News Shop
Events Chat

The Game Design of Codex

It might be fun to talk about the game design of Codex, because it is really one of the most “designed” games out there in a lot of ways. As a designer myself it feels really good to see something that is so intentional, so designed and yet works so well. To contrast, there are views out there that say that games should kind of be just a big unbalanced box of components, and let players figure out how it’s fun. Codex really isn’t like that - it embraces structure and I think that’s really great. :sparkles:

It’s only once every five years or so that there’s a game that comes out that I am really excited about. Codex is one of them! I wrote a big article about Codex and why I think it’s so great and such a feat of game design.

What do you think of the game design of Codex?


One thing that we’ve probably already started taking for granted is that the game is asynchronous, but that is really, really a great boon for this game. It not only allows for a viable online play-by-forum scene, but also speeds up the gameplay in real-time as well when you’re sitting face to face with your opponent, because there’s never a point where you have to wait and see how he reacts (unlike in Magic, where you have to do that essentially every time you take any action).

Not only that, but the fact that teching is finalized at the beginning of your next turn rather than at the end of your own turn really plays well into the asynchronous nature. It gives you something to do and something to think about while your opponent is going through the motions of his turn, so you’re not completely idle.

I know the game wasn’t originally like this, but overall I think that it wound up being a phenomenal change that really lets the game shine.


Originally, it was not asynchronous and played much more like MTG with stack like reactions. However, at that time, both garcia1000 and vivafringe were heavily playing outwitters and the former wished that Codex could be asynchronous so that in a future iOS app, he can play 10 games at the same time like Outwitters. While some people in chat were in disbelief, Sirlin did try an asynchronous version and he really liked what he saw.


Nice article! NItpick: the patrol zone lookout makes it cost $1 to target, not to attack.
What does it mean to say Codex has no core mechanism?


To me, Codex’s core mechanism is the patrol zone. Mastering the patrol zone is key to mastering codex.

Thanks for the correction, @garcia1000 - fixed.

Hmm. Well some games it’s easy to point to a core action or action-type that expresses on a fundamental level what a game is all about - like what you “basically do” in a game. Often, this thing relates to the core of the game, so in Smash Bros you basically are knocking people back so that you can knock them off the board, for example. Maybe in Codex the core mechanism is “attacking” to destroy the opponent’s base?

I tweaked the language to say that the game “seems to not have a clear core mechanism”, because I’m just not sure about what it is yet. But, even if it didn’t have one, that wouldn’t necessarily change anything about my opinion of it.

I have a lot of game designer friends on Twitter and they were all shitting themselves about Android: Netrunner when that came out. I really hope they give Codex a shot, because I think it is in the same general genre, but a million times better.


I feel like you’d have to say that there are several core mechanisms that all intertwine, and that’s where the issue is. When I’m showing the game to a friend/explaining it to someone who’s played a few games using those mechanics (Magic, Dominion, and like Starcraft or something, mostly), it sounds pretty straightforward right up until they play and have to figure out how not to mess up every turn.

Winning the game happens by destroying your opponent’s base, but that’s the goal, not the core thing you do.

Things I’d say qualify as what you “basically do” within the core gameplay loop:

  • Manage a gold resource pool (mostly by creating workers and playing cards)
  • Manage a cards-in-hand resource pool (mostly by not playing too many cards in a turn)
  • Manage a cards-to-draw resource pool (mostly during the tech phase and creating workers to add/remove them)
  • Manage a card-prerequisite resource pool (mostly by building Tech buildings, add-ons, or summoning heroes)
  • Manage an offensive capabilities resource pool (mostly by playing units, heroes, or spells)
  • Manage a defensive capabilities resource pool (mostly by playing units, heroes, spells, add-ons, or declaring patrollers)
  • Use offensive capabilities to affect opponent’s resource pools (mostly the card-prerequisite, offensive capabilities, and defensive capabilities pools by directly attacking them or using spells/abilities)

