So in hindsight it was probably a bad idea to post a discussion thread and then disappear for a few days.
Oh well.
So, I've been seeing a rather alarming number of abilities being proposed, although happily the existing abilities are being discussed quite a bit more towards the end. I think we have a very reasonable pool to choose from now - at any rate, there are enough abilities that I like that I can justify saying so. Once again, I feel as though some people are not quite grasping the point of an ability that provides Risk, though this is hardly the community's fault - I believe that all abilities can, in some situations, provide Risk to their users, if only because using them prevents the use of another, equally useful ability. This can be seen in Doug's examples at the beginning of this thread - it can also be seen in Pokemon like Relicanth, who have a choice between Swift Swim and Rock Head - each negating one of Relicanth's core weaknesses (either its Speed or its inability to hit hard). Perhaps Yanmega would be another good example. However, in both of these cases we see very basic, powerful abilities, wherein, although granting a certain Risk in that the use of one, one must necessarily forgo the use of the other, I believe that there are abilities that can hold this general rule while at the same time providing disadvantages that are not necessarily linked to loss of advantages provided by the alternative.
Now then, as for what I like (as well as what I could like, but am not sure about) so far:
I would first like to reaffirm my support for No Guard. I absolutely love this ability and the way it plays for and against the typing - with a Pokemon weak to Stone Edge, Fire Blast, and Hurricane, alongside a vulnerability to such moves as Thunder and Hydro Pump, there is a real Risk involved with using this ability, or at least to any battler who wishes to play the probability game. That's not the only reason why I like this ability - one thing it gives, which is so very important to nearly offensive Pokemon, but to a sweeper especially, is reliability - there is no need to trust to the will of the RNG or any other such variable outside one's control. This is a quite significant Reward in itself, but is also notable in that the gross Reward this ability brings is entirely under our jurisdiction, by the movepool stage. The final thing that I really like about it is that it, to a very significant degree, removes luck-based Risk from CAP4, both in terms of the inaccuracy of moves super effective against it and in any mildly inaccurate moves it chooses to use, which I find intriguing from an experimental perspective.
Illusion is another one I wanted to talk about - and as I have often been a fan of psychological warfare and the concept of player stress in CAP, I like how some of the principles involved can be applied to Risk. Unlike Zoroark, CAP4 is pretty much covered in weaknesses and resistances, so pulling off a successful bluff would be difficult - and made even more so by the fact that Stealth Rock, likely as not, would reveal the secret as soon as CAP4 switches in. I'm less fond of it because of its inherent riskiness, as really the only Risk involved is on the choices the player and opponent make - so it is not entirely within the user's control - and more because of the way it interacts with the typing, such that it enhances the ordinary Risk/Reward element of Bug/Psychic. I'm fascinated by this ability and its potential applications, though at the same time I am a bit wary.
Moxie is the only other ability on the list that I feel has some sort of connection to the concept - not by much, though. The Risk from this ability, I feel, comes from the timing - if you KO a Pokemon at the wrong time, and a revenge killer forces you out, you aren't getting your adequate Reward. Which is an enormous shame. The Reward, as with No Guard, is largely found in increased reliability for offensive sets. I can't say that I find the Risk with this ability to be all that great, though, and I don't like the way that it is pretty much solely good for physical sets, as this means that purely special sets will take the other ability by default, rather than there being any Risk attached, or else will simply be less viable.
Others? Sheer Force is probably my next most favoured, but really, it has very little Risk attached to its use, to be honest, as most side-effects, aside from being generally so little game-changing that no reliance on their activation is planned beforehand anyway, happen to pertain to the probability game that I want to avoid. However, it has merit in that it to an extent reduces passive damage by removing LO recoil, and possibly the removal of effects also tends to aid the concept in the same way that it does for No Guard. Simple used to be something I favoured for the combination it makes with Stored Power and a variety of set-up moves, but I am starting to believe it is not really feasible, aside from there not being any real danger in the current metagame of any stat-down tactics being even remotely common, so any potential Risk is very greatly lessened, and as such, really isn't that good for our purposes.
There are also some that seem to be extremely popular, though I'm not really sure why. Here are some that I do see the appeal of, though. Weak Armour is the only one of the "Get hit by <X> Move and Get a Boost" abilities that I can really see working. Type-specific abilities such as Lightningrod, Flash Fire, Sap Sipper, or whatever else was suggested, are quite significantly limited by whether or not the opponent happens to have a Pokemon with that move, and whether or not they are likely to use it - this is less the case for some abilities than others, but it makes me favour them less, as they are pretty limited. Rattled is the least of these, as it incorporates three different types, but even so, intentionally walking into super effective attacks is a pretty stupid Risk - and unlikely to pay off unless this thing has both godly defensive stats and high Speed - at which point it probably wouldn't care about Rattled. On the other hand, Weak Armour has a suitably large attack range (everybody runs physical attacks) and it really isn't hard to get the boost. The drop is what makes me wary - I am worried that it may be too much of a Risk. I imagine that this boost would be best suited for sweepers, who take a colossal Risk trying to set up with Defence slashed. Which means the realm of this ability falls to general attackers, and this seems to go against the "can't sweep without boosting moves" note we agreed in Concept Assessment. I also worry that it may be shunted aside if we choose a less Risky ability alongside it, or that we may be unwilling to give it another competitive ability if it is chosen. It's still an interesting option, and that archetype of abilities is rather neglected.
Swarm is simple and sweet, but it is hard to get any kind of strategy going involving the ability as a whole, without becoming easy prey for priority and passive damage of all sorts. Unburden is really in the same boat, but is one-time use and as such can't really be described as an alternative to another ability on the same set, so that element of Risk is removed. This is a valid criticism of Swarm as well.
So... talk, I guess. And please do feel free to lobby me if you think I'm missing something, wrong, or just plain insane. That's what I'm here for, after all.
EDIT: Forgot to mention. Wonder Guard / Magic Bounce / other stupid abilities are off-limits. I was going to put Zen Mode here too because it's dumb and I'm pretty sure it's programmed to be Darmanitan-exclusive, but then I realised that that hasn't stopped us in the past. So I'd like more opinions on that, please.