ANY form of strategy, is COMPLETELY alien.

The point of this spinoff is that I'm trying to make the boundaries of that blind spot clearer by way of examples that more clearly belong to it, but otherwise have aspects in common with the kinds of design problems that one might go about solving in order to ensure that one's group has a good time (whether they're technically playing a game or not).

I don't actually SEE it as a blind spot though! That's been my whole point. I don't see how design considerations don't apply to Nomic, and I feel like if I could actually look at the rules for Limit Break, I could point out exactly the kind of thing I described above, where there are still statistical things you can do to test if what you're doing is achieving the goals you set out for. Some mechanics are themselves mathematical, and thus admit direct analysis. Other mechanics are not mathematical in themselves, but we can collect data about user response and overall experience in such a way as to learn whether the players at least believe they are having the experience you want them to have.

So as a meta-goal for the conversation, my success conditions are: A lesser success would be that the claims which previously seemed incoherent to you now seem consistent with at least some set of assumptions, even if they're different than the ones you'd bring to a design problem. A greater success would be that you'd find a way to extend how you're thinking about games and balance such that it becomes possible to consider this sort of experientially-focused design, such that you could e.g. in the future suggest subsystems and alternatives to people who have those things as their goal without the suggestions falling flat or being rejected.

As an example point, you've described the 4ed skill challenges system as - to your design goals - a good example of balanced design.

Well, I'd call them an okay example. They required some tweaking after launch, and you kinda needed to "play beyond the rules" (not contradicting any of them, but setting up further rules and ideas than what the strict limits of RAW included). So: balanced, but only okay balance. They could admit a significant amount of improvement by asking carefully-designed survey questions about the experience itself (mostly because, as I said, I have had multiple DMs who "got" the rules, even ones theoretically "brand new" to 4e, and thus saw the Better Thing beyond the system-as-it-was.)

However, to this sort of goal of providing transformative experiences that need to be lived to really have their full effect, the 4ed skill challenge system has a problem - it takes a set of disparate contexts or situations and places them under the same mechanics, effectively making subsequent engagements with the skill system more similar to each-other than they would be otherwise. I think if I said that at the start of the conversation, it'd be hard to even get across why one would want such a thing.

I don't even understand what the phrase means. What is a "transformative experience"? How on earth can mere procedure create such "transformation"? That would be like having a procedure that, of necessity, simply by following its motions, one could achieve Buddhist enlightenment. The whole point of something like that is that it is in defiance of procedure. But the only constituent OF rules is procedure.

(And, as a meta-meta thing, I think this line of conversation has made more forward progress to some kind of mutual understanding and even potentially actionable outcomes than the previous debate about 'is D&D 3.5ed a bad game?' which has been done to death on these forums and won't actually change what anyone runs or plays, so I'm personally having more fun discussing this).

Personally I think the answer is quite simple. "Bad game" means different things to different people, hence one must provide one's definition first. If the standard is "did people have fun with it? if not, then it's a bad game," then 3.5e is just about the antithesis of a bad game, because it's historical fact that an enormous number of people had a huge amount of fun with it. But the standard is rather loose, as one can argue literally any game enjoyed by more than one person can't be a "bad game." My personal preference is: "does the game fulfill the goals its designers communicated to its players? if not, then it's a bad game." And by that standard, there's a lot of room to argue that 3.5e did badly--after all, not only its own creators, but the people who took up the mantle to keep it alive, have both admitted that in order to fulfill their goals better, they had to leave that system behind.

I think my players would probably be fairly put off if they were told that they didn't spend the last 12 months playing a tabletop RPG and could not discuss it here because there were aspects of the activity that put stress on the formal game theoretic sense of what a 'game' is...

Except I'm not the one claiming that. You did:

Strong sense of alienation when a player transfers from one table running the game to another. Inability for players to accurately describe their experiences with the game, or to agree with eachother about the nature of their experiences.

You have, repeatedly, referenced a game that cannot even be discussed. And that's why I have been pushing against this thing being, in any meaningful sense, "a game." If you cannot talk about it, how did someone tell you how to play it? You're describing a ruleset that somehow means two players cannot accurately describe to each other how it felt to play.

