[Chapter-delegates] ISOC open letter

[John Levine]

It appears that Mike Godwin via Chapter-delegates <mnemonic at gmail.com> said:
>
>Steve, this is one thing that has struck me as odd as well. My view of
>"layers" has been that it's a kind of metaphorical way to taxonomize
>different functions and services--at best a kind of heuristic for picturing
>and grasping how the internet works. But some people have decided
>(apparently) that the taxonomy or heuristic is an articulation of first
>principles.

This also shows a painful ignorance of history.

Back in the day there were implementations of network protocols that
carefully implemented each layer separately with clearly defined
interfaces between each. They all shared the characteristic of being
stupendously unusably slow, and in many cases impossible to debug
because all of the interfaces offered an exponential explosion of ways
to screw up.

The pragmatists that built the Internet defined interfaces where they
were useful and didn't obsess about places where they weren't.  There
were a lot of experiments some of which succeeded (most of Berkely sockets)
and some of which didn't (the Ethernet trailing header hack.)

Unfortunately, most of the people with that experience are close to
Steve's age, and there aren't a lot of them left to remind us of why
our networks work the way they do.

R's,
John