I'll tell you what's right about your notion, the half that baked. With a wealth of data you don't need a model so long as the data directly contains the information you want. But take the Ars counterexample -- the "cloud" really doesn't know how Chris Anderson will react to experimental drug X. You argue that correlation might do. But it is not enough -- without a theory the correlation is meaningless. That's what the maxim means. (Or, failing that, it means you're paranoiac.)

If you think data can be generally exhaustive, I'm sure Phaedrus will be tickled to hear about how wrong he was. Oh, by the way, you really should read it, it's a popular book. If you have read it, you must have skipped the part where it's pointed out that more data begets more hypotheses. Which, by the way, is also implied rather loudly by the unadultured portion of the Box quote you're abusing.

Sincerely,
etc.