Is This Theory Really Untestable?

I was reading an article that made the claim that this theory which gets lots of press has to "be taken on faith" and "no experiment can tell it it's right or wrong." No, no, not Intelligent Design--string theory.

The article is in the January 2006 Astronomy, and it is Bob Berman's column. I don't know enough about string theory to know if he knows what he is talking about with these criticisms--but if he is, it would suggest that non-testability isn't a problem just for Intelligent Design:

Problem one: String theory won't work in our reality of three dimensions plus the fourth dimension of time. To make it work, its creators had to invent six or seven additional dimensions, which contradicts our own senses and the rest of science. None of these extra dimensions can be possibly be tested. They have to be taken on faith.

Problem two: String theory is untestable -- no experiment can tell if it's right or wrong.

Problem three: According to [Columbia University professor Peter] Woit, string theory's only prediction (about the strength of the cosmological constant) proved to be incorrect by 55 orders of magnitude. Oops. That's like predicting mice are larger than stars.

I suspect that there are some cosmologists and physicists reading this blog who can answer the question: is Bob Berman right? Is string theory untestable--does it have to be "taken on faith"?