You should read the article in its entirety, this guy gets it, I'll add only two short points specific to psychiatry.

Groopman acknowledges that medicine has considerable data and science available to it, but doesn't like blindly following "evidence based medicine" because it fails to take into account the nuances of the individual sitting in front of you.  Through a good interview, a doctor has access to extra information, often subtle, that an algorithm or "evidence based medicine" doesn't have, that can increase his accuracy rate.  Fair enough. 

Lambdin, however, points out that doctors are susceptible to all sorts of errors and biases in the application and detection of this "extra" information, that can make these augmented decisions  even worse than simply applying the algorithm blindly.  Touche.

The problem in psychiatry is a little different, however: here, psychiatrists make biased and error-prone judgments-- often simply prejudices-- that they use to augment studies and evidence  which themselves are faulty and biased.  It never ceases to amaze me how people are suspicious of drug company studies, but not at all suspicious of studies from the universities or the NIH-- they don't have any biases there?  Is it magic that allows me to predict that an NIH study will find the generic the best choice?   Researchers not only can influence the data-- mostly unconsciously, but also by not publishing a study that didn't show the expected results-- but also which questions to investigate, and how questions are framed.  Is there any possible reason Harvard, Yale, and Cleveland will always find a reason to study Depakote or Lamictal?  And now Seroquel?   But not Zoloft (now generic?)

An example of this is the the doctor who has a patient who has no history of bipolar, and any rating scale or screen would not suggest bipolar, but uses his clinical skills (read: bias) and says "there's  something about him" and "discovers" that the patient is an undetected bipolar-- but then puts him on Depakote, because that's what the "evidence" shows is the first line for bipolars.  Really?   Which studies did you read, again?  The orange ones?  In the Trapper Keeper?

As anyone who has ever dated a girl who was too much into the occult will tell you, astrology is difficult. It has a highly structured set of rules-- math, really-- so precise and complex that, theoretically, any two astrologers should independently arrive at the same result, which is correct enough times to keep people from breaking out into hysterical laughter all the time.  However, astrology is crap, right?  Some other factors explain the few successes.  Is the fact that so many schizophrenics are born in the spring related to Mars rising in Orion, or to a virus women contract in the winter?  Etc.  In other words, just because a system is reliable, doesn't mean it's valid.

The other point is on the subject of "zebras"-- rare, outlier diagnoses that should be investigated last after more common and likely ones. Lambdin writes:

Oddly, Groopman rebukes doctors guilty of “zebra retreat,” but bungles the example this term is derived from. He quotes: “When you hear hoofbeats, think about horses, not zebras.” The actual lesson is, “When in Wyoming, if you hear hoofbeats and think you see stripes, it’s still probably a horse.”

In psychiatry, however, the problem isn't suspecting a rare disorder before a common one; it is inventing a disorder rather than dealing with the complexities of a person's life.  "It's not bipolar--he hasn't responded to Depakote or lithium-- I think it may be Intermittent Explosive Disorder."  If you apply the same rigor, analysis and logic, alien abduction is also on the differential.  As I've written before,

the real problem of a critique of our cultural models is to ask, when we see a unicorn, if by any chance it is not a rhinoceros.

-- Umberto Eco

(Thanks to reader Walter F for sending me Lambdin's review.)