You have described--at least in high-concept terms--how it feels to play Limit Break. We are already discussing it now. You are not at a loss for words. Ergo, Limit Break is not one of the things I'm saying cannot be a game. If it can be discussed, if you can ask people questions about their experiences so that you can have even a hope of shaping that experience (even if the shaping you choose to seek is "not shaping it at all"), then it admits design, and can be a game.

If one were to sit in on a session, it would be hard to say that its not some kind of tabletop gaming activity, even with internal strategies and consistent outcomes and stuff like that; its just that if you were to take the rules and run it yourself, it'd almost certainly end up being a different experience in the end, and copying over those internal strategies wouldn't work very well because they'd be highly tuned to the particular context of those players.

Okay so...what does that mean? I really, truly do not understand what this part means. I honestly think you're taking an overly-restrictive definition of "strategy," here, since (as I said earlier) I am allowing that even Nomic admits strategy, it's just going to look extremely different from other games' strategy, because it relies on a different set of skills than most games (mostly creativity and spontaneity, as opposed to within-the-rules analysis and calculation). "Strategy" is not limited to within-the-rules analysis of numerical values; it embraces a much wider variety of things. "Design" is not limited to setting probabilities for an internal random number generator; it too embraces a much wider variety of things--in short, anything you can gather data about, which is a very, very large set of things. (I am not a hyper-empiricist, I genuinely believe there are questions for which the answer cannot be found solely through measurement and counting, but I am skeptical of the notion that such questions are that relevant to game design.)

I think that's in agreement with what I'm saying? That basically, if you live in a world where everyone plays 'life' once, then life works a certain way. Take that same world, but let everyone play life N times (and remember the past experiences), and the world as a whole would look wildly different to the extent that at least some of the strategies or insights which would apply to the non-reincarnation world wouldn't apply very well to the reincarnation world.

Okay, but are you therefore asserting that there is no strategy at all? That there is no way, even in principle, for two players to sit down and compare relevant parts of their experiences, and to plan for future concerns? Because that's the kind of "game" you described above. One where trying to live your 7th "life" with a different group of people is so alien, you're not sure you're still playing "life."

Play Limit Break once and you get a game (or whatever we want to call it) about progressing past your own psychological boundaries. Play it a second time and your (and all other players') psychological boundaries have already been progressed past. I could go into details, but maybe a simple approximation would be easier to talk about concretely. Imagine the game: 'two players each pick a finite real number of their choice - higher wins'. A good strategy for that game might be to dive for named numbers in math journals, but once you've done it once, that strategy is burned and can't be used with the same opponent again. A decent meta-strategy might be that you want to escalate as slowly as possible while still maintaining a lead, but even that will only take you so far.

But such a game still admits discussion. Playing it with a different partner doesn't result in "alienation" from one's former experiences, nor would such new experiences with a new player be alien to an old hand. It admits a fair amount of analysis (that is, uh, kind of what led to Cantor's diagonal argument, and I'd argue that "number theory before Cantor" is a pretty substantial body of analysis). Even if you ignore the purely mathematical parts, leaving just the psychology of it, there's still a lot to learn from things like "what would the average joe think of as the largest number?" or "what's the first '-illion' term that most people don't know?" (e.g. most people know "trillion," a lot of people probably even know "quintillion," but how many people know nonillion?) All of this is under the umbrella of strategy and discussion. It's amenable to design (though which such an intentionally-reduced example, I admit there aren't many avenues of design available without changing the nature of the game somehow.)

I haven't used that, but it sounds interesting. I guess I'd say that a large part of my reason to engage in tabletop RPGs at all is to find those kinds of 'break your chains' moments in things that can only be understood through experience (because having other elements that are not-yourself involved means that you can get shoved in directions you weren't able to see just by pure thought).

