[e2e] the evolution of deployability

David P. Reed dpreed at reed.com
Wed Dec 4 14:59:14 PST 2002

At 11:09 AM 12/4/2002 -0500, Adrian Lahanas wrote:
>In US it was impossible to change inches and yards, once they were deployed,
>into centimeters and meters.
>In US it was impossible to change ounces or pounds into grams and kilograms.

         Yup, metric has failed.

>In US it was impossible to change from 120 Volt to 220 Volt.

         We didn't try this one.   For some good reasons, I may add.

>In US it was impossible to change from Fahrenheight to Celsius

         Agreed metric is a loser.

>In US it was impossible to change big cars into small cars.

         We did for a while.   It took the government creating a loophole 
for "trucks" to create the SUV.

>In US it was impossible to change steam engines into fast electric trains.

         No steam engines left.

But I think the Internet changes are not so subtle a problem.   The process 
by which the changes were developed and proposed was insufficiently 
aggressive.   Who's going to roll over the entire architecture with big 
risk, for a design that was deliberately held back so that it provides a 
very minor improvement of unclear value.   The current system works, more 
or less.   The new thing is not (in any practical sense) 2x or 10x  better 
on economically important dimensions.

There are much higher risk, higher return, really disruptive ideas (like 
the ideas I'm pursuing in radio, for example) that pay off in new economic 
opportunities.   They don't just make "fit and finish" improvements for 
existing applications while doing nothing to encourage out-of-the-box 

The disruptive changes will happen (but outside the current 
framework).   The current Internet will continue to incrementally absorb 
what fits inside its cultural and business context, because there's no 
great drive to do minor cleanups (like creating more address space, when 
the ISPs want to charge big bucks for each address anyway).

More information about the end2end-interest mailing list