Incremental Complexity

Announcement! In the future, I think I’ll do more articles in “video form”. Very lightly edited videos, mostly a voice over and some pictures/titles/video. I think that video seems to be where more of the conversation is happening these days. Here‘s the first video, on incremental complexity, a new way of thinking about strategy game design (designing them, and teaching them), inspired by Pandemic: Legacy.

Support my work on Patreon here! Special thanks to Patreon Patron Aaron Oman!

Today I wanted to talk about complexity, and a new way to think about designing strategy games.

A lot of videogames are really solvable, just when looked at in terms of their actual rules. They rely on execution and other forms of output randomness in order to make them seem a lot less solved than they are. You can be playing pretty much optimally, but get varying results because of the random rolls.

(I’ve written a lot about randomness in games before. Check the info below for links.)

The inefficiency of videogames and most board games means that the game doesn’t have to be all that deep. There doesn’t need to be much terrain to explore if “exploring terrain” happens really slowly.

But what about games that don’t rely on randomness, and DO have a lot of depth? There’s a good number of designer board games and Euro-games that meet this criteria. While I think that these games tend to represent the current peak of interactive-system-design-understanding, they are far from perfect.

One of the biggest problems a game like Puerto Rico has is that it’s hard to learn. I remember sitting down with the rulebook several times before I really figured out how to play. And I remember having this feeling like, “this is ridiculously, comically over-complicated.”

Of course, once you learn it, it’s really not all that complicated. But it’s hard enough to learn that the vast majority of people – even the vast majority of people who would love it if they did learn it, won’t learn it. It’s not accessible.

The simple answer to solving this problem is “just make the game simpler”. But the problem with that plan is that human beings are smart, and they are extremely good at solving systems. So even if you’re really good with emergent complexity and have a good “elegance ratio”, the system needs to be rather inherently complex to avoid being solvable. I’ve seen a lot of really good game designers who think about systems head too far in the “simple” direction and end up with a system that ultimately relied on randomness to work.

While the ridiculous amount of complexity in games like Magic and League of Legends is also an obvious problem in the other direction, the fact is that games just need to be pretty complicated to avoid being solvable. So if simplifying isn’t the answer, then what is?


I’ve been recently playing Pandemic: Legacy, which has been number one on BoardGameGeek pretty much since it came out. It’s the second game in the Legacy series spearheaded by Rob Daviau, with the first one being Risk Legacy. The basic idea is that you take a certain board game, and over time, after every match, you add/remove components and rules from the game, generally making it more and more complex.

Some new rules make some viruses behave differently than each other. Some  allow you to place road blocks between certain cities. Some allow two characters to have a “relationship” that allows them to interact with each other in a special way. In short, the game gets way, WAY more complicated over time.

At some point, I was just saying to myself, “man, this is the game Pandemic should have just shipped as”. It’s true that Pandemic, as it is, is a good game, but it could be a great game with some of the extra rules that develop in Legacy. But then you have the “too complicated game problem”. Pandemic is pretty successful as a game, partially because it’s rather easy to learn. But if we added four different viruses, complex characters, and two full sheets of player actions to the game, I’m pretty sure Pandemic would not be sold at Target.

So how do we fix this issue?


Here’s my suggestion. Game designers, develop your ideal system, with the ideal amount of complexity. Then, basically use the Legacy style system to have players, after every game, add one more component or rule to the game, until they’re left with the full game. Pandemic Legacy is designed mostly to be about the experience of opening the new boxes and getting the new rules and playing with them; the state it ends up in is not the “ideal play state”. Most Legacy players stop playing once they reach the “end” of the content, and that’s how it’s more or less designed to be played.

My proposal is to take that way of incrementally adding rules and use it to build players up to a complex game. The build-up would be a lot shorter than Pandemic Legacy, which takes 12-24 matches to “finish”. I would advocate something more like 6-8 matches.

