last post i hinted that the game i'm working on has deck construction, the type of play pioneered by Magic the Gathering where a player selects a set of cards before the game officially starts and then plays them out with some randomisation. part of why this works so well in Magic and its direct descendants is that it adds a singleplayer mode to a multiplayer game, so you can play the game without an opponent, but the singleplayer does not substitute for an opponent in the way a videogame AI might - instead it creates pressure towards finding an opponent as you build decks and want to try them out. the common problem we have making games for more than one player is that people "don't have anyone to play with" i.e. they're insufficiently motivated to ask their friends to play. (there is still stigma associated with the "nerdy" or "childish" activity of playing a game and for some people it is hard to overcome this.) but i am not making multiplayer games again yet, i've been interested in trying this mechanic in a roguelike game. this means it will lack the "metagame" of building your deck with reference to what your opponents decks might be, but it still seems compelling to optimise decks to play against random dungeons. (and it can still be massively multiplayer through the distributed discussion of deck construction, hopefully this doesn't mean it gets solved quickly and then everyone just plays the optimal netdeck or goes away.)

there's a lot in common between card games and roguelikes, in how you're constantly responding to random events, it seems natural to try to combine them. when i started thinking about this years back i hadn't seen anyone else really try it, but i doubt i was the first to think about it. now there seem to be quite a few examples around. i've played several unreleased ones people are working on. and then there's lots of games with names like "Card Dungeon" "Card Crawl" "Card Hunter" that i haven't gotten around to playing. (game developer curse: you end up playing way more unfinished games than finished ones.) also Dream Quest, though its dungeon grid is completely vestigial. (Deck De Dungeon has a similar battle mode while eschewing the grid entirely, which is to my mind more honest.)

i'd originally been thinking of using Dominion-style deckbuilding, where you start each game with a default deck and grow it throughout the game. this is a closer match for how roguelikes usually function, start with a weak character and gain abilities, it would just be adding a random element to when you can use those abilities. it is a nice system but i was finding it takes up a lot of space. the choices of cards to acquire and then the choices of cards to play, having interesting reasons for why you'd pick different alternatives in each case demands a lot of complexity, trying to fit that on top of roguelike tactical movement grid was too much. there's a reason why Dominion makes the entire game state be in the decks of cards. so i've settled on "constructed" play, build a combo then play it out in separate modes, get a score then maybe go back and tweak your deck again.

also thinking about Dota 2's ability draft mode. (ability dota is best dota.) you're building combos by selecting abilities that hopefully work together, but then there's still a lot of choices to make in how to play them out. you've drafted a character but then there are many possible builds for that character depending on the order you level your abilities, the items you buy, the role you play within your team. (hmm my game doesn't have a draft mode but maybe i should try that for multiplayer - take turns picking cards and then play it out and compare scores.)

i've written a lot of words and not even gotten to describing the actual problem i'm thinking about but i need lunch so i think i'll finish this here and write another post later.