It's certainly easy to  make fun of it, but as a slightly different point: why do psychiatrists-- educated, intelligent, dare I say on the intellectual side of the field-- do this in the first place?

Pharmacologically, it really doesn't make sense.  Ritalin is a dopamine/norepinephrine and sometimes serotonin reuptake inhibitor, as well as some degree of catecholamine releaser.  Risperdal and Thorazine are tight D2 blockers.  So while antipsychotics won't negate the stimulatory effects of amphetamines (different dopamine receptors, etc) the amphetamines do compete with antipsychotics for their sites, thus reducing their effect.  (NB: less relevant in quickly dissociating antipsychotics (clozaril, seroquel).)

I think the reason it is done is language, i.e. the child psychiatrist can say to a parent: "I'm giving this drug for attention and concentration, and this other drug for impulsivity/aggression."  Even if, pharmacologically, this can't possibly work, nor does it even make any structural sense to separate the two classes of "symptoms."

But "working" isn't the goal.  Psychiatrists in this regard are asked to do something impossible: counteract a gigantic social/economic/cultural matrix of pathology with two to three medications of dubious efficacy, given to only one of the players in the matrix, i.e. the kid. 

The reason they are called upon to do this herculean task is that society lacks the language and the power/commitment to handle these social ills in any other meaningful way, so it repackages them as psychiatric illnesses.  In this sense, pediatric bipolar is underdiagnosed, not overdiagnosed. 

So the goal isn't better treatment for the "patients"-- listen up everyone, I'm letting you in on the Trilateral Commission's big secret-- it's keeping the doctors believing the problem is medical, not social.

As long as doctors believe the problem is theirs, it will be.