The twisting deckbuilding post a few days ago has clearly set something going in my brain, since for a few days whenever I read a phrase that sounds anything like "A game should X" my immediate thought is "But what if...not X!". Rulebooks should be clear and concise, a player should be able to read a rulebook and then fully understand how to play the game.
Now I can think of games that skirt this requirement on a technicality in one way or another. Mao does not have a written rulebook and players are encouraged to modify their imagined version on a regular basis. Nomic often winds up with an unwieldy set of rules at some point during play, but doesn't start out that way and it's not technically required (It's just something that tends to happen so that someone can force a win). I don't think either of those answers the challenge though, which I see as "Can you think of a good game that comes with a physical rulebook that no player could reasonably interpret?"
"Translating dead languages - the game!" Hmm, no that's more of a career path for people who like frustration.
What about generating a rulebook such that no player could remember it? In some of the Myth Adventures series there's a fictional game called Dragon Poker in which the rules are so long and convoluted that many of them are generally forgotten and the rules do not apply unless a player can remember them. I can see a poker style game where player's build hands, bet, show them and then the real skill of the game is in strategically remembering the right rules in the right order such that their hand is the highest scoring hand. Rule disputes would be a problem though, how do you check that a remembered rule is a real rule? Given that such a check might be necessary, how do you stop people just looking through the rulebook for rules they want to use while looking up something specific? That rulebook would need to have one hell of an index.
I just spend an unjustifiable amount of time looking for an image about how cool and fun indexing is. Did you know that the search term "Index Adventures" redirects to a company called "Index Ventures"? Some days the world is just that little bit more boring than you imagined it to be. On the other hand maybe I just suck at searching. My point is that this frog is cool and I should get on with talking about this hypothetical game:
So what we have is the notion that the rulebook should be too long to memorise and that the rules need to be really well indexed. How about if players are able to take actions, but there are so many actions that nobody could possibly remember what all of them did and had to consult the rulebook? The actions would need to be such that they supported a really quick indexing system, but such a game would have some significant advantages.
Okay, here goes "On their turn players may take an A, B, C...X, Y or Z counter or they may use three counters in their possession. To determine the effects of using these counters, players consult the tome to see which of the 15,600 possible effects are generated. The acting player retains the counters that they have used. The first player to obtain twenty counters wins"
The initial dominant strategy would probably be just to pick up a counter every turn rather than taking a "use counters" action, since the average random unknown action would probably produce less than one counter. However this strategy is also deterministic, the player that went first will inevitably win the game, so every other player may as well take the 'use counters' option in the hopes of hitting into something that turns the tides. This generates a really neat property of the game, in that there is an excellent catch up mechanic. Unless every one of the 15,600 rules has been memorised there is always a chance of the current loser hitting into the ability that is perfect for this situation, so there's always a reason to keep playing.
Once players have hit into a few useful rules I can imagine that they might use the same action fairly often. For instance triggering a "Draw two Y counters" action every turn to hit the requisite twenty in half of the time would be an obvious move, but again it's deterministic, the player who started doing it first will win so the other players need to try new things. That could lead to hitting into a natural counter like "Gain two D counters for every Y counter your opponent has" at which point both rules are negated. It's not worth gaining the Y counters because your opponent will get twice as much, but there's no point in playing the double-D rule because your opponent has no Y counters.
The important thing is that those rules are now known, because they were dominant in how the game was played earlier once they hit into other rules that interact with them when they find effects like "The next time you would gain Y counters gain X counters instead". Over time and repeated plays strategies of plays and counter-plays, even diplomacy and cooperation would emerge as a consequence of the particular rules that this group had encountered. In essence, the game would start out very simply, with essentially random outcomes, and gain depth and complexity as the players became more experienced.
The thing about it that's cool is that it would grow entirely around a local meta-game, depending on which rules were popular with a particular group of players. There are so many rules that if two players from, say different wargames clubs, were to meet each other and play they'd each learn a whole new way to approach the game from each other.
Of course it is in the nature of some gamers to scour the rules looking for the "best" abilities and then memorise those combinations, but it's also in their nature to look for the counters to those abilities once their opponents use them. Such players would run into exactly the same situations, but for them the best strategy would be the most obscure one. There'd be no point running the one that generated twenty counters in the shortest timeframe, because everyone has learned the "Destroy all H counters in play" rule to nix that one. So do you go with the second-quickest approach? Third-quickest? There genuinely wouldn't be a dominant strategy, because it would depend so heavily on what your opponent has learned about playing the game.
Or maybe it wouldn't work. Perhaps nobody would remember enough for any truly interesting strategy to come out of it. Looking up rules might be too tedious even if a three letter combo is very easy to reference well. It could be, people simply wouldn't hit into useful rules fast enough to get past the "first player wins" stage of the game before they got bored. Perhaps it'd risk getting crushed under the weight the deliberately inaccessible rulebook.
I have to say though, for a stupid idea I came up with just for the sake of opposing something 'obviously true' it is tempting to write it and find out. Perhaps I'll write a rule a day for it, they'll only take a few seconds each and then we can try the game out in forty seven years.