In the latest issue of Current Biology there is a short two-page interview (sub-only) with Ronald Plasterk, current director of the Hubrecht Laboratory in Utrecht.
He had some very funny things to say about systems biology :
"The fundamental misconception of systems biology advocates is that one could create a virtual cell, and use big computers to model life and make discoveries. None of these modellers ever predicted that small microRNAs would play a role. One makes discoveries by watching, working, checking. They want to be Darwin, but do not want to waste years on the Beagle. They want sex but no love, icing but no cake. Scientific pornography."

I had a great laugh with this one :), however I happen to be working in a lab that is making software to exactly this and I disagree with this analogy. Of course you cannot discover something with your model about biological mechanisms that we know nothing about, but for sure that modeling approaches can help guide experimental work. If you model fails to explain an observation you can use the model to guide your next experiment. You go on by perfecting your model based on the results and so on. These cycles are not much different from what biologist have been doing intuitively but I think that few people would disagree that formalizing this process with the help of computational tools is a good idea.