Keith Moore writes:
it's even more absurd that widely used protocols should be
subject to arbitrary modification by any random individual
who thinks he understands what is going on better than
anybody else.
We aren't talking about ``arbitrary modification.'' We are talking about
_extensions_ to the Internet protocol suite, through mechanisms designed
to support these extensions: TCP ports, header fields, etc.
Since nothing limits the semantics of header fields, adding a header
field can cause an arbitrary modification to the protocol.
TCP and UDP ports are just a demultiplexing mechanism. They don't
change the semantics of IP, and different port assignments don't
interfere with one another.
You might notice that it's considerably more difficult to get approval
for new IP and TCP options than it is to get a port assignment.
Some ``random individual'' at CERN developed and deployed the World Wide
Web without approval from anyone other than his boss.
Sure, but he didn't change an existing protocol to do that. As it
happened, HTTP's poor design created tremendous headaches for network
operators. But at least the introduction of HTTP didn't disrupt other
network services (much).
Imagine going back to 1990 and demanding that he write protocol specs,
deal with people like Keith Moore, and build ``consensus,'' before
being allowed to use a TCP port!
And imagine if random vendors had been allowed to make arbitrary modifications
to HTTP without getting consent from anyone. Of course that's exactly what
happened, and the current version of HTTP is the mess that resulted. (more
precisely, HTTP 1.1 was a long and very expensive and only somewhat successful
attempt to clean up the mess that resulted)
Now you're arguing that IETF should endorse the practice of making
random modifications to protocols.
The past decade of Internet progress happened _despite_ IETF control,
not because of it.
I could make an equally valid statement about Dan Bernstein.
Keith