Human factors (was Re: Wish List for New Spec)

Harold A. Driscoll (
Thu, 04 Jan 1996 15:55:17 +0000

Message-Id: <>
Date: Thu, 04 Jan 1996 15:55:17 +0000
To: "Daniel W. Connolly" <>
From: "Harold A. Driscoll" <>
Subject: Human factors (was Re: Wish List for New Spec)

At 04:11 4/1/96 -0500, you wrote:
>Ack! I've done it again. Whenever I write team/WWW, change it to pub/WWW.
>So it's:

Thanks, Dan, that did the trick. And I must say, your collection is
certainly worth the trouble. Kudos and thanks for putting it together.

I'd asked Santa for a keyboard with a "Do What I Meant" key--maybe next
year. No doubt, if anybody comes up with one, it'll create a buying rush
which will outsell even "Holiday Barbie dolls." :-)

IMHO, the human-engineering aspects of URIs, URNs, etc. seems to have taken
a back seat to the technical aspects. We are accustomed to seeing a
"Security" section in every RFC proposal; it seems we need a similar section
for "human factors" as well. While human factors are certainly addressed in
a piece-meal basis, it seems to me that they also need to be specified (and
evaluated) in a top level manner, just as security (quite rightly) is treated.

Just to take one example, the popular ISBN, has within it a check-digit. No
doubt many transactions have had keying errors detected much earlier in
their processing, saving considerable hassles, due to such defined-in

Our local (Chicago) daily newspapers now have URLs printed almost daily, and
it has been months since I've seen one of them print
" tilde here)". At the same time, I'm
certain that if I had a dollar for every redirect or symbolic link set up on
a Web server to accommodate a mis-published URL, my
super-fantasy-laptop-computer would be sitting by my desk right now.

I'd encourage you to add to your document a section to the effect of "Human
factors are discussed in...", even if that list is presently very short. It
may be inevitable that at times people will have to use constructs which
might be described as "something only a computer could love."  But avoiding
such conditions, as often as practical, needs to be a major design criteria.

Harold A. Driscoll             
#include <std/disclaimer>