And eventually…

  • Use offensive capabilities (mostly by attacking or playing spells) to reduce opponent’s base health to 0 and win

I think those are the “core” mechanics, breaking out all of the ones that make Codex feel particularly unique (except the patrol zone, which I rolled up into defensive capabilities, and it’s important enough that zhavier above is arguing that it is the core mechanism). Moving resources around between the pools while destroying your opponent’s resources is the game (even the patrol zone helps do that). Base-killing is often a formality, because one player has created an insurmountable advantage in resources. This is a lot of why I’d argue the actual gameplay loop is the economics of war infrastructure.

The short version would be:

  • Build your economic infrastructure
  • Build your combat forces
  • Use your combat forces to defend your infrastructure and reduce your opponent’s infrastructure/forces until you can destroy your opponent’s base

That’s too generic, though. It describes every traditional RTS I can think of, the war elements of most turn-based strategy games, etc. It actually doesn’t address the mechanics at all. You could add “by playing cards” in there, but I’m not sure it much helps.

Codex ties so many resource mechanics together that simply haven’t ever been present all in one place. It’s hard to appreciate how much each one affects the game until you’ve screwed it up and lost because you ignored one.

I’d similarly argue Smash Bros. is more about maneuvering (through movement or attacks) to get your opponent into a relative position/health level where you can knock them off the board, so we might just look at things really differently. Thanks for raising the idea, though, as I’ve enjoyed thinking it through to write this.


That’s an interesting point — that the goal of the game doesn’t indicate the core mechanics, so much as what you are trying to accomplish by using said core mechanics. That the goal of Street Fighter is to reduce the opponent’s life bar to zero, but the core mechanics are controlling space and establishing offensive momentum in order to do so, etc. etc.


The core mechanic of Codex is lining up favorable combats.

There are a bunch of details both strategic and tactical that define what “favorable” is, and how to line them up, but everything is about making your attacks more profitable than your opponent’s.


I don’t quite get this. Mono-red, for example, could easily play games where they don’t give a whiff about lining up a favorable trade or combat. They simply want to get their opponent down to 0 Base before they lose. They could even play the entire game at a worse damage-per-turn ratio than their opponent, but still win as long as they went first.

I must confess, though, that I’m not exactly sure what a “core mechanic” is, or how to identify one. Codex is, to me, a toolbox for creatively achieving the game’s goal. The more creative a game allows you to be in pursuing its win condition, the less any one mechanism can be said to be “core”.


I feel like the “core mechanism” question might be a distraction from a potential broader discussion about what is really genius about this game, and what other games can take from it in the future. It’s not particularly relevant to this game - I only mentioned it in my article because in the context of my general writing it makes sense to talk about.

Some of the really great game design things from the game that other games should use in the future:

  • The patrol zone (or similar concepts) should be in EVERY future game of this type.
  • This card drawing system (or something similar) should be as well.
  • Some kind of dramatically limited customizability. Actually this is so great that I’m surprised it’s not getting complained about!
  • The codex book concept for a private bank (although this may have appeared elsewhere, not 100% sure).

If you are going for a base race, then a favorable trade is one that deals “more” base damage than you lose in board position.

It’s still about determining what an attack will cost, vs. the benefit.


I wonder if Keith is familiar with the concept of “coupling” from software engineering.

I think that the property of games with a strong core mechanism that actually makes them good is that they have strong coupling. This means that the various mechanics of the game are very strongly interconnected. Each action in any given system has a clear, direct effect on many other systems and you cannot understand any system of the game in isolation from the others.

In software, tight coupling is very bad, because it prevents you from breaking down your code into discrete comprehensible sections and thus understanding the entire system in such a way as to actually be able to solve problems. A system with only a few simple components can be a nightmare to work with if they’re highly coupled.