Okay but...doesn't that mean you see potential design even in that? You have just identified a discussion-capable, can-be-tested element: by including one or more sources of non-self-generated notions, do people get an experience of...well, you'd probably want to think about exactly how you phrase it. "Change in perspective," "broken (mental) chains," "epiphany," etc. That's still a statistical test, based on gathering user input after play. Even if the actual experience itself is always inherently unique, the concept of "epiphany" is not so anti-transferrable that we can't talk about it. That would be like saying that we can't talk about the flavor of bananas or apples just because I can't beam into your head what apples taste like to me--and, further, that apple pie recipes are impossible to design because any two people will always taste the pie a bit differently. (Of course, in food science, you're best served by catering to data clusters rather than finding a single data center, but that's going further into the statistics and psychophysics of it than I think befits the thread thus far.)

[NichG]

I'm going to try to consolidate a bit since some points are raised in multiple places in the post.

Thus, the most fully "un-game game" example you could give besides Limit Break (which...I honestly can't even find anywhere, so do you have a link?) was still perfectly well described by what I'm talking about.

...

I thought I had mentioned it earlier in the thread, but my general MO is to write a custom system for any campaign I intend to run, aiming at some particular experience or question. Limit Break is one of those. If you want to check the rules, they're here: https://drive.google.com/open?id=1pD...xMFGmT4isb9iGI

I am claiming a different design methodology, so I should explicitly state it. The design methodology basically amounts to trying to find things that I myself have difficulty understanding or anticipating, then writing the system in such a way that gameplay will center around those things and force them to be resolved. Alternately, I'll try to target some kind of standard belief or assumption which would be easy to just take as true, but then try to find a system that falsifies that belief.

With Limit Break, the thing I didn't understand which was used as a seed was the underlying motivational structure of the superhero genre - not just 'what happens' in superhero media, but ultimately why it happens and what it means. In particular, a common occurrence in campaigns I've run is that the PCs cross a threshold where things that once mattered to them become irrelevant (be it their home country, other NPCs, recurring villains, etc) unless those things also escalate. In the superhero genre, generally characters do big cosmic deeds motivated primarily by personal connections, and those connections are maintained in the spotlight. If there was something real to that, I could try to design a system around it to query that conceit, and I'd be likely to learn something in the process.

The assumption I targeted was that character power comes from mechanical options, and that sub-parts of a system with more powerful options lead to more powerful characters and vice versa. So the game was constructed to basically make any character capable of things far above the power level of pretty much any other tabletop RPG other than Nobilis - however, I wouldn't tell them too explicitly that this was the case. The rule in question was: 'as long as you can justify it to yourself in your character's theme, and as long as no one who would be directly affected by it opposes it, it automatically works' combined with 'you can at any point create a Lv0 power for free'. So a starting character with the power 'leadership' for example could create a universe in which they are worshipped by the populace and lead armies of minions out from the portal, or other such over the top things. This was the gamble on my part, the intentionally designed-in uncertainty - would the game fall apart, or would it actually work despite that?

And it ended up actually working, with a number of moments during play that were quite informative to me. So, I can say at least within the framework of my design methodology and goals, it was a successful campaign despite not being designed specifically with a mind towards balance (other than trying to make sure that there were enough wildcards in the system that it would be hard to obtain any guarantees).

But...okay. Let me be as direct as possible.

How can we discuss things that, as you yourself said, cannot be discussed?

That's my whole point here. I'm not constantly circling back around this because the play experience has to be mathematical. I've given examples where the test is mathematical, but the game is not--your very own Nomic example, where there is a clearly mathematically-testable aspect to it, based on survey data about what people want to do when exposed to the rules. The designers of Nomic very clearly wanted a particular, and communicatable, discussable, experience when playing the Nomic game, and designed it in such a way to naturally trigger that experience.

...

That's why I'm so deeply confused about this idea that there is this enormous swathe of gaming/design-goals for which ANY form of discussion, ANY form of analysis, ANY form of strategy, is COMPLETELY alien.

I don't actually SEE it as a blind spot though! That's been my whole point. I don't see how design considerations don't apply to Nomic, and I feel like if I could actually look at the rules for Limit Break, I could point out exactly the kind of thing I described above, where there are still statistical things you can do to test if what you're doing is achieving the goals you set out for. Some mechanics are themselves mathematical, and thus admit direct analysis. Other mechanics are not mathematical in themselves, but we can collect data about user response and overall experience in such a way as to learn whether the players at least believe they are having the experience you want them to have.

...