The first match is the most barebones and simple version that you could possibly play and have any fun with. This will be a design challenge, but keep in mind that for very new players, it feels good just to grasp the basics. After the first match, you add a new rule – now you have X! And so on, until before the player knows it, they’re playing the full, super-complicated game.

This actually isn’t that radical. It’s a lot like what single player videogames and puzzles tend to do – start the player in some tutorial area where they don’t have The Sword, and then the get The Sword and learn about The Sword Move, and then they get The Shield and learn about The Shield move, and so on. But I don’t see this model applied to strategy games. Strategy games, instead, give you the full game in “tutorial mode”, where they just explain everything to you one thing at a time. Strategy videogames like Europa Universalis or Civ would do well to implement something like the INCREMENTAL PLAY SYSTEM, and board games like Puerto Rico would be way more accessible if I could show friends the simplified version first.

It’s true that some games, such as Agricola, has a simplified “Family mode”. That’s good, but I think it would be better if there were more than just two steps in the learning process. I wonder if Agricola would have been released as a more complex – and possibly better – game, if it weren’t for the concern that the game would be “too complicated to learn” for most people.

My whope is that by embracing this incremental play system from the get go, designers will feel more comfortable making games that are complicated enough to avoid solvability.

What do you think? Thanks for watching.

  • Jake Forbes

    I am really glad to see you talking about this as it’s an area that might be holding back your games’ successes.

    Regarding Agricola in particular, there are actually a number of toggles for complexity that players can choose from: 1) Minor Improvements are rated Easy, Complex and Interactive with the latter two not recommended for beginners. 2) Drafting cards is a more advanced option for setup. 3) The expansion, Farmers on the Moors, supplements all the current rules and adds an additional set of Major improvements, horses, home heating, clearing the farm, and the additional action step, not to mention dozens of new cards. I would actually not consider the Family mode to even be a step in teaching an adult. In my experience, Agricola is relatively easy to onboard new amateur board gamers with because the theme is so strong and logical and inviting. And each session does an amazing job of rolling out complexity in increments. Consider that the first round, players really only have ~6 choices. Then 7, 8, 9, etc. as new spaces and concepts continue to be revealed up until the final turn. It’s an amazing game for how each session is almost a tutorial in and of itself.

    You ask about the “more complex and possibly better” Agricola — have you played Caverna? It is essentially Agricola with more components, more choices sooner, reduced randomness and a whole alternate victory track with expeditions. It’s currently ranked higher on BGG, albeit with a fraction of people rating. I’m in the camp that prefers Agricola + expansion as I find that the constraints reduce AP and the theme is stronger; I like Caverna, but it is more of a victory point salad.

    In the past you’ve written pretty strong words against designing games where players can choose the rules they want to follow. Are you softening that stance, or is this purely an argument for stronger tutorials? Thinking of commercial game design and not platonic ideals, I think you could pursue this line of thinking a little further and consider that maybe most of a game’s potential players are perfectly happy to plateau partway along the complexity curve and that it’s maybe okay to let that be called the “real” game with the highest tiers of complexity being an expansion or an opt-in supplemental ruleset. And if you can accept that, the next step might be to allow the community to make their own mods, custom modes and house rules to allow complexity to evolve to audience needs.

  • DukeZhou

    Haha. I had the same experience when I finally tried a Catan app. It took a ridiculously long time to learn, and at the end of the process, I found the game to be a great letdown–it was at once overly complex and frustratingly limited, and worse, required dice. (I much prefer Carcassonne;)

    I’m quite interested in complexity scaling. From a mathematical standpoint I wonder if it may provide insights into solving games, particularly in regards to very compact, strategic models. It’s also an key element of extended [M] games, which allows scaling in an effectively unlimited number of ways related both to board configuration, token qualities (neither of which require added mechanics,) as well as additional mechanics, and render [M] ultimately unsolvable, despite the presumed solvability of the basic game.

    I can definitely corroborate that incremental complexity in an applied sense is quite useful.