[e2e] the evolution of deployability

David P. Reed dpreed at reed.com
Wed Dec 18 06:48:23 PST 2002


I think you've got it exactly right.  Standards should compete for customers.

Interoperability is something customers want, but they also want 
innovation.   There's a balance here.

Yes, I bemoan the days when telephony was about black plastic phones and 
Strowger switches.  That was a great standard.   But AT&T's rate of 
innovation deployment in that environment was truly appalling.   They paid 
substantial dividends, for goodness' sake!   That's what happens when a 
standard leads to zero focus on risk/reward investment.

The only way we are going to get innovation is by encouraging competitive 
innovation.

My gripe with IPv6 is that it never was put in a competitive context as a 
standard that had to compete for adoption.   IPv4 had to compete, and it 
did so beautifully.  (especially since IETF didn't control HTML and HTTP 
standards!)   IPv6 was protected by IETF, and dragged down by a bunch of 
risk-averse decision makers  (some of us have been saying for more than 6 
years to IETF leaders that IPv6 is being killed by competitive 
solutions).   NAT (a really crappy step backwards) outcompeted 
IPv6.   Which would never have happened if people spent less time assuming 
its success, and more time thinking about the customers who needed it NOW 
(then).

My opinion may not matter to most, but it's probably time to declare IPv6 
dead.   In the commercial world we call it "Missing the Window".  Fire all 
the programmers and designers, and let some new competitors go after the 
market.

My gut feeling is that what's happened to the Internet is that it's become 
a maze of twisty passages and toll roads.   It's vulnerable, because it can 
no longer adapt to the big challenges before it, just as the Bell network 
could never do what the Internet opportunity demanded.

It's time to short the Internet, and take some of that money and invest in 
some big-deal, high-risk options.    My personal bet is on RF 
Interoperability and edge-based distributed capacity.




More information about the end2end-interest mailing list