Snitches & Activity

[John S. Quarterman]

Submitted-by: jsq at usenix.org (John S. Quarterman)

I'd like to thank Peter Salus for his input, and to encourage others
to comment on USENIX standards activities.

However, some of his comments were motivated by impressions produced by
unclear background.  This is my fault, since it's my job to explain
what USENIX is doing with standards.  I was planning to wait a week or
so after posting the poll results before posting anything from me about
them, but a bit of clarification now seems best.

>(1) I didn't think the manner of polling was in any way 
>	appropriate.  On the one hand, many individuals
>	concerned with IT standards don't get the news 
>	groups, on the other hand, many members of the 
>	various associations (USENIX, UniForum, IEEE, 
>	EUUG, etc.) don't care about standards reports.

Normally, we get almost no feedback about the snitch reports.  This
makes planning difficult.  So, we're trying several ways to get input.
A poll on the networks could be done quickly (defined as before the
next USENIX board meeting) and with a minimum of preparation, on the
theory that some results were better than none.

In addition to that poll, the paper mailing currently going out to TCOS
committee members has a page from me asking for responses.  It's not an
actual survey, because some TCOS Standards Subcommittee officers thought
that would be inappopriate (and I tend to agree), but it should reach
people that the network poll did not.  A request for responses was
printed in ;login: some months ago.  Unfortunately, it only got about
twenty answers.  But a direct mail survey of the USENIX membership is
in preparation, and will include questions about standards activities.

Suggestions for other approaches are welcome.

>(2) I thought the poll questions and methodology were 
>	slanted and unscientific.  The fact (noted in
>	some comments) that there was no option for 
>	no response/don't care/don't know is part of 
>	this.  The regression to the mean in most 
>	responses is evidence of just how badly 
>	designed the poll was.

This is one of the results of the above-mentioned minimum of preparation.
I spent maybe four hours writing the poll questions and the shell script,
with no prior experience.  In a later posting, I will explore the specific
problem Peter mentions, as well as some others.  I will probably claim
that, nonetheless, there were some useful results.

>(3) The paltry number of responses shows that even the 
>	majority of POSIX and X3 attendees didn't care 
>	about the questions.

I'll address that point in a later posting, noting here only that 34 people
answered yes to at least one of questions 6[XYZ], and that about 400 people
receive the average TCOS paper mailing.

>(4) I like the snitch reports.  I don't think that either 
>	John Quarterman or Jeff Haemer does an adequate job 
>	where the postings and the articles in ;login: are 
>	concerned.  I would like to see the snitch reports 
>	reduced in size to no more than 4-6 pp. in a quarterly
>	issue where POSIX is concerned; a page where 1201
>	is concerned; a page on WG15; etc.  I think that if
>	jsq & jsh are to do a job, it should be to filter 
>	the trash before it hits the printer.  Editing is 
>	more than merely spelling and punctuation and cute 
>	asides.

This is the big background problem.  Most people are unaware of how
much editing Jeff actually does.  I see both the original, unedited,
versions and the final edited form of every report.  I can attest that
Jeff does quite a bit of work on many of the reports.  Some are passed
through virtually unchanged, but others are very different after editing.
The result is that the published reports have a relatively uniform format,
although there is no attempt to disguise the individual styles of the
snitches.

Every article, whether changed from its raw form or not, goes back to
the original snitch for review before being sent on to me.  Some of
them go around several times, developing as they go.  Jeff's editorials
go by every snitch before they reach me.  This is a policy I insist on
in hopes of avoiding technical and political problems with the reports.
Occasionally, even after all that, I will ask Jeff to fix some problem
that I've spotted because I've been dealing with this for longer than
he has, but that's gotten to be pretty rare.

Several people have mentioned to me lately that they thought that all
the editing Jeff did was to add in the editorial remarks [in the brackets].
Nope.  Those comments are added after the real editing, and allow Jeff
to address the reader directly, often so he can encourage readers to
get involved (which is definitely one of our goals).  Very occasionally,
I will also add bracketed comments, usually about of political aspects
that Jeff didn't know about (usually because I was doing the politics
while he was editing).

There is essentially no editing done on the reports specifically for
;login:.  This is largely because there is no budget for it, and is one
of the reasons for considering a standards newsletter.

Jeff also finds the snitches, persuades them to write the reports,
and sits in on numerous committee meetings looking for topics suitable
for someone to report on.

In other words, Jeff is doing the job I asked him to do.

On the other hand, wanting shorter reports is certainly a valid
viewpoint.  We are actively soliciting such viewpoints.

>(5) I'd like to see more on the unmentioned ISO bodies:
>	what's gone on at the recent meetings of JTAP,
>	for example? [Joint Technical Committee on 
>	Application Portability].  They met in Copenhagen
>	in February and should be meeting about now in Ottawa.

Good question.  We've spent the last year or so cultivating
snitches in the more obvious groups, and haven't gotten to
JTAP yet.  If you know somebody who wants to volunteer, please
let us know.

>I think I need yet another newsletter about as much as I 
>need another of Dave Yost's wooden nickles.

I lost my last wooden nickle.  Where is Dave when you need him?  :-)

Once again, thanks for your input, Peter, and may it encourage
others to comment.

John S. Quarterman, USENIX Standards Liaison, jsq at usenix.org

Volume-Number: Volume 21, Number 107