21 June 2009

At Any t > 0...

Jay Lake points to some interesting, if ultimately incomplete (and therefore unlikely — but not impossible! — to be correct) comments on the dubious future of the publishing "industry". For one thing, Mr Stevens' article fails to acknowledge that there is no publishing industry; there are, instead, thirteen (or perhaps eleven) distinct publishing niches with various degrees of cross-niche consolidation, which in turn indicates pretty clearly that there is going to be more than one future to the "publishing industry".

[reduced size] Ransom Stevens, 15 Jun 2009    Second, it continues to accept the long-tail meme without doing the discrete math. On the right, you'll find an example of the long-tail graphic as used in that article. It's vastly more realistic than most (but for the cringe-producing mislabelling of one data element that, ironically enough, torpedoes the article's later imprecation to leave proofreading to the authors), and looks pretty convincing, doesn't it? The problem is that it's a freehand drawing based on the typical curves found in a non-calculus-based statistics course. Things get much, much more interesting once one acknowledges a few real-world difficulties:

  • Nobody has verifiable data to determine any aspect of the plotted curve, which in my own experience (based largely on the closest thing to "verifiable data" that exists: responses to discovery requests) is neither as smooth as nor as flat as the one in the diagram;
  • There's a huge difference between retrospective and prospective use of distribution functions, often simplified in Heisenberg's Principle but perhaps best exemplified by attempting to solve simple things like the Bessel function for a single electron... and then figuring out crystal structure from that data;
  • But, most grievously, there's a variable that is completely unaccounted for: t.

Ultimately, all of these "long-tail" theories fail because they are retrospective, cumulative snapshots that account only for aggregate income/sales — they do not account at all for the rate of the relevant transactions. I refuse to further mislead by trying to create a two-dimensional, web-friendly graphical representation of a four- (or more) dimensional reality that unfolds over time.

That said, there is one vastly superior model: Cellular energy acquisition, storage, and use. In this model, publishers aren't gatekeepers: They're enzymes (or, outside of the cell, catalysts) that reduce the activation energy required to actually start a reaction — say, the extraction and storage of energy through

ATP ←→ ADP + P

without disrupting other cellular processes... and is achievable without substantial further energy input. Of course, this all assumes some knowledge of basic thermodynamics — a part of physics and chemistry that thus far seems to apply to every natural system, even if the social "sciences" (let alone academic management!) have yet to figure that out.

So, in the end, this is another misleading snapshot. I guess what I'm really calling for is Publishing: The Motion Picture before I'll accept any of this stuff... because even under the 1934 Securities Act — what is really driving all of this nonsense — there's a nonzero temporal reporting horizon for publicly held companies (quarterly reports). It's rather ironic that the last graphic in Mr Stephens' {typo intended} ultimately misguided analysis does acknowledge time... and that he's a physicist!