Teachy feely

Staniford has a new Hubbert Linearization post up at TOD. I still don't feel that approach merits any worth, but many people feel differently; from a commenter at the site:

It doesn't seem to be out of the realm for this to be taught in college/grad school to a fairly broad audience. Why has this been held back so long?

I can answer that. Because this particular approach resorts to a grab-bag of heuristics and empirical relationships and does not rise far above mere fortuity. Staniford says hisself that

" (I confess that I still don't fully understand why this model works as well as it seems to in practice)"

Placing myself in such a situation, I wouldn't feel comfortable teaching anything I don't deeply understand to a bunch of hungry graduate students. On the other hand, I would feel perfectly fine teaching the oil shock model, because of its reliance on some physical reality and the properties of stochastic processes.