It quotes a Harvard psychologist as complaining that such effects are "by definition beyond the reach of science" - a foolish statement which Jonathan Ichikawa rightly calls him on.
As a naturalist myself, I look forward to rigorous research being done on this topic, to help convince the gullible that prayer really doesn't make a fig of difference. The current lack of hard evidence either way merely gives them an excuse to indulge their wishful thinking. So I don't understand why skeptics would be opposed to such research in principle. (In practice, of course, it's a bit of a waste of money - 2.3 million federal dollars could surely be better spent elsewhere.) It sounds like the kind of issue which CSICOP might have more info on.
There are a variety of variables which I'd like to see tested. For example:
A) Content of prayer: (1) to help or (2) to harm the patient.
B) Target of prayer: (1) Christian God, (2) Allah, (3) Hindu gods [etc.]
C) Sincerity of prayer: i.e. (1) prayed by a true believer or (2) prayed by a non-believer (of the targeted religion).
+ of course the control group, which doesn't get prayed for at all.
You could mix and match various combinations, taking one option from each row. Then you could do it all over again, except this time the new combination is not what gets done in fact, but instead is what you tell the patient. [Though this would square the total number of groups involved, so this 'ideal' experiment is fast becoming impractical. It would no doubt require a huge sample size to get enough people in each group to allow any statistically significant results to be inferred. So a simpler version may be preferable.]
For example, a patient might be told he is being prayed for (positively, by a Christian, to their God), when in fact an atheist is praying to Allah to strike him dead. Or vice versa. (Say, do you think an experiment like this would have trouble getting ethics approval?)
It would be interesting to see how the different combinations played out, so we could learn just what factors actually have a influence.
I'd be willing to bet that the fact of the matter makes no difference whatsoever. But what the patient believes (and his expectations regarding the results of this) might well influence things, as a form of placebo effect.
Incidentally, I think the genuinely religious should be embarrassed by the 'Santa Claus' conception of God implied by the 'believers' behind these experiments.
Chris Mooney suggests that "[t]he real reason intercessory prayer research is absurd is that[...] it not only assumes supernatural forces, but assumes that those forces will be benevolent". Not at all. A benevolent God wouldn't wait for people to ask him to help improve the world. He'd improve it right away.
Jason at Evolutionblog sums it up nicely:
It's hard to see how such studies could actually end up benefitting religion. If they are unsuccessful, that would be evidence that prayer is ineffective. But if they are successful, you are left with a fickle God who makes life and death decisions based on who asks him nicely for intervention.
Lastly, go read Positive Liberty's take on "what happens when you pray to the wrong god". Only semi-related to all this, but very funny.