Except I'm not the one claiming that. You did:

You have, repeatedly, referenced a game that cannot even be discussed. And that's why I have been pushing against this thing being, in any meaningful sense, "a game." If you cannot talk about it, how did someone tell you how to play it? You're describing a ruleset that somehow means two players cannot accurately describe to each other how it felt to play.

...

Okay but...doesn't that mean you see potential design even in that? You have just identified a discussion-capable, can-be-tested element: by including one or more sources of non-self-generated notions, do people get an experience of...well, you'd probably want to think about exactly how you phrase it. "Change in perspective," "broken (mental) chains," "epiphany," etc. That's still a statistical test, based on gathering user input after play. Even if the actual experience itself is always inherently unique, the concept of "epiphany" is not so anti-transferrable that we can't talk about it. That would be like saying that we can't talk about the flavor of bananas or apples just because I can't beam into your head what apples taste like to me--and, further, that apple pie recipes are impossible to design because any two people will always taste the pie a bit differently. (Of course, in food science, you're best served by catering to data clusters rather than finding a single data center, but that's going further into the statistics and psychophysics of it than I think befits the thread thus far.)

...

But such a game still admits discussion. Playing it with a different partner doesn't result in "alienation" from one's former experiences, nor would such new experiences with a new player be alien to an old hand. It admits a fair amount of analysis (that is, uh, kind of what led to Cantor's diagonal argument, and I'd argue that "number theory before Cantor" is a pretty substantial body of analysis). Even if you ignore the purely mathematical parts, leaving just the psychology of it, there's still a lot to learn from things like "what would the average joe think of as the largest number?" or "what's the first '-illion' term that most people don't know?" (e.g. most people know "trillion," a lot of people probably even know "quintillion," but how many people know nonillion?) All of this is under the umbrella of strategy and discussion. It's amenable to design (though which such an intentionally-reduced example, I admit there aren't many avenues of design available without changing the nature of the game somehow.)

For all of this, I'm going to go back to a point I raised earlier, and ask to be very specific about my wording here. When specifying those goals, I'm specifying a direction to move towards, not an end point which we must be at. Similarly, I'm not specifying that something be impossible to discuss, but just that discussion should not suffice to capture the experience of the thing and that players who have played at different tables end up finding that they disagree about what playing the game is actually like. This is not the same as saying that the game is impossible to discuss.

Also, whether this set of things means that design is impossible is a point of disagreement between us, so while this is incoherent from your definition of design, it isn't actually incoherent from my definition of design. That is to say, I am not at all claiming these things cannot be designed - I'm in fact claiming that designing them is not at all a problem, and that doing so is relatively easy in the design methodology I use (which is not based first and foremost on mathematical testing).

Since you claimed that these goals are incoherent, I'm taking that as you holding the position that these goals are not subject to the design methodology you are forwarding. If you've changed your mind on that and have decided that they aren't incoherent, then we can go from that point. If they feel simultaneously designable and incoherent to you, then that should seem paradoxical to you.

(As far as alienation in the real number game, if I just had a game where we started with Graham's number and went from there, and the next game my opponent said 'a trillion!', or worse handed me a terabyte harddrive filled with code that outputs a 1, executes Busy Beaver of 10^12, and outputs a zero at every step then I'd feel like it's not the same game, even though it has the same rules.)

I don't even understand what the phrase means. What is a "transformative experience"? How on earth can mere procedure create such "transformation"? That would be like having a procedure that, of necessity, simply by following its motions, one could achieve Buddhist enlightenment. The whole point of something like that is that it is in defiance of procedure. But the only constituent OF rules is procedure.

At least within how I'm thinking about these things, there isn't a contradiction here. Seeing a movie for the first time can be a transformative experience. Hearing music for the first time can be a transformative experience. Learning to actually do the math of quantum mechanics can be a transformative experience. Even playing tabletop RPGs for the first time (and the corresponding Stockholm syndrome people have with D&D) can be a transformative experience.
It's not supernatural or mystical, just trying to be the kind of thing that people remember for a very long time because it anchors some new way of thinking that they can engage in. Things like 'eureka!' moments.

