I've seen some very interesting and in depth discussion generated by this article on the web.
Overall I agree with Ellis and Silk - empiricism is vital to the scientific method and progress, and these notions of "post-empiricism" seem, to me at least, to be a dangerous path to go down.
I think what it comes down to is the degree to which we are willing to investigate hypotheses which have not been confirmed. I have not read the criticisms mentioned in the article, such as Smolin's or Woit's books, so I do not know exactly what they are advocating. If they are saying we should completely halt research on string theory and focus our efforts elsewhere, then Dawid's post can be seen as appropriately reactionary. I don't think string theory is without use, even though it has not (and likely will not be) experimentally confirmed, and research should continue. However, if the message in this criticisms is much more modest, then I don't see the use in Dawid's arguments. What do we gain by treating String theory as confirmed that we do not by treating it as unconfirmed? I guess I just don't see the point in Dawid's arguments, unless they are to argue for continued research in String Theory, which is already happening.