He acknowledges that Doctorow, Pilgrim, and Payne have a point and then fails to address the real issue
Except the "real issue" isn't actually an argument either. There is a tradeoff happening here: for the average non-programmer, devices which have at least a certain level of "no user-serviceable parts inside" can be made to be much simpler, much easier to use and at least slightly more secure (in that attacks which depend on duping a user into installing/executing malware will fail) then devices which are completely "tinkerable".
The problem lies in failing to acknowledge that this tradeoff exists or that it should be considered on its merits rather than ideologically rejected out of hand; this takes it from the level of rational argument down to the level of blindly applying dogma.
Most of the rest of the "argument" is merely a classic slippery slope (allowing just a few devices to have just a little bit of control built in is assumed to lead automatically to all devices having total control built in), and as such doesn't really stand up on its own.