Get updates via e-mail:

Main | The Secrets of Donkey Kong Country 2 »

The Interconnected Systems of Puzzle Strike

Here I talked about how it's best to have system where the subsystems aren't so tightly woven so that you can tinker with one subsystem without messing up the rest of them. Well, that's not how Puzzle Strike is. It's very tightly woven, and I didn't fully realize this until we were trying to balance it. This made it unlike any other game I have balanced. For months it changed radically, often multiple times per week, because changes to one system affected all the others. Some playtesters would come back a week later and feel like they were playing a different game. I've said before it's best to avoid such interwoven systems if possible, but I kind of wonder if they turn out even better and more fun *if* they are balanced--it's just that balancing such a delicate web is tough.

In my series on balancing multiplayer games, I talked about how "balance" can mean a couple different things. On the one hand, it can mean making sure that a set of different starting options, like characters in a fighting game or races in a real-time strategy game, are fair against each other. But there is also some concept of "balance" in games where everyone starts with the same stuff. The word can also mean making sure that the game system allows players to play in different ways, using different moves and tactics, and that it doesn't all boil down to just one thing (we'd call that "degenerate").

This is an abstract diagram of the card / chip game Puzzle Strike, not a screen from a video game.

In Puzzle Strike, I had to worry about both types of balance at the same time. You might think any asymmetric game has to worry about both, but actually I haven't really had to before. Street Fighter and Puzzle Fighter already had working systems, they only needed asymmetric balance, the kind that's about fairness. For Kongai, I created the system, but I didn't develop it in parallel to all the characters. I started by making the system, and it was basically fine from version 1. The only things that ever really changed about Kongai's system were adjusting the damage an intercept does and adding a switch-cooldown so characters can't switch two turns in a row. Other than that, all the balancing was about the characters, not the system itself. Same story for Flash Duel. I adjusted the system of En Garde, changing timing, edge cases, adding a push mechanic, etc, but Flash Duel's system was fine from day 1 and all the balancing was about the characters. Yomi comes the closest to Puzzle Strike in balancing system + characters, but because system changes happend over such a long period of time--6 years!--it was not quite as traumatic.

The Different Types of Chips

In Puzzle Strike, there are character chips, puzzle chips, gem chips, and purple chips. (There are also wounds, but those just clog up your deck and do nothing, so we won't worry about them for now.) You can see all 62 chip designs here.

Gem chips count as money when played from your hand.The character chips are the ones that need "asymmetric balance," meaning they need to all be fair against each other. But we can't even hope to fix problems there unless we have a game system in the first place. The puzzle, gem, and purple chips compose that system. The gem chips you play from your hand count as money to buy other chips for your deck. You can use gem chips to buy more gem chips to increase your buying power. You can use them to buy purple chips, the most important chips in the game. Purple chips let you empty your "gem pile" and fill up the gem piles of other players. (Each player has a special zone called the gem pile, and if a player's fills up, he loses. The gems in there don't count as money, they just get you closer to losing.) You can also use gem chips to buy puzzle chips so that you can do a bunch of extra stuff that gives you resources, gives you more actions per turn, slows down your opponents, and so forth.

The main balancing troubles here are that the money system affects the entire game, the way purple chips work affects the entire game, and that puzzle chips compete with the other types of chips for attention. Let's look at each of these situations.

Money System

Money works just like in the game Dominion. You draw several chips each turn, and discard your hand at the end of the turn. When you draw your entire deck, you reshuffle your discard pile and keep going. So the more money you have in your deck, the more likely you are to have a turn where you draw a lot of money and have the ability to buy expensive things that turn. This works fine in Dominion, but there are more factors in Puzzle Strike. One is that purple chips affect money, too. In the very first version of Puzzle Strike, I wanted to make it so that when you use a crash gem (that gets rid of a gem in your "gem pile" and sends it to an opponent), you *also* get money. Money is actually called "gem power" in this game, so it's like you're breaking open the gems in your gem pile and getting the money out of them. The reasons I wanted this mechanic are that it gives you more incentive to crash, which is the basic fun mechanic of the game, and it also means that the closer you are to losing (nearly full gem pile), the more money you potentially have if you can crash a lot of your gem pile.

