A couple comments:
1. At this point it's very unclear what people mean when they say string theory is "elegant". There are aspects of the theory that are mathematically very compelling, but these don't look at all like the real world (e.g. they have a lot of supersymmetry). If you try and relate string theory to the real world, the most popular way to do that now involves what Susskind calls the "Landscape". He explicitly argues that this picture of reality is highly inelegant and Rube-Goldbergesque.
2. The historical analogies are very strained. String theory is nearly 35 years old, has had literally thousands of very talented people working on it for 20 years (since the fall of 1984), and has produced not a single prediction of anything.
In the case of GR, there was one person working for 12 years, at which time he had real predictions and postdictions that could be checked.
In the case of QED, the theory was written down within a few years of QM, and it made a huge number of verifiable predictions. It did take 15-20 years to sort out renormalization, but those were years when many physicists were occupied with other things (staying alive, developing ways of killing large numbers of people, etc.)
The standard model took another 25 years, but even before it reached its final form, parts of it were making all sorts of detailed predictions that could be compared to experiment.
http://davidappell.com/archives/00000167.htm