At a meta level, generally things which force your brain to operate in new ways have at least the potential for qualifying. They're going to generally be somewhat personal though, so it's harder to hit the target if you don't know the players ahead of time. Recognize assumptions or modes of operation favored by the player, create a scenario in which those things are under tension, allow the player to work out that tension themselves via the vehicle of the game. I'm also not going to say I have a 100% hit rate on this kind of thing, but even if it's say 10% per campaign per player then I think that's still worth pursuing.

Okay so...what does that mean? I really, truly do not understand what this part means. I honestly think you're taking an overly-restrictive definition of "strategy," here, since (as I said earlier) I am allowing that even Nomic admits strategy, it's just going to look extremely different from other games' strategy, because it relies on a different set of skills than most games (mostly creativity and spontaneity, as opposed to within-the-rules analysis and calculation). "Strategy" is not limited to within-the-rules analysis of numerical values; it embraces a much wider variety of things. "Design" is not limited to setting probabilities for an internal random number generator; it too embraces a much wider variety of things--in short, anything you can gather data about, which is a very, very large set of things. (I am not a hyper-empiricist, I genuinely believe there are questions for which the answer cannot be found solely through measurement and counting, but I am skeptical of the notion that such questions are that relevant to game design.)

...

Okay, but are you therefore asserting that there is no strategy at all? That there is no way, even in principle, for two players to sit down and compare relevant parts of their experiences, and to plan for future concerns? Because that's the kind of "game" you described above. One where trying to live your 7th "life" with a different group of people is so alien, you're not sure you're still playing "life."

By 'strategy' I mean long-term plans or policies of interacting with the game in order to achieve a goal or reward. Responses you map out in advance of the events which will demand those responses. In math terms we could talk about this in terms of, for example, the way that you can have strong guarantees of asymptotic zero regret methods for bandit problems, but once you add the right kind of feedbacks or non-stationarities to the problem then you can force those methods to break. But I'm not sure that's a useful analogy for us to discuss?

Keeping in mind the previous thing about the difference between sitting at an extremal point and moving in the direction of that point, I'm not requiring strategies to be completely invalidated. Rather, I'm looking to increase the degree to which strategies are invalidated. Not everything from Life 1 will be transferrable to Life 2. Less will be transferable from life to life than from, say, chess game to chess game. It's possible to devise methods by which strategies will be destabilized, largely centering around non-stationarity and incomplete information. This element in particular I think you could probably target in your design methodology, though it would as a consequence make the kind of testing you want to do more difficult or expensive (because now you really do need to test multiple campaigns-in-sequence to get one data point, rather than getting multiple data points even within a given campaign).

Spoiler: D&D aside

Show

Personally I think the answer is quite simple. "Bad game" means different things to different people, hence one must provide one's definition first. If the standard is "did people have fun with it? if not, then it's a bad game," then 3.5e is just about the antithesis of a bad game, because it's historical fact that an enormous number of people had a huge amount of fun with it. But the standard is rather loose, as one can argue literally any game enjoyed by more than one person can't be a "bad game." My personal preference is: "does the game fulfill the goals its designers communicated to its players? if not, then it's a bad game." And by that standard, there's a lot of room to argue that 3.5e did badly--after all, not only its own creators, but the people who took up the mantle to keep it alive, have both admitted that in order to fulfill their goals better, they had to leave that system behind.

I just find it sort of boring conversation because, if you conclude with respect to some formal abstraction that 'the designers did a bad job' then so what? If you've been enjoying it, that conclusion would serve you poorly if it led you to stop. If you haven't been enjoying it, you don't need to prove that it's objectively bad in order to make the decision not to play it. So there's a high likelihood that someone discussing this honestly will be led to a point where they have to say 'okay, you can say all of that, but it seems that I don't care'. That tends to make it a frustrating discussion topic, because rather than just say that people try to find some logic to defend their tastes (or try to argue 'no, you can't possibly like that!'), and stuff goes in circles.

[ezekielraiden]

@NichG: If it pleases you, I can do a full reply to your post, but there's a thing I was reminded of tonight that, I think, may help show why I don't see the gap you seem to. That is, I see it as a sliding scale where some of the stuff you're looking for is difficult to design for, but still something that works within the framework I'm talking about. It's not a difference of kind, but of degree--and, as I think D&D and PF have shown, it's not like even the obviously-mathematical parts are really that easy to design either.

