Regarding joy of programming/playing: I play games for fun and challenge, but programming good games is also fun and challenging. And nothing beats designing a game, programming it, playing it, and discovering that I have created something great. I have won my Hydra Slayer several times with different races, and I think this was very fun. Interestingly, I have also tried to determine what set of weapon is the best for fighting very big hydras in Hydra Slayer. This was done with a program, and, to my surprise, the best sets found were very different that I have thought, apparently very sophisticated (they included 17-divisors, complemented by smaller divisors and meteorite blades in extremely smart ways, so even if one of the divisors caused doubling, you could later use a combination of other weapons which could fix it; and I thought that small divisors are generally better that larger ones). I am happy that there is so much emergent complexity. Now, I could set new Hydra Slayer challenges for myself, but there are so more other challenges to do: other roguelikes to beat, new roguelikes to write. So I mostly concentrate on writing Vapors of Insanity and NotEye, and on playing other good roguelikes from time to time.
Re rolls vs stats. Stats, and calculations behind them, this is an important design decision. There are lots of games which show you with lots of numerical stats (accuracy, dodging as a response to accuracy, damage, armor reduction and protection as a response to damage, armor penetration and critical hit chance as a response to armor protection, luck as a response to critical hit chance, all this split into 7 types of damage, whatever). Too many to think about, and there is no reason to have that in a board game, and even in a computer game usefulness of all that is dubious if the hidden or difficult mathematical rules make it hard to tell what weapon is the best against the given enemy. I have made some sophisticated rules in VoI and I am happy with them (experience system based on Elo's ranking, hit/dodge chances based on Cauchy distrubution instead of dice rolling which is overused in computer RPGs for no good reason, and "weapon class" as a third important feature in addition to damage and accuracy). But I am even more happy with clear but complex mechanics of Hydra Slayer. In a computer game, the developer has a choice of either creating a system based on real life intutions or beautiful mathematical principles, or something simple: as shown by some games by Jeff Lait and Darren Grey, and also Desktop Dungeons and Mage's Guild, you can get a very good game by making the battle system easy to understand, which goes as far as removing not only stats, but also dodge chance, and even HP. In case of a board/card game, you cannot get realism, since it is impossible to perform the necessary calculations, so aim for simplicity.
Hit points: As I pointed out above, you can avoid hit points and obtain a good game. The idea to lose cards instead of losing hit points sounds very nice, an original approach instead of the overused system.
Map: having a map certainly makes it more roguelike. But to have a game that can be played without a big table is as worthy goal as to have a roguelike. Maybe a small map, similar to those found in these pocket magnetic chess sets? Could be nice, but also would make it impossible to produce... (although these pocket magnetic chess sets are cheap, so maybe you could use them, have black pieces for dungeon generation and white pieces for monsters and players, or something like that, I like new games like Arimaa which reuse the pieces of old games in novel ways instead of requiring new sets).
Alliances in board/card games: I think that, if you have a game with more than two players, all of whom wants to obtain the highest score in the end, and where you directly fight other players, then this tends to make the actual rules of the game irrelevant, as everything is reduced to forming alliances and backstabbing, like in the classic board game Diplomacy. If you like such things, then I guess it is OK, but I suppose it gets boring after some times. The card game Dominion has cards which allow you attacking other players, but (IIRC) whenever you attack, you attack all the other players at once. This prevents it from becoming a new version of Diplomacy. BTW I think that Dominion is a very good game, and it has a certain very roguelike element: at the beginning of a game, card types which will be available are chosen randomly and shown, which corresponds to random equipment generation (just like in roguelikes you need to adapt your strategy to the equipment you find, in Dominion you adapt your strategy to the cards available, every game is different).
Low direct interaction in most recent Euro style board games is one solution, but indeed, I think it is a rather boring one (I think Dominion qualifies and I like it very much, because every game is different, but I find most other such games not very interesting). This can be solved by having only two players, only two teams, like in Bridge (2 vs 2), Mafia (informed minority vs uninformed majority), or a cooperative game.
