Re: Device Drivers as Entry Barriers

["Etchison, Michael" <etchison@puc.texas.gov>]

	Brett Glass, as forwarded 3/9 by James Love, writes about the very real 
difficulty would-be vendors of operating systems other than Windows (for 
Wintel machines, at least) have in persuading hardware vendors to write 
drivers for their OSes.  He suggests that some of the hardware vendors' 
reluctance may be fear of Redmond, and I'm in no hurry to disbelieve 
that.  But (as he also acknowledges) hardware vendors are reluctant to 
spend much time writing drivers for OSes which are not widely used.  As 
rapidly as change occurs, and matters, in this industry, I also would 
want to focus first -- and, therefore, probably _only_, on the platforms 
actually being used the most.  Call it increasing returns to scale, or 
network, or whatever.  It is quite sensible strategy.
	It is so sensible, it seems to me, that it will happen no matter how the 
starting lineup is jiggered.  Maybe this is an extension of the Coase 
theorem:  it doesn't matter how you start out, the market will end up 
with a single dominant OS because it is to the (entirely legitimate) 
interest of a great many participants that this happen.  If that's so, 
then using brute (or legal) force to make, say, Dell provide alternative 
OSes, or to make hardware vendors open their specs (talk to me of trade 
secret litigation!), is frivolous.  Wherever we start, we'll end up with 
one OS.  There may be other responses to the result, such as separating 
OS from application (although, as the IE thread demonstrates, that 
distinction is getting harder to know when you see it), but we ought not 
at this late date draft Canute to hold back the single-OS waves.
Michael E. Etchison
etchison@puc.texas.gov
[opinions mine, not the PUCT's]