Metrics and engineering tradeoffs

[Sam Ruby <rubys@xxxxxxxxxxxxxxxx>]

We now have a number of proposals on how to deal with composite (text + 
image) weblog entries, differing in a number of important ways: anywhere 
from what use cases they support to esthetical considerations.

We also have vendors who are concerned about the number of bytes that 
are transmitted, the number of connections that are required to achieve 
a given "transaction", and the number of layers of software are required 
to implement a given solution.  Tangible, measurable quantities.  One 
with demonstrable costs.

I'm not suggestion that we base all evaluations exclusively on such 
metrics, but that it would be helpful to have such information 
available.  One solution may require 20% more traffic, but another has 
compensating feature X... that sort of thing.

In order to try to swing the pendulum back towards what I perceive to be 
"center", I've begun work on a proposal based on MIME, focusing 
initially more on the bits and bytes and less on the use cases and the 
like.  Hopefully, this can give us a lower bound on certain metrics, and 
serve as a useful benchmark upon which we can evaluate the cost/benefit 
of additional features.

More detail is available at:

http://www.intertwingly.net/blog/1781.html

- Sam Ruby