General [Design Discussion] Perfect Imbalance

[Grillo_Parlante]

Yeah I think that some of these ideas apply much easier to a cube on the power level of the penny-pincer cube than most of our cubes. Even then, I don't think there will ever really be a cyclical thing happening. There's a linear optimization of the decks until the decks are as good as they can be. Then you probably just have 1 best deck.

You may be lucky and end up with 3-4 best decks. Then you may have a rock paper scissors thing, but this isn't standard where you can expect everyone to bring one deck because it did well last week, and bring the deck that beats that. You just draft the best deck presented to you in your packs.

Also note that nobody gets to build the best deck anyways, as we're all fighting over its cards. So there's not much point in talking about the best decks, just the best cards. So say Troll is the best card in the cube. So everybody wants the edicts. That doesn't mean you're gonna start seeing significantly more edicts. Sure, they may get maindecked a little more, but you wont start seeing 4x edict in every sideboard. The number of edicts you'll see stays about the same. So Troll is still the best card.

Ok, this is what I mean. I just addressed this:

1. Idea that metagames needs something to react to and counter-act to was introduced using a simple video and example.
2. Idea briefly discussed, and we were making some real progress on it, than a side debate began about the simplified nature of the video and example
3. Sigh clears this up around page 1 and 4, here and here, respectively.
4, Unfortunately, the same points that had been cleared up pages ago were still being re-argued (on page 6!). I think sigh's posts were maybe a little bit too subtle, or became buried under the barrage.

1. No one at this point should be discussing whether you can produce a truly cyclical relationship or not. We dealt with this back on page 1, and I provided two links to posts that directly address it, on this page.
2. This is an unsexy topic, and the purpose of the troll->edict relationships was just supposed to be a simplified relationship of the way a dynamic metagame provides tools for play and counter play. Here it is again:

If I want to give decks tools to encourage play and counter play between one another, exploiting the natural impetus of certain players to "break" a format, that should be fine, and is certainty preferable to the alternative of them actually breaking it (which they will do).

Outside of the original posts, where its referred to as a simplified example, there is this post here, here, here, and here. Around page 5 I just gave up, but here we are again. This relationship isn't going to play out literally as such, because its intended as a simplified example of a metagame back and forth. I tried to provide a real world anecedote on page 5, here, where I described how players began bounceland spamming as the format's most effective strategy, which caused the rest of the meta to shift dynamically in response. It was of course dismissed, until sigh, who thankfully ran bouncelands in his own format for a time, was able to confirm here something that I think we all should know: players will adapt to a predominate or overrepresented strategy by prioritizing the disruption they can use against it.

The nuance that is being lost here is that the dichotomy between talking about the best deck and the best cards is a false one:

An overabundance of archetype support or mana fixing leads to lazy drafters and a stagnant meta. How many times do the Riptide denizens see drafters go into a cubing session (or sometimes even retail draft) with tunnel vision locked on the archetype they want? If this mentality survives a few cubing sessions (and drafters with it end up with winning decks), I would hazard the environment isn't healthy. At least, I personally use this as signpost to make changes. (Diluting strategies that over perform or going into a more modular card that also supports a strategy has paid off with cards slotting into a greater variety of decks during deck building.)

And we can (and should) step way back on this. If your cube is about exploring a rainbow of <x> strategies within a narrow band (midrange, ramp, aggro, good stuff etc.) than your players may not be drafting a singular best deck, however, they are limiting themselves to a narrow band of best deck or strategy. That relationship is one that people can meta.

My mistake was simplifying it as troll vs. edict, which has caused people to repeatedly latch onto this idea that I am just talking about a singular troll deck, rather than a band of decks revolving around hardened threats, and the tools that a meta might use to keep such a deck from stagnating a meta, as well as how that deck might counter play.

Sorry to sound so crabby, and you know that I love you all, but I felt I had to be firm on this one.

 

[Rasmus Källqvist]

I can understand the frustration though if people stick more to the words than the idea, one which I agree with.

 

[FlowerSunRain]

I'm thinking there is a third factor variable issue here.

 

[Suicufnoc]

And we can (and should) step way back on this. If your cube is about exploring a rainbow of <x> strategies within a narrow band (midrange, ramp, aggro, good stuff etc.) than your players may not be drafting a singular best deck, however, they are limiting themselves to a narrow band of best deck or strategy. That relationship is one that people can meta.

