All models are wrong, but some are useful.
Data without a model is just noise. But faced with massive data, this approach to science — hypothesize, model, test — is becoming obsolete. […] Correlation supersedes causation, and science can advance even without coherent models, unified theories, or really any mechanistic explanation at all.
I humbly disagree. Understanding needs models, predictions need models. Of course, in order to find models, correlations – probably found by using computers – can show the way.