Finitistic objections to the current mathematical model

I recently read this pdf: Warning Signs of a Possible Collapse of Contemporary Mathematics, and I'm having some trouble understanding the issues it raises.

The author says that the consistency of Peano Arithmetic cannot be proved within the system, which is a direct application of GĂ¶del's second incompleteness theorem. It can, however, be proved using transfinite induction, but this is a problem in his opinion, because one has to trust the existence of infinite cardinals, contrary to his monotheistic faith. My first question is whether this is what he meant, or if I misunderstood some part of his argument.

Nelson then goes on to define some classes of numbers which are contained in one of them, that of "counting numbers", and proves that these counting numbers are closed under addition and multiplication. But he implies that it is not possible to prove that exponentiation and superexponentiation and, in general, faster-growing recursive functions, whose definition is completely permitted by the axiom of induction, are guaranteed to terminate, i.e., to be computable. My other questions are: can all recursive functions be shown to terminate after a finite number of steps, assuming the existence of some transfinite cardinals? And what exactly is the finitist objection to this assumption? Finally, what would be a finistically acceptable model for mathematics? Is it usable, in a general sense?

Thanks.

-
Note: the answer to the first question you ask is that Ed Nelson was somewhat joking in the first two pages. That paper was based on a talk he gave, and like most good public speakers, he started with a hook. The real mathematics starts roughly on the bottom of page 2. – Willie Wong Feb 22 '11 at 10:15
@Willie: Oh, I see. It would have made an interesting view if it hadn't been a joke. He does keep mentioning matters of religious faith later in the article, though. – Abel Feb 22 '11 at 10:24
Related (it may even contain the answers to some of your questions): math.stackexchange.com/questions/531/… – G. Rodrigues Feb 22 '11 at 14:54