Team Obama was winging it when it declared the stimulus would "save or create" 2.5 million, then three million, then 3.7 million, and then four million new jobs. These were arbitrary and erratic numbers, and they knew there's no way to count "saved" jobs.
Rove is certainly right that we have no precise way to estimate how many jobs are "saved" or created as a result of the stimulus bill, but the exact same critique applies to the Bush administration's bogus claims that President Bush's economic policies created [X] million jobs. In both cases, we want to estimate the causal effect of an economic policy. However, that requires us to compare the observed number of jobs with an unobserved counterfactual -- the number of jobs that would have occurred without the policy in question. We can try to estimate the latter quantity, but we certainly don't know it with certainty.
At a more general level, it's worth pointing out again that the Bush administration broke new ground in its use of misleading statistics to promote its tax and budget policies. (We devoted two chapters of All the President's Spin to the subject, and it only covers Bush's first term.) The idea that Karl Rove is suddenly concerned about the purity of economic statistics is laughable.