How hard is it to buy a given chip in Puzzle Strike? The answer depends not just on the money system, but also on the crash system. You have to know how hard it is to get crash gems (you start with one in your deck, by the way). Also, you can use crash gems offensively (on your own turn) which allows you to get money or defensively (on the other guy's turn when he crashes against you), but you don't get money that way (money from crashed gems only lasts until the end of the turn, so if you do it on the OTHER guy's turn, you can't actually use it; buying only happens on your own turn). Anyway, imagine trying to determine the cost of a chip. You have to take into account how easy it is to get crash gems, how easy it is to draw them, and how practical it is to use them on your turn versus the opponent's turn.

Purple Chips

In beta versions, the Crash Gem gave even more money when it broke open gems in your gem pile.I've already talked about how crash gems affect the money system, but in that first version, they basically broke the money system. There are different sized gems in the game: 1-gem, 2-gem, 3-gem, and 4-gem. You got 3 gem power (money) for crashing a 3-gem. This system's problem was that you simply had too much damn money all the time. The money from crashing was too much relative to the money you had in your deck, so managing how much money your deck has wasn't really something you had to think about. Though you could buy money chips to add to your deck, they just weren't worth buying. Even when they were made cheap or free, they often weren't worth buying because you'd rather get money the other way and draw non-money chips each turn.

We tried radically changing the money system so that money came from your gem pile and not from your hand. This was a completely different game, and it sounded interesting because it REALLY meant your power increased as you got closer to losing, but it ended up being not as strategic because managing money in your deck was a non-issue here too. There's also the option that money is what's in your deck plus what's in your gem pile, but that had problems too. I reverted it to the original money system and instead changed the purple chips. Now, crashing no longer gave money.

Imagine complaints from playtesters at this point that some character is too strong or too weak. There is nothing to be done about that if that character's chips have something to do with getting money or crash gems. The relative value of those chips is in so much flux that there is no way to fix characters that interact with money and crashes--pretty much all characters!


The stack of blocks symbol in the upper right gave you a threshold bonus so you could buy expensive chips sooner. This beta chips shows the old poker style border, too.If crashing doesn't give money, then there's no advantage to having a nearly full gem pile. But that is one of the main reasons to make the game at all--to explore the idea that almost-losing makes you more powerful, as it does in Puzzle Fighter. The next concept I added was called Threshold. To even be ALLOWED to buy a chip costing, say 6, you had to have 6 gems in your gem pile. This way, even if you had lots of money, you couldn't buy powerful chips until your gem pile got dangerously large. The game remained in this form for quite some time.

But...there were rumblings from playtesters that in actual practice, it did not enable more power for nearly dead players. What it did was control the pace of the early game so there's an interesting ramp up, and many liked that, but it wasn't really achieving my goal. Many were surprised when I removed this mechanic.

Height Bonus

Instead of threshold, I instituted the height bonus. This rule says that when you draw your chips each turn, the bigger your gem pile is, the more chips you get to draw. An extra +1/+2/+3 chips for gem piles that are 3 / 6 / 9 in size. This did have the effect of powering up a player close to losing (finally!), but think about it for moment. If you draw those extra chips, you are more likely to draw the crash gem you need to stay alive (good). You are more likely to draw a combo of chips that lets you do a lot of stuff at once (good). But you are also more likely to draw money, even if you have a fairly small pile size of 3. This is fine, but it affects the money system! At this point in development, having enough money was kind of hard, so it was actually a welcome change, but it's worth noting that interaction.

More on Money

Some characters have chips that help them get money, and some don't. That's fine, and it's part of the game's variety. But how strong do your other chips have to be to make up for it if you don't have money boosters? This depends on just how hard it is to get money. In the original version when crashing gave tons of money, who really cares if your character has a way to get more or not? With no money from crashing though, it was damn hard to afford things. Characters with no money boosting seemed like they needed to be impossibly strong in other ways to have any chance. This was made even worse by something else going on in another part of the game system: the dreaded mono-purple deck.

