He suggests we put aside the emotive debates about copyright and associated monsters like SOPA and ACTA and look at the numbers. The MPAA, for example, claim the internet is costing the US economy $58 billion a year "due to content theft". $58 billion which if you had it laid out in pennies would stretch all the way to Mars. It's also equivalent to the entire US corn crop failing, along with all fruit, wheat, cotton, rice and a number of other crops.
Music revenues are down about $8 billion a year since Napster's early days according to the music labels. Movie and cable TV along with publishing revenues are up so where's the missing $50 billion, once the $8 billion drop in music sales is accounted for? With copyright revenues up in most sectors the missing $50 billion must be "foregone growth in a market that has no historic norms", i.e. a market that didn't previously exist. His tongue in cheek answer to the missing link is "the insidious cost of ringtone piracy."
The MPAA also tell us the economy loses 373,000 jobs to content theft. Interesting given that in 1998 the US Bureau of Labour Statistics showed the entertainment companies were employing 270,000 people.
He goes on to talk about the $150,000 statutory damages available in the US for copyright infringement, (hence the notorious excessive damages in the Jammie Thomas and Joel Tenenbaum cases). He claims the original MP3 player the Rio could store 10 songs i.e. $1.5 million worth of songs. In an iPod with a 40,000 song capacity this runs to $6 billion "worth of stolen media." It's a pity he got his sums wrong here, as he actually says "$8 billion worth of stolen media or about 75,000 jobs." Maybe the error is deliberate and he's making a subtle point about the normal rules of mathematics being ignored in copyright debates and policy? His conclusion that you might find copyright math strange is an understatement. The five minute talk is nevertheless worth a look.
Update: Rob Reid explains his $8 billion calculation:
"In determining a given device’s maximum capacity for infringing material, I assumed an average song length of three minutes, and an encoding rate of 128 kilobits/second. I went with 128 kbps because using the AAC codec, this is the rate at which music achieves “hi-fi transparency — which is to say, it becomes indistinguishable from CD quality in most listening environments. This rounds very closely to 1 megabyte of data per minute of music. At 32 megabytes, the Rio (1999’s Christmas hit) therefore had room for about 10 songs, which, if pirated, could represent up to $1.5MM in liabilities under US law. Today’s iPod classic, with its 160GB capacity, can hold 53,333 songs, which at $150,000 a pop is precisely $8 billion. Incidentally, Apple markets the iPod classic as having room for just 40,000 songs, but by my math, that’s selling it short. I meant to note this in the presentation, but I was running way over time by then, and spared everyone the convoluted math (so if the leap from 40,000 songs to an $8 billion liability confused anyone, I apologize — I had meant to take a quick detour through that 53,333 figure!)."That's a really helpful clarification. The problem with simplifying is you end up losing precision to the point of getting it wrong. I hesitate to extend the pedantry but 53,333 songs at at $150,000 a pop is precisely $7,999,950,000 not $8 billion. Another 1/3 of a song at $50,000 is needed to reach the magic $8 billion. Of course the odd $50k in the copyright wars is barely discernible small change...