Dave did a good post about nofollow sculpting, and he got a very good quote from Matt Cutts:
Nofollowing your internals can affect your ranking in Google, but it’s a 2nd order effect.
My analogy is: suppose you’ve got $100. Would you rather work on getting $300, or would you spend your time planning how to spend your $100 more wisely.
Spending the $100 more wisely is a matter of good site architecture (and nofollowing/sculpting PageRank if you want). But most people would benefit more from looking at how to get to the $300 level.
Ok, so it’s of a “2nd order effect”. Let’s say, you could increase your traffic by 2% if you did it (and for the record, to my experience it’s a bit more, more like 10%). If you have 2 million visitors a month, 2% more visitors are 40,000 visitors a month, 480,000 visitors a year. Let’s say your conversion rate is as low as 1%, with a revenue per order of $25. That’s not too much, is it? Well… It’s $120,000. I can spend a few hours doing some sculpting for that, and still make a profit…
There’s of course a lot of “mis use” of PageRank sculpting. A lot of people are trying to compensate for bad site architecture and information architecture by nofollowing a lot of links, I’m even guilty of that on this site. Or people are calling the whole process of creating a good site architecture PageRank sculpting. It’s not. Good site architecture is one of the few main pillars of on-site SEO.
So, in all, PageRank sculpting is a nice technique. It’s useful for sites which have a lot of traffic to play around with. I agree with Matt, that if you have $100, you shouldn’t start with it. If you have $100,000,000 though, I’d definitly put it on my list.