This has become probably the most common argument made by string theorists when string theory is criticized for being non-predictive. When I first heard it, my reaction was that this was a joke. How could anyone seriously try to claim that the predictivity situation of the SM quantum field theory (our most successful fundamental theory of physics, which makes a wealth of detailed, accurate, tested predictions) and string theory unification (a theory that predicts nothing) was the same? There’s obviously some sort of sophistry going on here, an attempt to claim that white and black are much the same, since they’re both shades of gray.
QFT is successful because it includes a specific class of theories that have a lot of symmetry, and a very tight mathematical structure. Making a few of the simplest possible choices for theories in this class, you get the SM, with a huge amount of very non-trivial, highly testable predictions, all of which have turned out to work perfectly. Yes, it’s true that you could instead look at extremely complicated examples of QFTs, making them so complicated that you would lose predictivity and start to get something more like string theory. This is true of just about any theoretical set-up, you can make it worthless by adding in complexity until it gets to the point that it will fit anything, while explaining nothing.
That’s the problem with string theory unification schemes: you have to put more into them than you get out, the hall-mark of a failed idea. Here’s an old comment that goes the problem in more detail:
http://www.math.columbia.edu/~woit/wordpress/?p=4065&cpage=1#comment-98564