Puzzle Chips Compete for Attention with Purple Chips

This chip takes some time to get going, but it's powerful if you can survive long enough.At the start of each game, there are always 3 types of purple chips in the bank that all players can buy. There are also 10 types of puzzle chips, chosen from a larger set of 24 each game (so that you start with different puzzle chips available every time you play). You need purple chips to win because that's how you lower your gem pile and increase the other guy's. But the puzzle chips have all sorts of helpful bonuses and allow elaborate combos to make you even more effective. There's a tension there though: if puzzle chips are overal weaker than the effect of purple chips, players will simply ignore the entire set of puzzle chips...which means ignoring almost the entire set of moves in the game. This seems no problem to fix though, right? Just make puzzle chips powerful enough and it's fine. Except we also have to consider how much time (in turns) the player has to build up combos with puzzle chips. If games last absurdly long, then you have all day to build any combo you want. If games are very, very short, then you are under so much pressure from the other player's crash gems that you have no time to build up and you need to get your own purple chips ASAP. So how much time do you have, anyway? That depends on how crashing actually works.

Crash Systems

In the original version of the game, if I crash a 3-gem, it splits it apart and sends 3 1-gems to your gem pile, then you have the opportunity to counter-crash stuff in your own gem pile, to stop it. For example, when I crash my 3-gem at you, you could crash a 1-gem of your own and that would cancel one of the three incoming gems. Notice that it also destroyed one of your own gems so your gem pile will actually have a net increase of only 1 in this example, even though I crashed a 3-gem at you. The upshot: defense is good--really good. If defense is good, then you have all the time in the world to buy elaborate combos of puzzle chips...except that's not what happened at all. Puzzle chips were in general slightly too weak, but there was a much bigger problem than that. The problem was that players were finding strategies where they bought ONLY a few purple chips, then declined to buy ever again. They would use just purple chips and their character chips to set up degenerate interactions. Because these decks were so absurdly small, they had no real variation, and they could pull off the same degenerate tricks every turn reliably. Making puzzle chips a bit stronger was not going to fix that. Puzzle chips would have to be like 10x stronger to combat this, and that would break the game in other ways.

Death of Mono-Purple

Character chips were imbalanced at this point in development, but we can't even fix most of that yet. Money is a bit hard to come by, and maybe we should change all the costs of everything in the game, but we can't think about that either. We have to solve this mono-purple deck strategy thing right now because there is really no gameplay left with that happening. The answer comes in two parts, one I'll save for a future post (it has to do with the "combine" chip), but the one I'll tell you now is called "forced buy." We experimented with the rule that each turn, you must buy at least one chip. This means that it was impossible to have a super tiny deck that did exactly the same thing every turn. This took a bite out of the efficiency of mono-purple small decks. Now they had the same variation of turns as other decks, and they it gave other decks more time to setup different strategies against them. Players discovered that there many ways you could use combos of puzzle chips to beat people who only bought purple, so we finally had some balance. Not character balance, that was still all wrong, but at least system balance.


When two Chess pieces love each other very much, sometimes they have a stalemate.From the very first version, I always had stalemates in the back of my mind. Every turn, each player must take a gem from the bank and put it in his gem pile. This provides constant pressure on everyone and pushes the game toward some sort of finish. But the defensive power of crash gems means that in theory, players could cancel each other out with crashes and counter-crashes indefinitely. I hoped that the pressure of ever-increasing gem piles would be enough that these stalemates would be very rare, and we'd figure out some rule for if that happened. As players got better, the situation arose more and more (especially because a couple chips enabled stalemates even more), and we had a lot of trouble coming up with an elegant rule to address it.

The culprit seemed to be counter-crashing. Perhaps it was just too good of a tool defensively. We tried a different type of counter-crashing where it doesn't reduce the number of gems sent to you. So in my earlier example, if I break a 3-gem, it sends 3 1-gems to you no matter what. If you counter-crash a 1-gem of your own that 1-gem goes to me, but you STILL get 3 1-gems sent to you. This makes defense radically worse than the other system. We tried this for weeks and it changed everything. Offense was king, it was all about rushdown. It was tooooo much about rushdown, in fact. Some said that there was an early game, a mid game, but no end game. You were dead before you could put together cool combos for end game. This rule had too little defense, the other rule had too much defense (stalemates too). Hybrid solutions where stop "some" incoming gems with a counter-crash were inelegant and too complicated. We looked for different solutions to stop stalemates that didn't make the game end before it even really started.