But in games having total understanding is bad, that means the game is close to dead. And having a game with lots of simple systems that are easy to understand but an overall system that is very complicated and multifaceted is like, the holy grail of game design.

When Keith talks about a “clockwork game” I think what he means is “games with very strong dependency between systems” And yeah, games with a strong “core mechanism” will be have high coupling.

Codex doesn’t have a core mechanism but it’s got really, really strong coupling. All the various different systems, hand size, tech level, building damage, board state, money economy, deck size, board state, patrol zone, cycle speed, tech progression. They’re all really tightly linked. No decision that you make in game affects only one of these systems. Every decision has effects on multiple axes which makes them super ambiguous and makes the game strategically deep.

A game with a strong core mechanism will have strong coupling/dependency because each subsystem or supporting mechanic will touch the core mechanism. But it’s not the core mechanic that’s important. It’s the property of high dependency that arises from that structure that leads to compelling gameplay.


Really well said, @Fenrir!


I think I learned something. I’m totally not going to use this in some bad fan-fic thatmost likely won’t see the light of day.


I disagree. Or at least, the win-more aspect only kicks in when one player is already winning by a large margin. When someone is winning by a smaller amount, the catch-up mechanisms are quite strong.

That’s a result of 2 things. First, Codex has a huge amount of very powerful effects, so even if you are down you can do some crazy thing to get right back into it. Second, resources are usually very tightly constrained, and you just barely can or can’t make the play you want. That extra gold from Scavenger is often the difference between a good turn and a great turn. And great turns in Codex are really really great.


I had an opponent once who won by playing two stampedes with Guargum, and after the match he was surprised he won, because before that explosive turn it looked like the match could be a dragged out fight. Codex reaches a critical mass where the game is just going to end, one player or the other is going to get an explosive turn eventually. Its not exactly random, as you have to be at least a little clever in teching in the right cards. But Barrelfish’s point is valid, there are a large number of very powerful effects that can swing the board state back and forth. With the right tech choices, the game isn’t really over until someone’s base is at 0.


I don’t tend to like the snowball effect either. But I agree with @Barrelfish that it’s not as delicate in Codex as you might believe. Once you reach that point there are some really silly comeback mechanics that are contained in Ultimate Spells and some colors’ T2 cards that act as a last resort. You probably won’t be able to play these if your opponent is doing a good job though. But sometimes your opponent takes a big loss in economy to put you in a bad position; at that point you can exploit it.

One thing that I enjoy far less is when there is always a way for the opponent to come back despite all of the work you did to place them in an unfavorable game state. This sort of thing really betrays what I take to be the goal of the game and most competitive games. The game should encourage optimal plays, smart responses, and small risks. If you achieve these things then you should have an advantage over the opponent because you have performed well and shown superiority that is reflected by the game state. If the opponent can turn their situation around from this point in a non-skill-based way, it appears arbitrary and undercuts the goals of the game and the incentivized actions, because actually those actions did not benefit you as much as the comeback mechanic did the opponent.

This sort of thing is pretty hotly debated in the fighting game community (who often do not have the best heads for game design) with respect to the ‘comeback meter’ and more recently in the discussion of removing chip-death from Street Fighter V. I dislike comeback meters for precisely the reasons described above; basically it can level the playing field with a simple low-skill command. Further, it might incentivize poor play in order to gain the advantage, which I take to be confused. The latter feature allows for comebacks in a more mitigated way: if the player can block and execute perfectly from 1HP, then that player can still win. This is an example of allowing a comeback mechanic that still involves a great deal of skill.

I don’t value arbitrary comeback mechanics. I wonder what kind of mechanic you would envision in Codex that would not benefit the player simply because she is losing. Do you value these sorts of comeback mechanisms? It seems to me that they create the illusion of a competitive match rather than displaying a true representation of skill.

Sirlin has more to say about these mechanics here, and while I’m not sure that I agree with his thoughts on the matter, we might see how some of these topics were reflected in the design of Codex.