But they can't really. Let's say the hardened green deck is dominating. If people don't metagame, all the answers to the deck that are in the cube get spread out among the players. Then somebody figures it out, drafts all of those answers, and maybe beats the dominating deck. But the dominating deck then roflstomps the rest of the field, as they have no answers. Then people realise they need those answers, and value them a bit more highly, and end up where they were before, with the hardened green deck still dominating. Its then the designers job to weaken the deck or add more answers.

Of course, that's still probably too simplified, and maybe every other deck can just adjust their strategy a bit to beat the Troll Deck. So black decks pick up edicts, white picks up wall of omens, blue learns to hold counters for problem cards, red goes wider or something. Those changes lead to a few more slight shifts in the cards people want. Then that makes even less effect the next time around, and after some number of iterations, the meta is "solved".

Am I just totally missing the point? I don't understand what it is we're trying to figure out.

Edit: Went back and reread some stuff. I will say that I don't see how we can create what I would call a "dynamic metagame". So I guess I'm hoping for arguments for how its possible or examples of how to do it.

 

[Jason Waddell]

To this effect I'd say that an underpowered archetype is far more of a problem than an overpowered one. Given sufficient fixing, players can all divvy up something strong until it is sufficiently weak. You just have to make your archetypes flexible enough. People at my local store believed Infect was the best strategy, but sometimes 5 people would try to draft it and it would suck. The problem, as you've noticed, is that there's no smooth transition out of infect.

But let's say something like black aggro is underpowered. If it's underpowered even when you get the nut black aggro deck, there's nothing your players can do to compensate in their drafting.

The drafting process itself does a lot to balance things, and as long as archetypes interlock well and each one stands a fighting chance, your players will do some balancing for you.

Note how in constructed, this isn't really true, because any number of players can play Dark Jeskai. WOTC prints sideboard cards as tools to let the players balance the meta.

 

[Grillo_Parlante]

Of course, that's still probably too simplified, and maybe every other deck can just adjust their strategy a bit to beat the Troll Deck. So black decks pick up edicts, white picks up wall of omens, blue learns to hold counters for problem cards, red goes wider or something. Those changes lead to a few more slight shifts in the cards people want. Then that makes even less effect the next time around, and after some number of iterations, the meta is "solved".

This isn't bad, though its again simplistic, but we're at last moving forward somewhat. I think it might be better to step back and look at this from a macro level. Even when you're choosing to frame this as the most anemic meta response ever, you can at least still see a meta response, which I have to at least interpret as a good sign.

Let me provide a real world example. We're going to have to leave the world of simplified examples, so this is going to be a bit less eloquent.

I run these:





20 bounce, 10 gain, 10 scry.

At the very beginning of the format, I only had a vague idea of how these types of cards would work out, and just had a vague idea that the meta would be bifurcated between slow multi-color decks and faster 2 colors decks. People always approach this type of format (as did we) by drafting fairly traditional mana bases, which meant 16-18 lands. There was actually a point where we were seriously talking about going down to 10 bounces, because they were wheeling.

Eventually, however, people realized that you couldn't use traditional mana math with the bounces, and that they allowed fundamentally broken deck construction. A 17 land deck with 2-3 bouncelands in it, often felt mana flooded, and the decks began to become more and more extreme. People realized that you could run essentially any deck off of 15 lands, if you drafted enough bouncelands, which was a huge advantage: you essentially could be a non-aggro deck with an aggro decks ability to top deck. I have an example of a ramp deck, of all things, in the penny thread thats runs off of 15 lands.

In addition, because of the way bouncelands work, they are extremely popular with weaker drafters and players, as they sort of magically correct for negative variance by their design. This means we had a card type that was both popular with good drafters (who wanted to exploit the cards mathematically) and bad drafters (who could lean on them as a sort of learning crutch), and was present in the cube in density. Now, there wasn't enough density where everyone could do this, but with 20 bounces in the format, its a safe bet that one or two people at the table are going to be running these more extreme decks. The last point is important, and should help conceptualize Chris' point of:

An overabundance of archetype support or mana fixing leads to lazy drafters and a stagnant meta. How many times do the Riptide denizens see drafters go into a cubing session (or sometimes even retail draft) with tunnel vision locked on the archetype they want? If this mentality survives a few cubing sessions (and drafters with it end up with winning decks), I would hazard the environment isn't healthy.

