I don't know if that's a counterpoint per se. That's the heart of why ML is actually challenging and knowing how fragile a notion like convergence is is importa

I don't know if that's a counterpoint per se. That's the heart of why ML is actually challenging and knowing how fragile a notion like convergence is is important and requires at least part of that mathematical jumbo jumbo.