(First off, I disagree with $1.92 million in damages. It's absolutely ridiculous, IMNSHO. I want to explain my point a bit further, though)
Wait, that doesn't make sense: the profit should be used up covering the fines for the compensatory damages (by definition!)
well, yes, and that works if and only if copyright infringement is always found out and prosecuted successfully.
Imagine, for a moment, punitive damages were small or nonexistent. I start a DVD pirating company, selling copied DVDs for $5 apiece.
I sell 100,000 each of two particular titles. my $5 is basically pure profit, because of the almost nonexistent cost of copying. If I don't get caught, that's, say, $450,000 per title profit for me. If I do get caught, I have to pay compensatory damages of $1,000,000. ($10/DVD).
The profits from one or two titles (in this example) completely offset the fines from getting caught for one.
So, as long as I only get caught for a third of my titles, I make a rather nice profit. While this example uses made-up numbers, the basic idea stands: Due to the low cost, and high profit potential, of copying copyrighted works, straight-up compensatory damages don't act as an effective deterrent
As a side note about 'only getting caught for 1/3rd". In the Thomas case, do you think she really uploaded only 24 files, ever? She quite likely had uploaded more than that, it's only the 24 songs she's caught for
Say Thomas had uploaded 100 songs, to 100 people, and somehow made $0.10 on each upload. (she didn't) The RIAA can prove 25 songs to 25 people (They can't). Straight compensation would be 25*25*$1, or $125. Her profits would be 100*100*$0.10, or $1000. A contrived example, yes, but it illustrates the idea. The expected value of for-profit piracy is a positive amount, without damages beyond straight compensatory.
the fundamental disconnect here is that Thomas's infringement WASN'T FOR PROFIT. In this situation, the law doesn't correctly account for reality