For example, sales guys are compensated based on sticker price, not profit margin, which means that the guy who sells $2MM of work for $1MM (i.e. at a loss) gets more commission than the guy who sells $0.5MM worth of work for $0.75MM. For another example, grading programmers by lines of code... 'nuf said ;). Another totally ubiquitous example: grading office workers by ass-in-seat time rather than value produced. Another also ubiquitous example: Departments in a company encouraged to spend as much money as they can at the end of the quarter, so their next budget doesn't get cut.
I could go on, but I'll stop here. The point is, even if "publish or perish" refers to misaligned incentives, that still means it's basically the same as 90% of other jobs out there.
It's not so much "publish" in itself that's the problem[1], it's how publishing is defined and handled, and that a certain form of publishing is seen as right to the detriment of any other form of published material, whereas the preferred method is often not the best method for a given project.
[1]: though it certainly is a big part of the overall problem of uniting teaching, research and publishing