So lets recognize this for what it is: a danger point in the format's history, and I have been in this spot so many times before with other formats. I am about to have a stagnant meta, because everyone is going to be on a defacto bounceland ramp strategy. And how do I fix that by patching? Do I go in and rip out the bouncelands, reduce them by 10, nerf one of the best features of the cube? Or should I go in and buff the aggro decks, breaking singleton on pushed 1 or 2 drops that race under the bouncelands: effectively putting into place another nerf. All of these solutions are terrible, because they put me in a position where I am having to constantly react piecemeal to how my drafters are playing the cube, and I end up spending months trying to calibrate exactly how to balance this. Since cube is holistic, whatever changes I make will never be in a vacuum, and can create more problems that I have to address in other parts of the cube.

Its so much easier to just give the meta the tools to correct itself; and if you want to get really greedy with bouncelands, any disruption of them in your high CC, 15 land monstrosity is going to be a disaster.



These are non-narrow cards where the resolution of any single one of them can effectively beat a deck looking to abuse bouncelands. Resolving a single copy of mold shambler in some of these matchups feels like resolving blood moon in modern: a strong hoser. Of course, unlike actual sideboard cards (which are too narrow for cube) these cards are generically fine, but can contextually police our dominate strategy. And just as the presence of blood moon in a certain meta can cause people to play different decks or adjust their mana bases (at least until blood moon becomes less popular), we had a similar reaction here.

The temples have been good since they were added, but we are also at a point where we are seeing very few of the extreme 15 land decks, and now people are at least supplementing their mana bases with a higher % of gainlands and scry lands, or sometimes even forgoing bouncelands completely. Of course, right now, people are highly prioritizing the disruptive tools (I first picked petravark a couple weeks ago), but if this strategic approach recedes, I would expect to see the bounceland abusing strategies to become more common again. There will probably be about 3 or 4 different approaches to mana base construction that will develop, their popularity rising and falling with the popularity of the format's disruptive tools.

Most importantly, this feels fun rather than punishing; a puzzle to solve, rather than a heavy handed rebuff from me the designer, to a drafter just using the tools that I gave them.

While this back and forth, will not be cyclical in nature, it will be more like an oscillation (as Rasmus originally suggested, though this is probably more multi-dimensional), with the format using its own tools to continually push in different directions of strategic optimization, but never (if well designed) allowing for any single approach to settle and stagnate at a single point of optimization. Thats what I mean by a dynamic meta.

 

[Suicufnoc]

What's keeping the meta from eventually settling, though, beyond changing the cards before it gets solved?

If the bouncelands answers is strong enough that bouncelands become unpopular, those answers are too strong. If those answers are flexible, they will stay popular and people just won't play bouncelands anymore.

If the answers are only good against bouncelands, then maybe you get the oscillation effect, but you also end up with non games where people who happen to pick the overpowered hate while bouncelands are popular win, or people who pick bouncelands while the overpowered hate is popular lose.

If the hate/answer cards aren't overpowered, then you just have normal, balanced games of magic, and there's nothing for the players to adjust.

Of course we're talking about making something just slightly overpowered, which might put the whole thing somewhere in between? I don't think so. If aggro is slightly overpowered in general, people will pick those cards a little higher, which should solve the balance "problem"

 

[Jason Waddell]

These are non-narrow cards where the resolution of any single one of them can effectively beat a deck looking to abuse bouncelands. Resolving a single copy of mold shambler in some of these matchups feels like resolving blood moon in modern: a strong hoser. Of course, unlike actual sideboard cards (which are too narrow for cube) these cards are generically fine, but can contextually police our dominate strategy. And just as the presence of blood moon in a certain meta can cause people to play different decks or adjust their mana bases (at least until blood moon becomes less popular), we had a similar reaction here.

Is this a good thing?

In the abstract if somebody described a set of cards as "maindeckable Blood Moons against half their non-basics that aren't dead in other matches", I'd be super wary.

 

[quanta]

Although formally speaking, there is a static Nash equilibrium for any game, and I don't see how magic gets around it; in practice, there's not a snowball's chance in hell of players actually finding it. Human players are probably better thought of formally from an evolutionary game theory point of view. And in this case, there can be formal solutions which oscillate forever essentially because the players aren't really trying to solve the game exactly. But this is all largely pointless in application to a game as complex as magic.

