To be clear, this type of profiling is about categorizing people psychologically. This is not "Muslim extremists claim credit for a terrorist act, therefore we should look for Muslim extremists" profiling. I have no problem with the latter type of profiling. I don't even call it profiling; I call it commonsense.
In comparison, the kind of profiling John Douglas writes about is where a crime with no discernible suspect is attached to a specific type of criminal; one discovers the type of criminal by looking at the psychology of the crime itself.
I have mixed feelings about this type of profiling. I love reading the books, and I'm a big fan of the first season of Criminal Minds (it got too yucky in season 2 for my tastes). Still, the "one random bit = conclusion" angle seems rather hit-and-miss, like Sherlock Holmes (whom I also admire) saying to Watson, "Ah, I know you walked here because of the mud splashed on your boots." Don't you always half-expect Watson to say, "Actually, I took a cab; a passing cart splashed mud on me"?
Here's my personal theory concerning John Douglas. I think he was/is (he's retired) the type of guy who could go into a crime scene and see what elements belonged to an ongoing investigation and what didn't. Through experience and pure talent, he could exclude the unimportant information and focus on the important information. He could see the forests and the trees but never get distracted by either.
The whole profiling conundrum arose when he decided that his ability was a science, not a gift. So he, and a bunch of other people, created these nifty categories and clear-cut applicable definitions, and I'm not just sure that can be done. (To be fair, Douglas does say over and over, "Don't be misled by superficial applications," but he doesn't seem to realize that not being misled by superficial applications has more to do with the nature of the man rather than the beast.)
I think Douglas' ability is legitimate. That is, I'm perfectly okay with him testifying in a jury trial: he has the expertise, the experience; he knows where-of he speaks. But I wouldn't let just anyone with profiler training testified. And I wouldn't let Douglas testify about anything outside his expertise.
I say this because although Douglas makes insightful observations about serial killers, including Jack the Ripper, in his books, his comments about "ordinary" criminals are surprisingly blah. Although he accepts Lizzie Borden's guilt, he insists on perceiving her in serial killer terms. I think Lizzie Borden was the original all-American/home-spun/no-frills crime-of-greed chick. Looking at her in other terms leads to all that silly "blacking-out" and fugue-state stuff. Not very helpful.
Douglas is right about Jack the Ripper (no, it wasn't the Duke of York) which means Douglas is good in his speciality. He can cut through the crap when it comes to what he knows.
I think the desire to generalize from the speciality--create a science out of one man's ability--is a desire that surfaces beyond law enforcement. You get a manager who is good at seeing the forest for the trees, good at pinpointing problems, good at separating the wheat (useful suggestions) from the chaff (stupid, wasteful solutions), and there's this "Hey, how do we duplicate this, so every manager is as good?" reaction.
And I'm not sure you can.
That is, you would probably duplicate techniques with someone who already gets "it," "it" being the talent or perception or whatever, but you can't really teach it to people who don't. It's like trying to teach irony to people who don't get irony or trying to teach conceptual thinking to people who think concretely (all you end up with is a bunch of people who want to make rules about using the "spirit of the law."). It's like (major segue into politics here), Democrats trying to win elections by duplicating Republicans and coming over kind of flat.
You can't duplicate people if you don't understand where their hearts lie.