The Liberal Democrats’ deputy leader, Menzies Campbell, said the government should use its influence with the US president, George Bush, to secure [the] freedom [of nine Britons detained for more than a year at Camp X-Ray and at Bagram air base after being taken prisoner during military action in Afghanistan].
The Guardian, 25 April 2003

The principle that “ought” implies “can” is usually attributed to Immanuel Kant, although he never actually said anything quite so pithy. (See chapter eight of the Critique of Practical Reason for his more convoluted expression on the idea.) Whether we credit Kant with the discovery or not, the principle itself is pretty self-evident. It makes no sense to say we ought to do something unless we can actually do it. It is absurd to say “you ought to be eight foot tall” or “you ought to eradicate world poverty by dinner time” since neither of these are genuine possibilities. How can you have a duty to do what is impossible?

The logic of the principle is clear enough, and frequently ignored. People call on politicians to do what is not within their power, or athletes to perform above their capabilities. For instance, in Britain, many people felt that the world’s greatest woman distance runner, Paula Radcliffe, ought to have performed better at the Olympics, where she retired from the two races she competed in. What they don’t seem to have taken seriously is the probability that Radcliffe was performing as well as she could. For whatever reason, she wasn’t able to win on those days.

The example of Paula Radcliffe is instructive, because although the “ought” implies “can” principle is crystal clear in theory, in the real world, “impossible” carries more than one sense, some looser than others. Radcliffe could never have run the marathon in one hour, not because it is logically impossible but because it is physically impossible. That means there is nothing logically wrong with saying she ought to run that quickly. Rather, the problem is a failure to appreciate the facts concerning what is physically possible.

Where things become more tricky, and more interesting, is that in ordinary discourse it is perfectly legitimate to say something is not possible even though there is nothing either physically or logically preventing us doing it. Radcliffe, for example, could almost certainly have finished the race if she had put her all into it. To say she couldn’t do so, and that therefore we are wrong to say she should have done so, is to say that the obstacles were such that it would be unreasonable or unrealistic to expect her to have done so.

That may seem to extend the reach of “ought” implies “can” too much. We have moved from can’t (logically) through can’t (physically) to can’t (realistically) where the notion of what is “realistic” is somewhat vague. It would certainly be accurate to describe this version of the principle as an adjunct or extension of the core one and not simply a corollary of it. But I think we need something like it in order to make the principle really effective in critical thinking about the real world.

Consider, for example, Menzies Campbell’s statement about what the British government ought to have done about the prisoners at Guantanamo Bay. It may be that he was reported a little loosely, and what he meant to say was that the government should use its influence to try to get the prisoners released. Whether you agree with him or not, there is nothing incoherent about that. However, even if that is what Campbell meant, there were plenty at the time who thought the government ought to do more, namely, actually secure the release of the prisoners. And it can only make sense to say it ought to have done that if it actually could do so. But it is not at all obvious that the government could realistically have done this. Certain measures, such as threatening a trade war or resorting to other extreme measures, remained physically possible, but politically impossible.

We can see the “ought” implies “can” principle as therefore having two versions. One concerns the link between duty and what is logically or physically possible and should be uncontroversial. But arguably the more interesting and useful version, though also the more controversial and imprecise one, is that we cannot say people ought to do what “realistically speaking” they cannot. Any controversies such a principle would generate I would suggest are more a consequence of how we define what is realistic, and do not show any intrinsic flaw with the principle itself.