I think it's more a question of how long before your players' heuristics reach a sort of evolutionary dead end. I agree with grillo that in a good cube, ideally we should have built in answers that aren't embarrassing when they aren't totally hosing the strategy they are meant to answer. Practically speaking, I think unless you jam your cube a lot a lot, it should be pretty doable to keep things moving around even if you only rotate cards in and out very slowly and in a way that doesn't make any seismic shifts in the meta.

 

[Grillo_Parlante]

What's keeping the meta from eventually settling, though, beyond changing the cards before it gets solved?

If the bouncelands answers is strong enough that bouncelands become unpopular, those answers are too strong. If those answers are flexible, they will stay popular and people just won't play bouncelands anymore.

If the answers are only good against bouncelands, then maybe you get the oscillation effect, but you also end up with non games where people who happen to pick the overpowered hate while bouncelands are popular win, or people who pick bouncelands while the overpowered hate is popular lose.

If the hate/answer cards aren't overpowered, then you just have normal, balanced games of magic, and there's nothing for the players to adjust.

Of course we're talking about making something just slightly overpowered, which might put the whole thing somewhere in between? I don't think so. If aggro is slightly overpowered in general, people will pick those cards a little higher, which should solve the balance "problem"

Of the 8 cards I mentioned, 4 of them I would consider strongly favored to find their way into a final 40, while the other 4 are more contextually, and do not fit in every deck. This means that their popularity as disruptive pieces can wax and wane in accordance with what the meta is doing. Once they move down the pick order (and I think acidic slime is the only slam first pick) you than have them wheeling, maybe ending up with the wrong drafter, or someone takes one as a speculative pick that doesn't pan out, or theres a hate draft: there are a lot of draft dynamics at work here.

Like I said before, I don't think this is really that radical of a concept, as at the end of the day, we are just talking about disruption in cube. I can assure you that its possible to calibrate a disruptive package (even LD, and you'll notice this LD is all high CC, for the express reason that well designed decks should be able to recover from it) that does its job without completely eliminating a strategy,

Like I mentioned:

While this back and forth, will not be cyclical in nature, it will be more like an oscillation (as Rasmus originally suggested, though this is probably more multi-dimensional), with the format using its own tools to continually push in different directions of strategic optimization, but never (if well designed) allowing for any single approach to settle and stagnate at a single point of optimization.

Imagine, for example, three points on a grid that represent three different optimal strategies for drafting and building mana bases. The drafting habits of your players can be tracked on this grid. After a certain period of time , as players learn what works and doesn't work, a disproportionate percentage of our data points begin to congregate around one of the optimal strategies. We want to provide our drafters with tools so that they can began to disperse that concentration, by making it only contextually optimal rather than objectively optimal, our ultimate goal is to prevent the meta from stagnating at any one of those points. The better we are as designers, the more we can try to make these shifts more dynamic and interesting.

Now, this isn't going to be a cyclical model, its not going to promise a never ending perpetuation of the exploration phase (for that you are stuck constantly patching or building new cubes). There is probably no reason to deliberately create unbalance (as the original video suggested) because you will naturally end up running some pushed or unbalanced strategy because cube design is hard (though this may make someone feel less bad about certain inclusions).

What this does do is help the meta to settle in a constructive way, where this is some continual push and pull between different strategic points, and we aren't settling on any single point.

Is this a good thing?
In the abstract if somebody described a set of cards as "maindeckable Blood Moons against half their non-basics that aren't dead in other matches", I'd be super wary.

I didn't describe it as such

The people that are losing to mold shambler are people that are getting very greedy with both their mana bases and their curves. They are effectively running a 17 land deck on 15 lands as a strategy, and making themselves extremely soft to a particular form of disruption because of the strategic line they choose. That decision is on them.

I know that wasteland is very popular here, but I never understood why its inclusion is brilliant, but if its any other type of land destruction, even an effect stapled to a 6CC creature, its GRBS. It take some fairly extreme circumstances for something like brutalizer exarch to produce non-games.

 

[Suicufnoc]

Wasn't really knocking mold shambled itself, just saying that any hate that's powerful enough to make the targets undesirable is too strong, and anything that's not too on strong probably isn't going to have much effect on the meta.

