How Many Legs Does Science Have?

Back in 2010, Moshe Vardi wrote an opinion piece titled "Science Has Only Two Legs", in which he argued that computational science is just another form of experimental science, and that programs should be held to the same standards as other pieces of experimental apparatus. It's an interesting counterpoint to recent excitement around "big data" (as evidenced in works like The Fourth Paradigm), and, I think, directly relevant to what we're trying to teach. What are your views?