Optimal intellectual property protection (part 1)

You often see stuff from the MPAA and RIAA about how piracy is destroying all artists' livelihoods always forever.  The natural solution to this problem is to somehow end piracy through legislative penalties.  So what does that look like?  Let's look at some horribly skewed and biased numbers!

Apparently, there's a man named Stephen Siwek who is happy to produce research showing the devastating effects of copyright infringement.  He isn't very creative about names though; one might even think he plagiarized himself.  But don't take my word for it, look at the names of his two "studies":

For the RIAA:   The True Cost of Sound Recording Piracy to the U.S. Economy
For the MPAA:  The True Cost of Copyright Industry Piracy to the U.S. Economy

It's a little tricky, because the Sound Recording Industry is a subset of the Copyright industry, and, actually, so is the Software Industry, which I won't discuss here.  Both these studies have some important characteristics though.

• They assume substantial full price substitution rates.   In fairness, I read through the "Sound Recording" report, and found that its estimated substitution rate is 60.7% for physical piracy and 20% for digital piracy.  The 60.7% is absurdly high based on my brief survey of the literature, but the digital piracy number is actually on the low end.  The most recent research (admittedly using a sample of U. Penn undergrads... not exactly a representative demographic in any way) puts it (and physical piracy, incidentally) at between 15% and 30%.  Of course, some researchers show no effect on physical sales and slight positive effects for online and concert sales (that's on an international sample, but its methods are a little harder to follow and its credentials aren't as towering).  The "Copyright Industry" report only says "less than one", a value it claims is "conservative" on the grounds that some internal industry "estimates" claim that it is exactly one.  Furthermore, each lost purchase is imagined to have replaced one at full price and new.  
• They calculate losses in the global market. So piracy in China is lumped right in and used to justify stronger copyright enforcement in the U.S.  I mean, maybe fair enough for treaties, but in general? 
• They show "total economic output lost".  This goes beyond lost sales to include the hypothetical economic impacts up and down the supply chain of the copyright industry. 
• They neglect the obvious economic benefits to consumers of receiving goods for free. This seems particularly relevant given their lost sales model of substantial purchase substitution, and their insistence on evaluating macroeconomic effects.  Every dollar not spent on frivolous entertainment is available to be spent elsewhere, in some other industry (or in the same one, for that matter).  Effectively, they are assuming that people choose between purchasing music and "pirating while putting the money they save under a rock".
• They rely substantially on confidential and industry provided data.  Well of course they do; how else could you produce credible looking studies without any means of verification?  Well, I guess they aren't that credible looking.
• They include both physical and digital piracy.  Bootlegs, bit-torrent, mix-tape? All the same.

Anyway, the number Siwek comes up with for the cost of worldwide, savings-burying, any format, substantial pirated-copy-to-lost-new-sale substitution, all-copyright-industry piracy is:  

58 billion dollars.

So, that's a lot of money, I guess.  By contrast, if a naive but mathematically capable person tried to estimate the cost of pre-screening all copyrighted content on youtube alone through manual human curation they'd come up with something between $441,029,692 and $36,829,468,840 per year. Ahem:

37 billion dollars.

And that's just youtube! Imagine all the streaming sites and content lockers...  Plus, the comparison is unfair because youtube doesn't host software or videogames--both major contributors to that headline number.  If you chop out Software and Videogames from the 58 billion total estimate you are left with 19.256 billion dollars of loss per year (approximately, since they don't actually break this out by industry I have to do some funny stuff--basically I multiply the total loss by the percentage of direct loss in the music and media industries: 33.2%)

Now, those numbers are quite silly.  Also, they naively ignore the existence of Google's Content ID system, which basically does this automatically in software.  One might imagine it is substantially cheaper than any of the estimates of human manual labor curation.  The point (for tonight) is merely that anyone with a bit of time and some trumped up numbers can make pseudo-reasonable apocalyptic claims about the cost of enforcing or not enforcing copyright.  Next post, a closer look at the genuine economics and law of the situation.