I don't like wasteland, for what its worth...

 

[Jason Waddell]

I don't think Wasteland is brilliant, nor do I think 6CMC land destruction is broken.

Just to ground the conversation, here are the hits for "destroy target land" from Ravnica block:


They take things to a lower CMC than you, but with the exception of Helldozer, none leave a creature behind. Rolling Spoil also potentially gives card advantage. (I am sure there are RAV bounce spells that hit lands too)

It's fine for bouncelands to have cards that destroy them. Out of curiosity, do you have any ways to "protect" bouncelands?

 

[Jason Waddell]

 

[Grillo_Parlante]

Don't really have ways to protect bouncelands, though I suppose you could put counterspells and discard in that camp. If a player wants to counter play, they can make changes to their mana base or deck structure, which provides its own set of strategic rewards. The blue bounce effects (or the vark), for example, are poor against gainlands or scry lands. There are also cards that reward players for running higher land count: stuff like magmatic insight, tormenting voice, and compulsive research. I think my favorite card in that regard is frantic search, because while you really want to run that card with bouncelands, you also want to have redundant things (like lands) to discard. Mana base construction can get really involved.

Wasn't really knocking mold shambled itself, just saying that any hate that's powerful enough to make the targets undesirable is too strong, and anything that's not too on strong probably isn't going to have much effect on the meta.

I don't like wasteland, for what its worth...

Sorry if I seemed a little over zealous there. My comments on wasteland weren't directed at anyone in particular.
With disruption there is a middle ground between too strong and too weak, which is well worth attempting to achieve.

 

[Suicufnoc]

No offense taken.

With disruption there is a middle ground between too strong and too weak, which is well worth attempting to achieve.

Absolutely, but this has always been a goal of most riptiders(lab maniacs?). Though Jason was talking about building a removal metagame from the ground up, which could be a fun experiment.

 

[Grillo_Parlante]

Though Jason was talking about building a removal metagame from the ground up, which could be a fun experiment.

Look no further than here. I finished up my experiment from pages ago. This is short the rest of the BOZ lands that have yet to be printed, but I wanted to experiment with removal and threat relationships.

360 design, close power band, and I tried to run only cards that have been vetted in our various riptide formats, so as to make things more relatable. My goal was to run as many cards as possible that could fit into a removal metagame structure. I think it came out ok, though I wish I could have made space for a goyf and a KOTR, both which feel like they should be in a list.

Also, I have no idea how you guys build your artifact sections, because there is not much there at this power level.

Note that we ended up with very few build arounds, minimal singleton breaking, and the cards in the list don't feel very narrow. The titans also mostly didn't make it in (though I tried). I think frost titan would be fine, and sun titan is great. Grave and inferno just don't interact with things, while primeval titan feels a bit on the weak side without more control over its tutor targets.

Here is a list of the removal interactions I was experimenting with:

1. Protection beats spot removal and evades tokens, but is trumped by devoid, which has problems with larger scale threats




2. Recursive and indestructible theats overwhelm or avoid damage based removal, but are are beat by smaller scale exile effects, but in turn have problems with larger scale threats





3. Indestructible threats beat damage based removal but has issues with minus counters or edict effects, which have problems with larger creatures or edict fodder.




4. Hexproof or shroud beats targeted effects but has issues with sweepers, death touch, or edict effects, which in turn have issues with dash creatures, and edict fodder.




5. Ramp cards that enable above the curve plays, but are vulnerable to early disruption that doesn't always scale well into the rest of the game.




6. Regeneration which beats damage based removal, but has problems with everything else



7. Wrath dodgers, which increase the value of instant speed removal which can be trumped by the other interactions




8. Enchantment or artifact based creatures, which increase the utility of artifact or enchantment destruction




9. Tokens are awesome (they are), they blank some removal, but kept in check by token killers and evasive protection cards.




There are also some size and growth based relationships, as well as color based based relationships




I also tried to sprinkle reasonable LD through the cube, and provide some more ambitious mana costs, while running 10 colorless producing utility lands, so as to strain people's mana bases.

Running some protection cards was interesting, and I wonder if protection doesn't get an unfair rep. Protection provides such an easy way to avoid board stall, and as long as you provide some ways to address those creatures, I don't think they should be too bad.

