I've been thinking a lot lately about fragile games. By this I don't mean games with easily-damageable components, but games in which the game state is itself extremely malleable, changing in response to player decisions. A highly fragile game is capable of being systemically skewed, with certain actions becoming easier or harder (and more viable or less), and the pace of the game becoming faster or slower, all depending on what the players have done previously.
You don't actually see this kind of thing a lot in your standard move-combat wargames (though there is some degree of malleability there). It's more often found in economics games like Container, the 18XX, or Cole Wehrle's An Infamous Traffic. In Mr. Wehrle's game, player actions not only dictate the demand for the opium (and thus their opportunities for profit) but also the political stability of China. Because of this, it can play very differently from one session to the next.
I find this sort of thing utterly fascinating as a player, this sense that my decisions, and yours, can fundamentally alter the decision space. I don't get to play with it much as a designer, as it's very much a feature of games that seat three or more, and most of my games are just for two. Arguably and somewhat ironically, I've mucked about with something similar in my solitaire design Agricola, Master of Britain. In that game, the contents of the three tribal cups is gradually and systemically skewed in response to the actions you take. If you utilize a military solution to every problem, for example, the tribal chits will lean heavily in the Hostile direction, making the game harder.
One of the "problems" with fragile games is that because they have these kind of feedback loops built-in, leads tend to widen over time; a player who gets ahead in points or money or widgets or what-have-you tends to stay ahead. Players are able to nudge and push the game state to diminish that lead, but only if they do so in a timely fashion; advantages, once established, can be increased quickly and exponentially. Of course, the player who's taken the lead is going to happily capitalize on his privileged position, pushing back hard against anything that might level the playing field. What makes it even trickier is that by the time players realize there's a problem, it might already be too late; opaque butterfly effects are the bread-and-butter of fragile games.
I don't really see that as a problem, myself - it's not a bug, it's a feature. There exists the ever-present threat of someone gaining an advantage, and everyone knows that if that happens, it's because we let it happen. If good play is rewarded, and bad play mercilessly punished, then everyone at the table is challenged to bring their "A" game. It makes for a rich, intense, exciting, and involving play experience.
But it's not for everyone. I mentioned the 18XX in the above, and while I love the genre, it's extremely hard to get one on the table. In fact, every time I've played one, it's more-or-less been with a completely different crowd. After the first game, no one's wanted to come back for a second helping, my long-suffering spouse being the sole exception, and I think that's less because she wants to develop the perfect train rush strategy and more because she doesn't want me to mope. A lot of what turns people off is not so much the opacity or the theme - the Winsome train games especially are good about distilling the essentials of 18XX, including in some cases a high degree of fragility - but the sheer epic length of the games. While having lost a game isn't necessarily terrible, losing the game for eight hours without any chance of a comeback is the pits.
But the threat of my position being rendered unwinnable if I play badly gives my decisions real weight and meaning. It's something that I need to guard against, as does everyone else at the table. To tie this a little more directly into wargaming, Twilight Struggle has a wonderful mechanism by which a player can trigger an atomic holocaust, resulting in instant defeat. In order to get there, things need to escalate dramatically, and since neither player wants to risk losing the game in that fashion, in my experience they'll both be very careful. My assumption is that the number of games of Twilight Struggle that end in thermonuclear war are generally very few in number, simply because the players are working to avoid it.
But instead of being a clever sub-system or something you have to worry about once in a while, in a highly fragile, highly malleable game, you're worrying about this all the time. It's great.
Like I said, since most of my designs are two-player wargames, these aren't concepts that I myself get to play with very often. That said, Optimates et Populares, my game about the political struggles of the Roman Republic, was created with this sort of fragility in mind. The amount of Political Will each player has is dependent on his opponent's actions, and his own reactions to that. In some games, PW can be hard to come by, and in others, you can find it just laying around - but that's all dependent on the aggregate of the back-and-forth.
Scoring in Optimates et Populares is cumulative from turn to turn. That is, at the end of each Game Turn, you look at the game state - who has what offices, the current position of the laws - and you score VP accordingly. So, if you have a Law on the "2 VP" space, you're going to score 2 VP each and every turn, not just once. Because of this, if one player is scoring more VP than the other, he's going to keep scoring more VP, gradually and systemically widening his lead.
This isn't great for his opponent, and so he needs to try to narrow that gap - knocking that Law from the "2" to the "1" space, for example. More than that, however, he needs to work to ensure that gap doesn't occur in the first place, because it's something that's going to quickly compound over time. The closer he can keep the scores, and the better job he does maintaining that closeness, the better his chances of being a viable contender in the end game, and/or in reversing the situation in his favor. He also needs to watch for an "income gap" - if your opponent's out-earning you two or even four-to-one, you're going to have a really hard time of it.
Given the inherent fragility, and the cumulative scoring, a mistake can be very costly. And because the game is fairly interactive, with for example your decision to attempt a piece of legislation being subject to veto, and your opponent's veto being subject to being overridden by the appropriate bribe, and each player's decision being somewhat influenced by what they think his opponent will decide to do, and how the costs balance out with the benefits - well, it can be hard to know what's good play and what's bad. Because of this, even though there are only six pages of rules, and most games run about an hour, I would describe the game as being on the heavy side.
There is one bit of advice I'd give to anyone playing Optimates et Populares. Try to get all three Major Offices if you can, but if you can't, then you want to have one Consul and a Tribune. If it's split the other way, with one player having both Consuls and his opponent the Tribune, the game can become much less malleable, and will generally reward whatever trend is already in motion (though, because the game isn't entirely symmetrical, with each of the two factions having different advantages, this is far from an absolute). So, you don't want to have both Consuls and not the Tribune, or vice-versa - except for when you do.