Something I really, really appreciate about the design of the game is how the deckbuilding aspect works. In a traditional CCG like Magic/Hearthstone/whatever, you have this fundamental feel-bad problem where you need different cards in different stages of the game. So you’ll have a bunch of early-game cards, then some midgame cards, then some lategame cards, and you have a few cards that are useful all around.

But, well, for the sake of example, I’ll talk a little bit about my experiences with Anyfin Paladin in Hearthstone: I’d have tons of games where I’d be in a tough situation, and I’d have the ideal card to deal with that situation…except that the card was still in my deck, and I hadn’t gotten lucky enough to draw it. That feels really bad. And instead of having those cards, I’d maybe draw cards that aren’t useful in the early game. So I’d spend the first few turns getting blown out of the water by an aggressive deck for the mere reason that I had only drawn early-game cards, or I hadn’t lucked into any of my card draw.

I think it’s telling that in traditional CCGs, cantrip cards and card draw are valued immensely, because they increase the chance that you’ll get lucky and draw into the cards you need, out of the entire deck you have. Also, every traditional CCG has a mulligan rule, because of the possibility of getting a terrible starting hand of cards you can’t use in the early game (or not getting enough early-game cards)–and there’s no mulligan in Codex, because it’s not needed.

But in Codex? If you don’t need a card until Turns 5-7, then don’t tech it in at first. Your Tech 0 cards are exactly what you need to handle the early game, and you can tech in midgame and lategame cards as the turns pass. It’s a really brilliant bit of design that makes a huge amount of difference. Suddenly, instead of crossing your fingers and hoping that you draw the right cards for your situation (or, in the case of Hearthstone, a card you can play on the mana curve), you mostly get decent plays for your situation, presuming you’ve teched well. To top it off, you get to start workering cards from the get-go, slowly filtering the weaker cards out of your deck.

It’s a great cycle that, I think, promotes decision-making and skilled use of your resources. Plus, the game doesn’t totally remove randomness, because randomness does often lead to more-skilled play. You have to figure out what the best response is, given a limited set of options. I think that Codex hits a Goldilocks level of randomness here: it dispenses with the feel-bad “I hope I draw Card X” of traditional CCGs but keeps a “well I don’t have exactly the card I need, but I can use these cards” level of random draw. I really like that.


Was thinking about this, and wanted to add something to this and how this post relates to the core mechanism concept. The “coupling” idea is a good word for referring to the kind of elegance I think a lot about in game design. I have talked before about an abstract illustration of the concept in this way: imagine you have every rule in your game written on an index card, and each card is placed on a large gymnasium floor. (You can be as granular/not as you want with this; instead of a “rule” being on each card, you could have a “mechanism” or even a whole “system” written there.) Then tie a piece of string from each card to each card it has a relationship with.

Looking at it just from a “coupling” perspective, the more string you have down, the better. Right? Because coupling just refers to how interconnected everything is. And I do agree that coupling, in this way, is important.

But at the same time, it has to be ordered - it actually seems like it’d be a bad thing to have literally every rule/mechanism connected to every other one in some way.

So since there is going to be some order to the coupling, I think it makes sense to ask if some kinds of orders are maybe inherently stronger than others (at least in the ideal, if not necessarily in practice for a given system). To me, it makes sense that, in a vacuum, the most elegant order would be one core system/mechanism/rule - a spine, a thesis statement, etc, and everything else in support of that thing.

I also recognize that this kind of theory does not necessarily apply all of the time in all cases, and I think Codex is a case where it doesn’t apply. I do not think you can just “improve” Codex by “giving” it a core mechanism (nor do I think that’s even possible).

I think Codex does have strong coupling, and I think if we really broke it down, it probably has 2 or maybe 3 core mechanisms - which is what works for this system.

Hopefully this kind of clears up the whole “coupling” concept and how it is related to core mechanism concept.

1 Like