I haven't read the blog post Daniel linked to yet, but I would suspect that the reasons based on humanities/social science apply to us as CS people more than you might think.
As far as [REDACTED]'s points: I don't know of any CS Ph.D programs that will put you into debt either... well, sort of. I went from having ~ $5000 in student loans post-Wellesley to having ~ $40000 in student loans after being a Ph.D student at Portland State for 4 years (and not graduating). Why? Because my grant didn't pay for health insurance, so I had to pay for that out-of-pocket (and the cost of student health insurance went up from about $1200/year to $4000/year over that four years). I had a lot of other expenses, mostly medical, that I couldn't pay for just out of my grad student stipend either. So while a CS Ph.D program may seem like it's not going to cost you anything, you also have to think about the opportunity cost of spending 4-10 years making $15K-$25K/year, most likely with no benefits (since RAs and TAs at many schools are hired at 0.45 FTE so the university won't have to give them benefits -- the exception is schools where RAs and TAs are unionized, which anyone looking to go to grad school should seek out). Especially for CS majors, those 4-10 years could be critical in your career -- people who don't go to grad school spend them developing and gaining experience that you're likely going to have to start over with anyway if/when you leave or graduate and can't or don't want to work in academia.
I agree that being your own boss is very appealing, and it's a big reason why i once wanted to be a professor. But it's not quite right to say that professors get to decide what they work on, and especially not grad students. Applying for grants limits what you can work on and also means you have to do a little or a lot of "spin" to make your work sound like something that will "help the war fight" (at least if you're applying for defense funding, which is what a lot of CS funding is). For some people, this may feel dishonest.
Teaching and mentoring are also appealing. There are many jobs that have mentoring as a component.
I think there should be more women in CS, but I also wouldn't want to make any individual woman deciding on a career feel like she *has* to do something that's difficult, costly (financially and emotionally), and will likely expose her to various kinds of harassment just to further the cause of equality. Ultimately, it's men's job to stop driving women out of CS.
In the end, I think if somebody reads all of this stuff and makes an informed decision that they still want to go to grad school, then they should go. That's why I went, two separate times! I won't exactly say that I wish I hadn't, but I also wonder how it could have been if I had spent those years working on projects with clear goals and expectations where I would have been compensated fairly.