Specifically, I'm thinking of the Thanksgiving Day, 1983 episode of Sesame Street, an episode that actually aired before I was born but which still manages to be moving today as an adult. That is, this is the episode where the cast explains to Big Bird that Mr. Hooper died, and won't be coming back, and that there is no "good reason" for it, it's just...because.

This was a serious issue, and the crew knew they had to deal with it correctly--a stumble on a topic of this magnitude would be bad for the audience and the show. So they did their homework. They consulted with psychologists, writers, religious experts, and more, and they made sure to make use of test audience children. In other words, they did do a significant amount of real statistical testing, backed up by careful, respectful understanding of the topic and how hard it could be to communicate it effectively to children. Test-audience children were carefully interviewed before and after, and were in other ways tested to determine what effect the episode might have, even though dealing with something as complex (and I suspect you'd agree) obviously transformative as "accepting death" is just about the definition of individual and non-transferable.

This, to me, is a perfect example of a carefully-designed experience, that admitted very important rigorous (and statistical) testing, while still ultimately being about exposure to a difficult, painful concept for young children who might struggle to understand it. Admittedly, a TV show is not a game, so there's no rules-construction and no equivalent to "multiple trials," etc. But as far as a comparison goes between an episode of a TV show and the design of a game? This seems good enough.

It's also worth noting that the scene was only filmed with one take--the cast couldn't bring themselves to go through it a second time. But in doing it that way, they made their reactions as thoroughly honest as possible, and on review, they realized that that was very important. You see some of the characters tear up, struggle to speak, look away, etc. Big Bird presents the perfect image of the insistent child, frightened and upset while the adults struggle with a subject profoundly difficult for them and the child. Showing children that there are things that can even make adults sad, and for which even adults have no good answers and no power to change the situation, was extremely important. Based on their pre-airing interviews and post-airing studies, the overwhelming majority of children understood the message, paid attention through the whole scene, and had a fruitful and positive conversation with their parents about the subject of death--and this was millions of children, because it was aired on Thanksgiving Day, when most parents would be able to have such a conversation with their children.

So...yeah. That's a transformative experience that was very carefully designed, with extreme care and consideration put to the specific questions asked during the design process. It included elements that cannot be "designed" in any meaningful way, like the emotional reactions of the cast members, but which could with review be understood for how they would influence the intended result (communicating to children about death).

Anything that cannot be grasped by this kind of design--anything sufficiently difficult to share that no meaningful data can be collected about the experience--would need to be pretty dang alien. If even teaching children about death (and the emotions and problems that result from it) is quite meaningfully "designable," I'm not really sure that there's all that much within the "transcendental experience" umbrella that can't be "designed" but is still worth pursuing. It requires genuine creativity and effort to ask the right questions for this kind of design. You have to work hard and maybe be unusually smart to do it, but you can still do it.

[AdAstra]

Gotta agree with Ezekielraiden. Just because something is impossible to quantify in objective terms, doesn’t make it inimicable to design, discussion, and testing. Something as hard to quantify as say, the taste of a really good and complex meal, can still be tested, discussed, refined, rebuilt, and tested again. Language is ALWAYS an imperfect description of the world around us. You can never describe something and be sure the other person is thinking and feeling exactly as you do. Chances are they are not. Testing and discussion are still possible, though

[NichG]

@NichG: If it pleases you, I can do a full reply to your post, but there's a thing I was reminded of tonight that, I think, may help show why I don't see the gap you seem to. That is, I see it as a sliding scale where some of the stuff you're looking for is difficult to design for, but still something that works within the framework I'm talking about. It's not a difference of kind, but of degree--and, as I think D&D and PF have shown, it's not like even the obviously-mathematical parts are really that easy to design either.

Specifically, I'm thinking of the Thanksgiving Day, 1983 episode of Sesame Street, an episode that actually aired before I was born but which still manages to be moving today as an adult. That is, this is the episode where the cast explains to Big Bird that Mr. Hooper died, and won't be coming back, and that there is no "good reason" for it, it's just...because.

