The argument has a little trick card up it's sleeve:

It says "Believing in X" is rationally justifiable, not "X is rationally justifiable". Nonetheless, this isn't a very good argument. First of all, if one tries to formulate a belief by saying "I should believe in this because it has positive consequences" then one is explicitly acknowledging that they aren't actually "Believing in X" they are merely believing in believing in X. Here's a more helpful way to think about it. Replace the sentence "I believe in X" with the equivalent statement "I believe X is true" and we immediately see that one cannot begin with "I believe in X because X has positive consequences". If that were the case, you wouldn't necessarily believe in X. Believing X has positive consequences does not meet the criteria where we could say you "Believe in X". To believe in X you have to believe X is true. If you decide to believe X is true because you believe X has positive consequences then you have committed an ad consequentiam fallacy and therefore your belief is not rationally justifiable.

Isn't that just an example of where the consequences aren't positive?
Surely if the premise holds (that good consequences follow) then it is valid.
I think Todangst's criticism of it was that when people starting thinking that way, they started under-estimating the bad consequences of falsity.

You're right that someone couldn't come to believe something is true through the line of thought: "It is good to believe it, therefore I do believe it..."
That said, I think I can find some examples that evade your criticism.

If we were judging the beliefs of a person we were observing, we evaluate the rationality of their beliefs without their knowing.
So if a particular theist had a particular theology which had a positive psychological effect on them (good consequences) then an atheist observer would be able to acknowledge that this belief was good for the theist, even though it wasn't a true belief.
Because the theist doesn't hear this judgement (or doesn't believe the atheist when he says so), the theist still genuinely believes.
The atheist can then use the reasoning above to conclude that it's in the theist's best interests to believe, that believing in it is rationally justifiable, even if the theist does not know the justification.

Another point is how complex our minds our and how belief doesn't always come from our rationality.
We can allow ourselves to become "swept away" in an idea without worrying about the rational justification behind it.
There are methods we can use to make ourselves believe something, and sometimes we can use this to positive effect.
e.g. self fulfilling prophecies:
Someone with low confidence might have strong experience of failing and therefore the evidence would point to future failure.
But this would lessen confidence further and thereby make future failure even more likely.
It would be in their best interests to believe that they can, which would provide them with confidence and give them a better chance.
However, in order to do this they would need to somehow be able to ignore all the previous evidence.

Perhaps the argument could be seen as justifying taking a break from rational thought processes at appropriate times and places?

 DG beat me to the punch on the main point.  I'd like to add that once we start talking about "belief" being "rationally justifiable" we're opening up a gigantic can of worms that doesn't have any solution I can think of.

Look, belief is just holding something to be true, and it is not a choice.  It's an end product of the brain equivalent of algorythms chugging through data.  We are biologically designed as data processing and pattern finding machines.  If we find a pattern in a few bits of data, we are acting precisely as we have been designed to act.  As I've pointed out numerous times before, if a particular data processing machine has three things: a system of measure, an algorythm, and a goal, it can make a rational decision within that paradigm.  In other words, "rationally justified" in this context isn't well defined enough to make the argument coherent.

As DG has pointed out, beliefs are what they are, and it's very hard to address them other than to examine the object of the belief for empirical truth or epistemological validity.  If we decide to arbitrarily assign pragmatism as the measure of rationality in beliefs, sure, pragmatic beliefs are rational.  But that's just circular.  I can just as easily say that poetic beauty is the measure of rationality and beliefs, and then assert that any belief that is poetically beautiful is rational.  (After all, poetic beauty makes people happy!)  It's pretty hard to argue against any conclusion when you make it a given.

This sounds like an argument for the placebo effect. If a sugar pill can cure cancer by using the power of believing the pill were an actual cure, then I’d say it was justified. With X being religion, a lot of people need the placebo effect of religion to get them through certain times of their lives, whether it be the death of a loved one, the need to feel like a part of something or just the feeling that everything will work out in the end. What people do with their own lives doesn’t concern me as long as they don’t try to grind up those sugar pills and slip them into my morning coffee.