[ivxvm]

> has made all of its progress by throwing chairs (data) at the problem.

To me it seems like the most tricky thing about the whole field is that quantitative adjustments make qualitative breakthroughs. Simply increasing parameters / layers / data make sudden rapid jumps in generalization / abstraction / etc. It's somewhat unintuitive because many things in world don't work like this, and the metaphor you're discussing is more about "normal" things, it doesn't really apply to ANNs well, those are more of anomaly, like quantum mechanics (yeah, it's probably a bad analogy too).