'Ought implies can' is an old principle, usually attributed to Kant (although you can certainly find older arguments requiring something analogous). There has been an increasing tendency to criticize it in recent years. I think these criticisms are often extraordinarily naive, and tend to confuse the often sloganish way in which the principle is stated with what it was originally intended to mean. If we look at Kant, he is quite clear that it is entirely possible that you can't do what you ought to do -- Kant actually doesn't think it's the sort of thing you could know for sure one way or another. But to command something as a categorical 'ought' is to put it forward as something to be done, and trying to comply, practically speaking, requires taking it as something you can do. The point is that it is reasonable to treat the action as possible for practical purposes; but this is different from having a proof that we can in fact do it, because that would require a proof of free will, which Kant doesn't think we have. Thus it's reasonable to hope you can do what you ought, although you can't in fact guarantee that this is true. Or to put it in other words, an 'ought' morally and practically requires that we treat it as implying 'can', although we limited human minds are not able to prove that it does in fact do so for us in particular. This is quite clearly not a naive interpretation of 'ought implies can'.
But even more naive interpretations are not as naive as the versions sometimes criticized as if they were the way to understand the principle. Here is a case that, as far as I am aware, everyone who has ever claimed that 'ought implies can' has accepted could happen:
Jay has an obligation to do X at a certain time T. Jay fails to do X at T, and thus fails to fulfill his obligation.
But it is also quite self-evident that if you are not fulfilling your obligation, you cannot also be doing what fulfills it. Therefore if Jay is not doing X at T, he can't fulfill his obligation to do X at T. Therefore, one might conclude, he can't do it, so he has no obligation to do it. This would obviously be a problem for the principle, if it required a conclusion like this -- it would mean that no one ought to do anything they don't actually do, so 'ought' would actually imply 'does'. But, of course, this does not appear to be what anyone has ever thought the principle required, which is a sign that this is likely a bad interpretation of the principle. And the culprit (one that seems quite common among criticisms of the principle) would appear to be the idea that the principle 'ought implies can' requires that 'ought' implies every kind of 'can' rather than just some kind of 'can'.
Consider the following scenario:
Kay has an obligation to meet Jay at no later than noon. Kay puts off going to the meeting until it is physically impossible to meet Jay by noon.
Kay can't meet Jay; she's guaranteed she can't. Thus, one might say, it's not true that she ought to meet Jay. But this is not really any less absurd (in philosophy, the technical term for 'blatantly and hopelessly stupid' is 'absurd')as an interpretation of the principle than the previous one: she only can't fulfill her obligation because she's already violating it. That she can't fulfill it in this way does not imply that the obligation does not imply any possibility at all. And, indeed, we know that Kay's obligation was not an impossible obligation to meet; the whole set-up requires that meeting Jay at no later than noon is possible.
To be sure, the possibility is fairly attenuated; but this does not mean that it is not a genuine possibility. I cannot be at home now, for the obvious reason that I am currently somewhere else, namely, on campus finishing up required office hours. But I can be at home now in the limited sense that it is a possible state of affairs that is not inconsistent with my abilities: I could have stayed home rather than doing what I ought, and I am perfectly capable of going home at any point. And the reverse works exactly the same way. Suppose I were at home instead of on campus where I am required to be at this time. I cannot be at home and away from home at the same time and in the same way; therefore, I couldn't be doing what I ought. But the obligation is not linked to what I am able to do when I am violating the obligation; that 'ought' does not imply that 'can'. But this does not show that the 'ought' does not imply some 'can'. And, indeed, if it were literally impossible to do something, one might well say that this shows that you have no such obligation. For instance, if I went around claiming that everybody has an obligation to jump over the moon, it's perfectly legitimate to reject this claim on the grounds that we know for sure that no human being could possibly do so.
Consider another kind of case.
You have an obligation to meet Jay and Kay at noon. As it happens, God, who is omniscient, knows that you won't. If God knows that you won't, however, then you won't. And if you won't meet Jay and Kay at noon, you can't also meet them at noon. So you will not fulfill your obligation.
This works exactly the same way: the impossibility is conditional (it is impossible only given that you in fact won't), and is not in fact an impossibility relevant to that particular obligation.
It's likely that there are cases where the relationship between 'ought' and 'can' is non-trivial. But it is a confusion to hold that the principle requires that if you ought to do something, you can do it, simpliciter; 'ought' does not imply every 'can'. Things may be possible in one way and not possible in another way. And, what's more, you can entirely make sense of the notion that you might have different kinds of 'ought' depending on the different kinds of 'can' to which the 'ought' is linked. What you would need to counter the principle is not to find situations where you can't do what you ought, in some particular sense of 'can't', but to find an 'ought' that is linked to no particular 'can' at all. It would at least be hard work to argue that anyone ought to do things that are absolutely impossible in every way. Without such an argument, however, it seems that every obligation does imply some kind of possibility, even though it doesn't imply every kind of possibility.
(There are, of course, other questions in the vicinity of this, which complicate the question of how obligation and possibility are related. For instance, we sometimes treat trying to fulfill an obligation as if it were not significantly different from fulfilling it, and we sometimes don't. But if trying ever counts, then this changes what counts as 'being able to fulfill the obligation'. And there are a number of other things that suggest that we should also not be too facile in assuming what it even means to say that one can fulfill an obligation.)
