hit metering as 'Proposed Standard'

There's been some private mail on the topic of Hit Metering as
Proposed Standard, but I thought I should respond publicly:

It is not useful or relevant for the "major vendors" to give
testimonials as to the number or intensity of desire on the part of
their customers to have Hit Metering.  I think everyone believes that
customers want help in gathering demographic information without
interfering with caching. We all believe that there are many many
customers who want the problem solved. We don't need testimonials.

I don't actually think these customers would be well served if we push
through a Proposed Standard that doesn't actually solve the problem
they want solved.

The IETF process (RFC 1602) says that a Proposed Standard has resolved
all the known design choices. There's an obvious design choice for how
we go about letting people gather data: explicit hit metering, or
statistical sampling. I'm not seeing 'rough consensus' in the group
that we've resolved this design choice.

# It is because of the above that I don't think Experimental is the right
# path. Experimental is for when there are doubts about the soundness, and
# expreimental use is required to test it. And Experimental explictly
# forbids deployment in operational use -- but it is exactly operational
# use that we need to permit in order to decide the question of utility.
 
Yes, exactly. There are doubts about the soundness. I think the doubts
have been expressed clearly, and I don't think that sufficient
evidence has been presented to remove those doubts.

This is on the agenda, and we can at least see from the folks in the
room whether we have consensus on whether Hit Metering should go
forward as Proposed Standard.

If there are others who have not sent mail on the topic who have a
strong opinion about moving Hit Metering forward, you can mail either
me privately or the group as a whole.

Larry

Received on Monday, 2 December 1996 19:24:40 UTC