So BoingBoing had a pointer to this article by some dude about Turing machines and Turing computability. This guy allegedly has a PhD in CS from Duke. Wtf.
I posted a critical comment (comment #2) which I'm going to expound upon briefly here.
Turing machines are a model of computation. What makes them
universal is that, for any set of separate Turing machines that do different things, including the infinite set of all possible Turing machines, one can find/construct a single Turing machine that does all of those tasks. In other words,
adding one Turing machine to another does not remove you from the world of Turing machines—you simply end up with a third Turing machine.
Now what makes Turing machines a powerful idea is the Church-Turing hypothesis, which essentially states that anything that we as humans would call
computation can be performed by a Turing machine. The C-T hypothesis, like many scientific theories, can never be logically proved, but it has not been disproved, and people believe it.
Computatability theorists have studied more powerful models of computation than Turing machines, in particular ones where you assume the presence of a magical
oracle which answers correctly any question you pose it, but these models do not correspond to anything in the real world. Saying that because people have studied these models, Turing machines are somehow irrelevant or insufficient or outdated, is simply incorrect.
Anyways, I just find it a bit bizarre that this guy attacks Turing machines. It's a little difficult to really understand his arguments because he confuses a lot of terminology and ideas.