Many suggestions required players to remember keeping track of some other stat that would change the rules toward the end to apply more pressure, and it was just inelegant. One solution involved only having a limited number of 1-gems that you ante (put in your gem pile each turn) and when you run out, you start putting 2-gems, 3-gems, 4-gems in there. THAT should end the game quickly. didn't. When players reached a state where they could crash consistently to force stalemate, they could stalemate just the same with bigger antes.

We sure tried a lot of things, and it was a very hard problem to solve. Notice that like everything in Puzzle Strike, it was affecting the whole game. How much defense/offense, whether games end very fast or very slow, it all affects character balance. Characters would rise and fall in tiers without even being changed, because the game system was changing around them.


We needed to solve this stalemate thing once and for all, and establish the correct amount of offense and defense. For inspiration, I thought about StarCraft. Imagine you show up at the Terran's base in the very early game. He has bunkers, so he's hard to attack. That lets the Terran tech up, and it lets you tech up or expand if you don't feel like directly attacking. Maybe you build a bit of an army and show up at his base later, but now he has bunkers and siege tanks. He's even harder to attack now. StarCraft is designed so that defense can be very powerful, even overwhelmingly powerful during the early game on some maps. If this continued throughout the entire game, it would be horrible. But the thing is, while that early defense might stop your Zealots and Dragoons (or Stalkers), there's also a late game where you just come in with a bunch of Carriers and completely own him. While defense can be good early, it's possible to build up overwhelming attack power that crushes defense in the late-game. That's what we needed. We needed Carriers.

Carriers have arrived.


I went back to the original counter-crashing rules, the ones that gave great defense and gave you time to build up combos. Then I added something allows you to attack with overwhelming power so you can crack any defense and prevent any stalemates: the uncounter-crashable 4-gem. If you crash a 4-gem--the largest possible gem--then the opponent cannot counter-crash it. In fact, they can't play any chips at all that trigger on crashing. The 4-gem is special, super powerful, and I gave it new artwork that emphasized that. Would this actually prevent stalemates though? Perhaps players can STILL make the game never end? Testing showed that the solution worked. The deadly 4-gem does indeed end games, and it works exactly as intended.

Lack of Money

We're almost out of the woods, meaning we're almost to the point where we can actually balance the characters. But there's still that old question about how good a character's chips need to be if none of them give money. Forced buy means you won't get killed instantly by stupid degenerate mono-purple decks. Uncounter-crashable 4-gem means the game ends at about the right time. Ok great. The trouble is that money is still a bit hard to come by and there's an amplifying effect that can make this problem spiral out of control. Whether you have much money or not, you must still buy a chip each turn. That means the money you do have is now diluted, and you're less likely to draw it (unless you bought money, but the whole point is what if you can't afford a 2-gem). There's a hump you have to get over, kind of like an escape velocity. If you have enough money to keep up with your increasing deck size, you can buy enough extra money to keep it all going. So if you're a character that has no money boosts, but one of your chips says "survive pretty long, then win the game instantly," you could still lose, which illustrated the trouble we were in. You could potentially fall into a money trap where you just can't do anything and you'd totally die before you ever get to play your win-the-game chip. There should be no such chip as that, by the way, I'm just illustrating the problem here.

Most testers wanted to fix this by giving all characters a way to get extra money. I thought this was a terrible idea though, as that is part of the variety of the game. Some characters have money boosts, while other characters have other advantages. I thought what we needed was not specific abilities that gave money, but rather a system-wide change (again...) that ensured everyone had slightly more money to work with. You'd think we could lower prices of everything, but we kind of couldn't. We already had established which chips should cost 1, 2, and 3, and we can't really lower those any further (nothing should cost 0 except for Wounds).

Crash For Money, Take 2

Looking back, I now see that I had it almost right in the very first version of the game. We already saw that getting money from your hand (version 1) was a better idea than getting it from your gem pile (experimental versions after version 1). We already saw that counter-crashing where you really do get to stop incoming gems was good (version 1), better than the purely offensive version (tried later). And now, I realized that getting money for crashing (version 1) was the right way to do it after's just that version 1 gave too much damn money. Instead of getting 3 money for breaking a 3-gem, what you really needed to get is 1 money for breaking anything, no matter what it was. It turns out that this is just enough money so that no character falls into the spiraling pit of "can't buy anything." But it's not so much money that it obsoletes entire systems of the game. It's just enough bonus to make you happy about crashing and to keep you from falling into degenerate game states.


