> I don't think there are any memorable PhD theses that significantly changed anything, save for Shannon's.
I think that's going way too far in the other direction.
I certainly know of PhD theses in my research area that have made contributions to the area, and these contributions (or derivative ones) are very likely to make it into real products eventually.
A lot of PhD computer science students have the benefit of taking 5 years or so to really work hard on something consistently that would be very hard to accomplish with the same level of rigour in industry.
I think that's going way too far in the other direction.
I certainly know of PhD theses in my research area that have made contributions to the area, and these contributions (or derivative ones) are very likely to make it into real products eventually.
A lot of PhD computer science students have the benefit of taking 5 years or so to really work hard on something consistently that would be very hard to accomplish with the same level of rigour in industry.