All of these more robust threats also means (I think) you can afford a higher removal count, which makes it somewhat easier for control to get their spot removal. I think it would be difficult for removal based decks to take over, however, since threat based decks can always just diversify their hardened threats in response.

 

[Quirk]

Ugh.

I've been staying off this thread because I'm tired, and busy, and this requires an epic injection of game theory to be useful, but it feels like it has to be done. I don't want to spend my weekend on this though.

Grillo, for the love of all things beautiful, go read up on minimax and the Nash equilibrium. Without you having an understanding of how the basics of game theory work, we're going to end up in some messed up conversations.

Minimax is pretty straightforward conceptually to most people who've played a lot of strategy games. There are a bunch of great examples of how it works in Magic in LSV's What's the play with LSV articles.

The Nash equilibrium is basically a state where changing only one person's strategy does no good if everyone else's remains the same. It is the point at which everyone has learned how best not to be taken advantage of.

So let's use these to address the initial contention in the video linked at the very start: that by providing imbalanced elements to the game we can keep some kind of cycle going.

I'm going to use the game their actual theoretical model is most based round: rock paper scissors.

The Nash equilibrium of rock paper scissors is to play all three equally at random - 1/3, 1/3, 1/3. That way no other strategy has a positive probability of gaining an advantage against it. It may not be obvious that you want to start by doing this, but it should be obvious that any time you have an advantage and need to hold it, this is where you want to be.

(Consider this against a player who plays purely rock, one who plays purely paper, one who plays purely scissors. Against any of them we win 1/3rd, lose 1/3rd, draw 1/3rd, for a neutral expectation. Any mixed strategy at all will be composed of up to three neutral parts.)

Let's imbalance rock paper scissors and see what happens. Let's make rock wins worth double.

The new equilibrium is to play at random 1/2 paper, 1/4 rock, 1/4 scissors. Check for yourself that this also scores evenly against all rock, all paper, all scissors.

But... wait. We're at a new perfectly stable point. There is no more dynamism, no cycles, no instability. All that's happened is that we have a different stable strategy.

The contention in the video is mathematically wrong. Imbalancing an existing game may change where the optimal strategy lies, but it does not prevent an optimal strategy from existing.

All we've learned is that changing the rules of the game (or the contents of your cube) will force a change in strategy. But that's what we didn't want to do, right?

 

[Jason Waddell]

But that's what we didn't want to do, right?

How did we reach this conclusion?

Maybe I've misinterpreted.

 

[Grillo_Parlante]

So let's use these to address the initial contention in the video linked at the very start: that by providing imbalanced elements to the game we can keep some kind of cycle going.

The contention in the video is mathematically wrong. Imbalancing an existing game may change where the optimal strategy lies, but it does not prevent an optimal strategy from existing.

I've not been responding to your posts because you keep on wanting to argue the same non-issues over and over again, and I figured it would just lead to a non-constructive argument. I will link you back to this post and this post. I'm not sure how much more clear I can be that I am not interested in making a cyclical relationship.

I will quote for you:

Now, this isn't going to be a cyclical model, its not going to promise a never ending perpetuation of the exploration phase (for that you are stuck constantly patching or building new cubes). There is probably no reason to deliberately create unbalance (as the original video suggested) because you will naturally end up running some pushed or unbalanced strategy because cube design is hard (though this may make someone feel less bad about certain inclusions).

Hopefully you won't take that personally, and we can move on now? Chalk it up to a misunderstanding?

 

[Quirk]

Now, it's not much use looking at the Nash equilibrium for Brainstorm or whatever. We aren't going to have data at a level which we can use to make useful pronouncements on it.

The interesting level is more at the point of decks and archetypes. The Nash equilibrium effectively arrives when your players decide that they know what's good and what's not, and they have converged on a common algorithm, something like: "take RDW if it's open, else ramp, else control, else WW, avoid black aggro at all costs."

(Note in reality that players only behave as ideally as this in highly competitive contexts, and there's a strong argument that Cube drafting is usually not like this. Players have personal preferences for archetypes and cards, like challenges, are stuck in their ways, etc.)

This is a thing that could happen. "Imbalances" won't change that. "First-order optimal strategies" won't either: a "first-order optimal strategy" is either the stuff that lets newbies win against other newbies, or some bit of unbalanced brokenness that acts as an "I win" for whoever picks it up. It's not a coherent idea in a strategy sense.