There were a lot of casualties in the process. Mono-purple, threshold, crash-for-lots-of-money, offensive counter-crash, and progressively-increasing-antes all died off. It turns out that crash-for-$1, defensive counter-crashes, forced buy, and uncounter-crashable 4-gems are the way to go. Or to put it another way, it was a year-long exploration of a game system that revealed that the original concept was completely broken and degenerate, but that it only needed slight fixes to be totally great. It just needed crashing to give a bit less money, forced buy to prevent tiny degenerate decks, and uncounter-crashable 4-gems to end the game, just like Carriers do in StarCraft.

In the end, this is what most middle-aged actors posing as office workers from various ethnic backgrounds thought of Puzzle Strike's game system.

References (13)

References allow you to track sources for this article, as well as articles that were written in response to this article.

Reader Comments (8)

Thank you for the good read. I can't wait to play it.

June 18, 2010 | Unregistered CommenterZ

Very insightful behind-the-scenes development and game mechanics theory for the game here. Also enjoyed the Starcraft reference. Really considering purchasing it, but not at the moment due to the price.

June 18, 2010 | Unregistered CommenterMe

I do think that games turn out more fun when their subsystems are an interwoven, difficult-to-balance mess... as long as they end up balanced. Games get more complex as you try to guard against subsystems affecting each other.

One example is Starcraft's damage types and unit sizes. Here are three different approaches they could have taken to balance the units, from least complex to the most complex (in terms of game rules):

1. Don't use damage types and unit sizes at all. Balance the game based on HP and damage alone. If I'm not mistaken this is somewhat possible, but would create such an interwoven mess of unit balance that it would be unmaintainable by the designers.

2. Give each unit damage type and size stats. This is extra complexity that players have to know and remember, but it gives room to more easily balance the units without affecting every matchup with every change. For example, giving Hydralisks an explosive damage type helped make them versatile later in the game while not making them overpowered early-game, since they do less damage to smaller units.

3. Adjust damage on a matchup-by-matchup basis. This is a complete decoupling of the unit balance, but players now have to memorize a grid of matchups instead of knowing some intuitive rules.

Option 2 is obviously the best of these choices, since the unit sizes and damage types can be somewhat conveyed via flavor (visuals, sound effects, etc). Option 3 is done by some RTS I can't remember, which is pretty unintuitive.

June 18, 2010 | Unregistered CommenterPoisonDagger

Strifeshadow (a semi-indie, now pretty much dead) RTS made by Zileas (SC player, now League of Legends guy) among other people had the "each unit does something different against each other unit" balance theory, if i remember correctly. Kind of a nice idea, but really clunky.

Puzzle Strike hasn't really been on my radar, but wow that's a lot of stuff. Interesting that you sort of go full circle. I'm a little surprised forced buy survived, but hey... if it works!

June 20, 2010 | Unregistered CommenterWinter

Total Annihilation used method #1- units were balanced on HP, damage and cost alone. That the game ended up playable at all was a stroke of luck akin to Marvel vs. Capcom 2: the super-efficient light tank was countered by a critical mass of anti-air trucks, which lost to "inefficient" rocket guys from an entirely different tech tree. Unless you plan to get lucky with emergent systems like they did, I don't recommend it.

This is an excellent post, by the way.

June 21, 2010 | Unregistered CommenterRobyrt

I think the biggest take away from this is that if you don't have a well iterated, polished, and solid core to your mechanics, asymmetry isn't even something you should be thinking about much.

It's something you've touched on a number of times, especially when referencing Guilty Gear, but this really makes it very clear just *how* important that solid core is to everything else.

June 29, 2010 | Unregistered CommenterEolirin

Fascinating article... it inspires me to get Puzzle Strike, even though I'll probably enjoy it more than my non-gamer gf and buddies... I live in hope. I look forward to reading about the character balance. Bravo!

September 26, 2010 | Unregistered CommenterJohnny

Sirlin, you mention something about the combine chip re: stopping mono purple decks. Have you posted about it yet? I'd be very curious to hear more.

November 5, 2010 | Unregistered CommenterDoug

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>