I spent seven years doing a PhD in Computer Science. It was an unpleasant experience from about year three or four on and I would have gotten the masters' and left before finishing if I knew then what I know now, and probably a) not gone through a suicidally depressive period b) have done more interesting work c) traveled more since I could afford it and d) also been paid more.

The educational aspects of a PhD, at least in my program, were essentially equivalent to the educational aspects of getting a masters degree in the same subject area. The difference between them was after you finished the educational aspects you were now expected to produce between 3 to 6 papers a year corresponding roughly to the important conferences in your field and get most of them published. Speaking personally it takes me about three to four weeks to write a good research paper assuming I already have the results, so I worked out at one point that I had about two weeks to generate meaningful work to write about if I was to publish to that schedule. It is impossible to do anything more than the famous Least Publishable Increment of Work in that environment, and as a result most papers in Computer Science aren't especially compelling. A healthier field (say, chemistry, physics, mathematics) has you write perhaps one or two a year, with actual data.