DSM diagnoses would be just like the signs of the zodiac
if it weren't for the fact they're very different in many ways.
In the Bentall profile, the New Scientist interviewer opens by asking if comparing the DSM with astrology isn't "a bit strong". "No" says Bentall. I happen to agree with him, but not with the reasons he gives. Comparing the DSM with astrology isn't "a bit strong" (criticism of something you dislike should be strong) it's wildly off the mark. Why? Well although you can derive some superficial comparison between the categories of the DSM and the signs of the zodiac (both describe classes of people; both aspire to some degree of reliable prediction) there are also clear differences.
The most obvious and important difference is the way the way the categories are derived and assigned. The signs of the zodiac are assigned to people on the basis of their date of birth and based upon the idea that these dates are linked to personality characteristics in a meaningful way. Meanwhile DSM diagnoses are applied on the basis of set of criteria describing patterns of behaviour. Someone designated as having, say, OCD, can be expected to resemble a particular broad set of clearly defined features. However a Libra is not just someone who is "on an even keel" (which may way not be an unreasonable classification in itself) but someone who was born between the 22nd September and 23rd October and is regarded as "on an even keel" in virtue of this fact. In short, astrology makes a needless jump--the linking of birth dates and personality traits--that the DSM doesn't.