In a discussion recently on the formerly useful DVInfo Forums I was trying to explain to some self appointed experts that they had the wrong impression of how Log Gamma encoding works.
Some misleading marketing graphs gave them the idea Log crushes the highlights, giving you even less information than than regular gamma curves, rather than evenly distributing the information giving roughly the same amount of data bits per stop. Effectively linearizing the stop per data bit bucket.
I gave up on the discussion quickly as it was evidently pointless, but I wish this Art Adams article had been available then.