If I’m going to try to apply lessons learned in the paper world to the digital one, the two had better be close enough that the lessons are relevant. I won’t try to argue for that directly; instead, the individual points I make will, I hope, make that case. This section will briefly — in the interests of full disclosure — point out some of the ways that the two design problems are different. These differences are important enough to matter, but not enough to doom the enterprise.
The most obvious difference with the paper design process is how much cheaper it is. The first playable prototype can be made very early on with a paper game. And if the designer wants to change a rule, he says to the person opposite him “let’s try playing this way!” and they do that. If he wants to change a game object, he takes a pen and crosses out some bit of game data on a card or slip of paper and writes in the new value. This is enormously powerful, and allows a lot more time and effort to be put into the game design than into things like waiting around for a new build to show up. In other words, paper designs are much easier to prototype and test.
Tied in with this question of prototyping is that of interface. Interface is a challenge for both paper and digital games, but a much bigger one for the latter. One can imagine a good paper game with a bad interface (e.g. a poorly designed cardface) — also, as a designer, one can work quite productively on the game while its interface is in bad shape. For many computer games, the interface and the enjoyability of the game are so tightly tied together that it can be quite hard to separate them.
Another difference between the paper and digital worlds that’s tied in with the difficulty of coding is what percentage of design ideas are implementable at all. In a game like Magic, most new card ideas are at least possible to make (a few might not be due to technical problems with the rules system). With a computer game, a great many design ideas might not be practical to implement. So a large amount of Magic design work can be done before bringing in the technical experts. This would probably not be a good idea with a computer game.
Paper games have both the advantage and disadvantage that their rules are written largely (although not entirely, as anyone who has read the Magic rules knows) in English, rather than in computer code. On the plus side, the game can’t really “crash”… if something is poorly written, players will make some reasonable decision about what to do and move forward. English being less than precise, though, means that paper game rules will contain ambiguities: a mild annoyance in casual play, but a more serious problem in highly competitive play. Efforts to be extremely precise with rules (and remember, if there are many different gameplay objects, there are many rules, and many rules interactions) can lead to unpleasant legalese. Computer games can hide complexities in code that players never see.
Although paper games are on balance surely easier to “debug” than electronic ones — it’s easier to catch a typo than a subtle code bug — they are harder to patch. If a card has a misprint, it has it forever. If a unit in Starcraft needs to cost 20 more crystal, the next patch can make it so, in both gameplay and tooltip; declaring that a card with a printed cost of 4 should be played as if it costs 6 (by whom? how do they know? what if they meet someone else who doesn’t know?) is imposing such a burden on the user that it’s arguably never worth it.
All of these factors influence an important large-scale decision: how much effort to put into game design versus coding (think for the moment of things like laying out the rulebook or designing the graphics for the card face as the paper equivalent of coding) that game design. It’s not surprising that game design takes up a higher percentage of the effort in the paper world — the “coding” part is just plain easier — but given that a bit of time spent on game design can save a great deal of time coding, it’s perhaps surprising how big that difference is. Wizards of the Coast is not a large company by the standards of the computer industry, but we have around 40 people devoted full-time to game design. They don’t do graphic design, or layout, or printing… they just work on gameplay. Even a small trading card game will have 4 or 5 designers working on it, and Magic has many more. A typical computer game will have a much larger overall budget, but probably has fewer people devoted to gameplay, and many of them will have other duties. Designers of games that involve a large number of collectable objects, though, are discovering this may not always be the best mix.
So although the demands of coding might push one to spend almost all one’s resources on it, the fact that coding is so expensive should make one look for ways to substitute cheaper (but not easier!) things for it… like careful planning. I don’t mean this to be taken as an argument for long spec documents and no prototyping. On the contrary, I think there’s no substitute for having a playable version of the game to see if it’s fun or not. But once the basics of the gameplay are there, careful thought about balance and costing can save a great deal of time and effort over the more usual “keep trying stuff and adjust the broken bits”.