This was a serious issue, and the crew knew they had to deal with it correctly--a stumble on a topic of this magnitude would be bad for the audience and the show. So they did their homework. They consulted with psychologists, writers, religious experts, and more, and they made sure to make use of test audience children. In other words, they did do a significant amount of real statistical testing, backed up by careful, respectful understanding of the topic and how hard it could be to communicate it effectively to children. Test-audience children were carefully interviewed before and after, and were in other ways tested to determine what effect the episode might have, even though dealing with something as complex (and I suspect you'd agree) obviously transformative as "accepting death" is just about the definition of individual and non-transferable.

This, to me, is a perfect example of a carefully-designed experience, that admitted very important rigorous (and statistical) testing, while still ultimately being about exposure to a difficult, painful concept for young children who might struggle to understand it. Admittedly, a TV show is not a game, so there's no rules-construction and no equivalent to "multiple trials," etc. But as far as a comparison goes between an episode of a TV show and the design of a game? This seems good enough.

It's also worth noting that the scene was only filmed with one take--the cast couldn't bring themselves to go through it a second time. But in doing it that way, they made their reactions as thoroughly honest as possible, and on review, they realized that that was very important. You see some of the characters tear up, struggle to speak, look away, etc. Big Bird presents the perfect image of the insistent child, frightened and upset while the adults struggle with a subject profoundly difficult for them and the child. Showing children that there are things that can even make adults sad, and for which even adults have no good answers and no power to change the situation, was extremely important. Based on their pre-airing interviews and post-airing studies, the overwhelming majority of children understood the message, paid attention through the whole scene, and had a fruitful and positive conversation with their parents about the subject of death--and this was millions of children, because it was aired on Thanksgiving Day, when most parents would be able to have such a conversation with their children.

So...yeah. That's a transformative experience that was very carefully designed, with extreme care and consideration put to the specific questions asked during the design process. It included elements that cannot be "designed" in any meaningful way, like the emotional reactions of the cast members, but which could with review be understood for how they would influence the intended result (communicating to children about death).

Anything that cannot be grasped by this kind of design--anything sufficiently difficult to share that no meaningful data can be collected about the experience--would need to be pretty dang alien. If even teaching children about death (and the emotions and problems that result from it) is quite meaningfully "designable," I'm not really sure that there's all that much within the "transcendental experience" umbrella that can't be "designed" but is still worth pursuing. It requires genuine creativity and effort to ask the right questions for this kind of design. You have to work hard and maybe be unusually smart to do it, but you can still do it.

I generally agree with this post. I think the only point of deviation is perhaps what we're each considering to be the 'hard part'. In the sesame street episode example, the transformative element was essentially already in hand, and the people involved understood that it would be so already (because e.g. they could remember when they first learned about death as kids, or telling their own kids about death). So the design question was then about how to modify the reception of that - sort of polishing the technical details around an already solid core.

In my eyes, it's an interesting design challenge to, e.g., 'find something as impactful as learning about death for the first time, that we as adults do not yet know'. Since we probably can't guarantee finding such things, how do we maximize our chance of happening upon them? One has to be very careful in using historical data to guide such a search, since it's easy to overfit: 'Minecraft was a transformative experience, so if I make a Minecraft clone it will also be transformative' doesn't work - therein lies the path to another fantasy heartbreaker. I'm not going to say it's impossible to use historical data for it, but it has to be used very indirectly and with open eyes to the changing context in order to have any hope of generalizing.

Gotta agree with Ezekielraiden. Just because something is impossible to quantify in objective terms, doesn’t make it inimicable to design, discussion, and testing. Something as hard to quantify as say, the taste of a really good and complex meal, can still be tested, discussed, refined, rebuilt, and tested again. Language is ALWAYS an imperfect description of the world around us. You can never describe something and be sure the other person is thinking and feeling exactly as you do. Chances are they are not. Testing and discussion are still possible, though

Non-designability wasn't my claim, though you'll probably have to wade through 2-3 pages again to trace how we got onto this. Basically it goes back to something Ezekielraiden originally said about defining balance as matching quantifiable targets, and the discussion which followed had to do with establishing whether or not that position implied that balance (in that sense) equaled design.