As for what games are CSI games, he has three lists. Games which defined themselves as CSI games include his own game "Cranium Rats"; "Apocalypse Girl" (by Sydney Freedberg); "The Dragon Vs. The Gun" (by J.J. Prince); "Gnostigmata" (by John Kirk); "The President is superf***d" (by Matthijs Holter); "Threads" (by Filip Luszczyk); and "Fantasy Game Engine" (by Andrew Cooper). The concept is mentioned in "Shooting the Moon" (by Emily Care Boss); "Agora" (by Joshua BishopRoby); "Bacchanal" (by Paul Czege); "The Dynasty" (by Adam Kleizer); and "House of Card" (by Brennan Taylor). There are two or three games which he suggests may be related: "Capes" (by Tony Lower-Basch) and "Final Stand" (by Tim Denee).
This is mostly documented in the CSI Games Wiki.
So Guy asked for general thoughts on the blog and wiki. Now, I've never played any of the games listed in the wiki. I've played Shab Al-Hiri Roach, which Guy suggests may be CSI-like. I've also played Pantheon, The Extraordinary Adventures of Baron Munchausen, and Once Upon A Time. On the other hand, I've also played many tabletop RPGs with in-character competition like Paranoia and Amber Diceless.
I generally enjoy in-character competition with a few caveats. For tabletop RPGs, I prefer for the player characters to remain in communication -- to allow for frequent in-character dialogue between the players. Separated PCs often reduces communication to only one player to GM, which is less interesting to me. While I don't mind players having secrets, I don't like note-passing because it is too slow compared to the spoken word. This doesn't apply to online play or larp, though. Basically, as long as the competition doesn't shut down inter-PC communication, it usually makes the dynamics at the table more interesting. So, for example, I love Paranoia where you have a team of characters who all are trying to backstab each other but have to work together to complete the mission.
However, I'm ambivalent about games where the competition is made into a formal, mechanical contest between the players. In my experience, games like Baron Munchausen or OUAT are poor arenas for true competition (as is Paranoia, but it doesn't imply that it should be). Basically, if you as a player actually play to win, the game doesn't really work. Elliot, who also played in the recent Roach game, commented on my last post:
The weakness of Roach was the rules problems; that's compounded by having a win/loss condition, which invites attempts to argue the rules and bad feelings if you don't get your way. I think it might be better to feed Reputation directly back into game play so that it can affect the flow and the outcome of conflicts, and then to just define the Endgame outcomes in plain English instead of win/loss--taking into account, for each character, whether the Roach takes over the world, whether that character is Roach-bound, the character's Reputation, and whether the character still has any Enthusiasms.What Elliot describes is roughly the way that My Life With Master handles its Endgame. Depending on your stats at the conclusion of the Endgame, a player's minion will come to certain endings (like dying, living happily ever after, or becoming an evil master of their own). However, none of the endings are defined as "winning".
I'm interested enough to play RPGs which define a winner. However, I see very common, perhaps insurmountable pitfalls to true competition. The first is relying on self-limiting. For example, as I understand it winning contests in the Roach is easy: just narrate in a whole bunch of NPCs on your side. It advises you not to push it too far, but there is no penalty. This basically rewards being a jerk: stretching the group contract as much as you can. This applies similarly to using equipment dice in Dogs in the Vineyard -- it's free, so the only limit is how much you want to push things.
The other common pitfall is becoming a popularity contest. A number of games (including board and card games) rely on player voting. However, unless that is narrowly constrained, people will tend to simply vote for the people who they like anyway. This isn't necessarily a problem, but it's a tendency I dislike.
Now, it could be that there is a design which gets around these. I play such games and I'll try out new ideas, but I don't like continually butting up against these two issues. In part, I wondering what the purpose of the competition is. Competitive games are often designed to train -- to develop particular skills, like the tactical skill in chess, or the ability to read other people's expressions in poker. Winning then gives a feeling of accomplishment since your skill was measured against others. For imaginative works, though, I think there isn't the same feeling since you can't rate them as objectively.