I've been reading his 1999 book A History of Modern Computing, grabbed from the library pretty much at random, which has a very nice overview of the U.S. side of computer development from UNIVAC onwards. The surprise was his apparent anti-UNIX bias. One of the apparent faults of early UNIX is that it was closely associated with networking so "it made it easy to write programs that acted like a 'virus'". Another apparent fault was that circulating the source allowed people to create other, potentially incompatible, versions. Instead the author hails MS-DOS as a real (author's emphasis) desktop standard. This is a bizarre choice as he also complains about UNIX's difficult command line interface. The complaint against openness also destroys one of his earlier points where he points out that Rockford Research Institute's tight legal control of its TRAC language likely led to its downfall, in comparison to the then contemporary and more open SHARE (collaboratively developed) and MAD (University of Michigan) languages. I am assuming that there was a long delay in submission and publishing as there's no mention of Linux (actually the book coverage seems to stop at about 1995 with the formation of Netscape).
Having said that, if you want an overview of 1950's - 1980's hardware, computer languages and system software it is a nice book.