What we're really talking about now is how games get solved, and it's nothing like the Extra Credits guys imagine.

We all know how to play tic-tac-toe optimally (I hope). The game has a relatively small statespace - very few positions compared to something like draughts or chess. We exhaust its depth easily.

It's tempting to think that more choices makes a game strategically sophisticated. This is partly true, in that it is hard to have much sophistication in a small statespace, but alone is not enough.

Chess is a well-known game with a staggeringly large state space which has undergone centuries of human analysis, latterly aided by computers. It is not nearing solution and any solution will be exceedingly complex to describe.

Maharajah is a chess variant that pits a king with the powers of a queen and knight against the whole chess army on the other side. The state-space is lower than chess, but still very high. Unfortunately a simple strategy wins for black: advance the rooks' pawns until they queen, then queen the knights' pawns. With five queens and two rooks it is easy to deliver checkmate without ever exposing the black king to the threat of mate.

(Or, simpler still, Battleships on a 100x100 board exceeds the statespace of Go with ease, but the strategy remains very basic).

It is hard to build in a raw resilience to being solved. One thing that does matter is that if you've half-solved it as a designer already, the chances are good that your players will catch you up in a short period. Ideally as a designer you want to design a game that you know you have little realistic prospect of solving.

However, it is possible to make solving games harder in other ways. If you have a simple game with three choices to make, one of which always scores you 55 points, one 45, one 40, it takes you at most one play with each choice to determine which will always be best. If you have a similar game in which the choices give you a 55% chance of winning, a 45% chance of winning and a 40% chance of winning, it may take you many games before you determine with certainty which choice is best.

The more difficult it is to evaluate game elements, the more work that needs to be done to solve the game. Of course, this only works when the difficult-to-evaluate elements are not clearly worse than easy-to-evaluate elements: if you know that the tricksy option wins somewhere between one-tenth and one-half the time, and the dumb option wins three-quarters of the time, you are not much impeded in your progress towards an optimal strategy.

So, what is good in cubes if you're purely trying to prevent people reaching the optimal strategy? Hard-to-evaluate cards, and things that multiply game complexity, and caps on things that are too powerful.

 

[Quirk]

Hopefully you won't take that personally, and we can move on now? Chalk it up to a misunderstanding?

The problem is you still keep using terms like "pushed threat", as though imbalancing things made any difference, and "first order optimal strategy", like it was meaningful, and saying things like this:

Sure, and this thread is about an alternate way to approach the problem that doesn't involve constantly making changes to a format, and isn't soft to player psychological profiles.

WHICH DOES NOT EXIST.
You can make a format deeper, harder to solve, etc, but that does not get you away from player psychological profiles. Your players are going to roam across a format like a hill-climbing optimisation algorithm. Depending on their psychology, they may well get stuck on local maxima. You may well have to change things to shake them off it.

 

[Quirk]

Basically, my frustration is that the terms being used are still terms from the videos which are not simplified, but flat wrong in almost every respect. We end up discussing real phenomena but the model is nonsense.
e.g.

Idea that metagames needs something to react to and counter-act to was introduced using a simple video and example.

This idea is bad and wrong and it needs to die.

 

[Grillo_Parlante]

Ugh, this is exactly what I hoped to avoid: tedious semantics arguments and a brewing flame war. Sorry guys.

 

[Quirk]

The problem is the notion of metagames, of reacting to metagames, of treating the format as a metagame rather than a solveable game. It's absolutely true that people's evaluations of cards will change as they get used to them within the format, that their evaluations will grow nuanced as they build different decks and see them play different roles. But this is not a metagame thing. It's not a metagame reaction. It's a learning process. The concept of metagames they're talking about is not a sound concept. The models they draw their ideas from - LoL, Magic - are mostly kept interesting by injections of new stuff, by changes. When we talk about metagames, we draw away from the real models we actually have which give us some way to think about and talk about depth and unsolveability and all that kind of thing.

 

[Quirk]

Grillo: this is not semantics. Semantics would be if we were discussing the same concepts using different language. We do not appear to be conceptually on the same page. That you do not realise the very substantial difference in the concepts concerns me, and yes, frustrates me. I feel like you are not engaging with me, but merely assuming I've found some new way of putting the thing you were saying already.