He can’t pronounce fresnel correctly (few can in this business it seems), do we want to trust him? 
Whatever gives you the results you’re looking for, that’s the “correct” approach. There is nothing wrong with using fresnel input (heck, even with a curve modifier) or face input (has to be curve driven, usually). Often I will exaggerate the “IOR” to raise reflections, or divide it a bit to lower it.
Do your own research and observe how well (or badly, quite often) the result vary from so-called “official” values. Buy a LEGO set and see how ABS plastics in this form isn’t quite as reflective at glancing angles as tables tend to suggest. Another thing I’ve often noticed is to raise “IOR” value significantly for the “main rough gloss of the actual plastic”, and have the “sharp shiny gloss of plastic” be a separate “coating layer”. If it looks coated, it often is The “IOR” value of the “coat” would then depend on the thickness of the transparent coat. If thin, you’d get edge favored sharp reflections. If thick, sharp reflections would pay a more major part. In addition to the rougher gloss of the plastic material itself.
Also, you’re showing chrome. What is chrome? It’s a chemical element, hence a pure metal (assuming no contaminants, and thickly enough plated showing no signs of transmittance - the usual way of applying chromium plating at least), so where is your diffuse coming from? The thing that ruins the left image is the amount of diffuse coming through. Chrome isn’t as reflective as you’d think (although I often exaggerate as well for artistic purposes), it’s actually a bit absorbent (compared to aluminium, silver, gold), but it’s preferred for its durability/hardness, price, and protective abilities.
So, did you miss something? Not really, except chrome shouldn’t have a diffuse component (if pure). Even steel (alloy, not a pure metal) has such a low non metal content (carbon) that you could safely ignore diffuse. If anything, mix in another glossy shader with varying roughness to simulate imperfections caused by scratches or less perfect surface treatment (i.e. polished).
Can you use curves for everything? Sure, if it gives you the additional controls you need. It’s not “typically” needed though, and you might want to apply to a high recursive scene to test for speed differences (I haven’t tried yet myself). If it works, then it’s ok. Scientific values is an ok starting point, don’t get me wrong. But don’t trust them blindly if the results doesn’t seem to pan out, as the experiment might have other mistakes.
One exception to “not typically needed”, is that of glassy materials. Pretty much every time I need glass, I go with custom curves, or at least a custom node setup, because the builtin glass shader offer zero flexibility wrt control. Real life glass is never as perfect as using a glass shader on its own, or I just need to tweak glass properties that doesn’t exist in the shader version (like very transparent but using a layer of anti reflection coating, heck, even dust :)).
In the end, for me, the quickest to render solution that produces acceptable results, is what is good for me. That said, I othen tend to overdo complexity myself 
As an end note: Pure polished metals tend to be so reflective that you can safely ignore any kind of “IOR” or curved facing tricks. Not a soul would notice in the grand scheme of things.
