...and, following on from that, how I really struggled to come up with one that was, and was fun.
I'll be referencing what I shall refer to as "the L4D Boardgame", which doesn't exist, but various design notes and thoughts can be seen here.
In summary, after playing through Doom and Descent campaign games for 2-3 years, I came to a number of theories about boardgames;
1) Co-Op games are rarely that. Most de-generate into a single player ordering the others around. There is rarely any individualism, and the players combine to form a gestalt, typically with a single leader doing 90% of the game, while others are simply piece-movers, and dice-rollers.
Typical Culprit : Pandemic
2) Dice are quick, yet inherently random when used in low numbers. Too often success or failure comes down to luck rather than good play. Boardgames often have low volumes of rolls compared to (for example) wargames, where dice are rolled in 10's, rather than 1's, and thus the bell-curve of probability often does not properly develop.
Typical Culprit : Arkham Horror
3) Player elimination should be avoided whenever possible. It's no fun watching others play.
4) A game that can be lost by everyone should not take too long to play (it's no fun). Also, individual turns should be quick, keeping downtime for individual players to a minimum.
Typical Culprit : Descent
With Co-Op games it is a "us against the system" style of play, and results can fit into 1 of 3 brackets
1) System is impossible to beat
2) System is possible to beat, however error/luck causes failure
3) System is possible to beat, and skill/luck allows win
Looking at Pandemic again, often the lie of the card decks means that victory is impossible, whereas other times it may almost be a certainty. The game is fairly effective at disguising these scenarios, however at the end of the game there is no way of knowing for sure if you ever had a chance of winning, or if victory was almost pre-determined. There ideally needs to be some way of reducing the impact of luck (while not eliminating it, otherwise optimal routes are found and following, leading to a rail-roaded playing style), while maximising skill returns, and also reducing "System cannot be beaten" chances, so you end closer to the "ideal" state of;
1) System is possible to beat, however error AND/OR bad luck causes failure
2) System is possible to beat, skill allows win (maybe a bit of luck)
So I challenged myself to come up with a game concept that rewarded skillful and true co-operative play, didn't take too long to play, and was fun. I fucked up. The basis was taken from Left4Dead, which is an exceptional co-operative game in that it didn't just involve 4 people running the same way, but player interaction and support was required (and fun).
The initial concepts can be seen on the Wiki, but boiled down to;
1) No dice, instead cards would be used which held dice roll results, ensuring that a given player would always produce results within a standard probability bell-curve. They could choose when to roll well, and when to roll badly within certain parameters (hand size/deck cycling etc)
2) Individual secret player goals to stop gestalt play.To win the game to your own goals, you also had to support the team, or risk failure.
3) All players on same side (no "dungeonmaster"), requiring a simple and robust system for the opposition monsters
4) 2 levels of zombie. The "horde" zombies, which were due to be mowed down, and "specials", which would provide a challenge. These would be introduced against the players by some method to keep challenge high, yet close (the mythical "sweet point" for a game, which I think we hit once or twice within Descent and Doom, where both sides feel victory can be reached if played well, rather than a cakewalk or butchering)
5) Random boards. These speed up gameplay, by minimising set-up time, as well as increasing re-playability.
I bashed together some rules, based on hexagonal floorpieces, and generation of a map based on drawn cards. Challenge would be provided from a "Director Deck", which would vary the level of opposition based on the overall state of the players (for example, a well-tooled, healthy team would be more lightly to meet swarms and specials, while a weak team may be lucky enough to find a health-pack). There was a basic time system for introduction of special zombies, and some basic ideas for individual goals (the team goal being to reach the end), and even a way to link them into a campaign basis (so in return for XP you could either increase the ability of your character, or select potential win-goals).
I did a couple of solo playtests, and the immediat eissue that came up was that re-presenting a swarm of zombies with models is extremely slow, even moreso if you want to have a good visceral feeling of mowing them down. Specials were very hard to make effective in the L4D-style (a simple automated set of move rules does not really properly translate the challenge of taking down a smoker, whereas simple zombie horde moves of "approach closest and bite" is fairly quick).
Balancing the guns was a complete nightmare...
I put the project onto a back-burner, though since then I have had some other thoughts. Rather than re-present all the models in a horde, instead you have an abstracted phase where the players combine their firepower to "suppress" hordes, and only where the suppression is not enough then models are placed, such as;
4 players each have a hand(clip) of cards (ammo), with values from 1-4, and possibly "MISS". In the zombie phase a number is determined (probably from the Director Deck) to indicate the number of targets visible. Players must then add in one card from their hand hidden to counter the threat (think BSG challenge overcome). These cards are taken off the number of zombies, and then remaining models are placed, and become much harder to kill (think of these as the swervy bastards that get past your shots in the computer game). This reduces the number of models to move, adds in more "co-op" (and fucking over the team if your individual goals require it).
For Specials, I think it would probably be better to ignore Smokers, and stick with "approach and attack" type zombies, as the move rules are far simpler. This is the route that games such as "Last Night on Earth" have gone with...