Re: [Terminology] WG Review: Effective Terminology in IETF Documents (term)

[[e-mail]]

Maybe because of my different background (which did include learning 
networking by reading RFCs, though), I would think IETF usual practices 
could eliminate a lot of uncertainty and maybe emotional involvement.  
Below are some paragraphs in which I try to show what I mean.


from the draft charter, v.07:
"The TERM working group is therefore chartered to produce an 
Informational RFC
containing guidance to IETF participants on the use of effective 
terminology
that also minimizes exclusionary effects. ...
... guidance to IETF participants on the use of effective 
terminology....
The RFC will express general principles for assessing when language is 
effective. ...
The output of this WG will provide guidance to IETF participants and 
will not
restrict the type or content of contributions that can be made to the 
IETF
standards process."


Looking at the situation described in ITEF's draft charter from the 
point of view of Information Theory, and simplifying WLOG to 1-way 
communication between IETF as Sender and users of RFCs as Receivers, the 
issue at hand seems to be at its base a mismatch between Sender 
encodings and Receiver encodings, whereas a foundational idea in Info 
Theory is that Sender and Receiver must share the same encodings.  RFCs 
are encoded using words in bare plain text, but some users are reporting 
that some words in RFCs messages introduce imprecisions which limit 
successful communication.  There are meaning tokens which have become so 
loaded with extra, non-computing meaning that for some users they are 
equivocal, and what is encoded in a word may not be the same for all 
participants.  From this channel efficiency viewpoint, the requests for 
changes seems like just another instance of usual IETF work to improve 
clarity of language in a multi-lingual world.

Focusing on just such cases of submitted complaints, it seems 
straightforward to get good data by assembling fully documented 
suggested improvements--ie some standard format list of them--from all 
interested parties, then review and decide what encoding tokens are to 
be rectified.  Which is what IETF is starting on, as far as I can see. 
The banality and democracy of the list format should even help maintain 
objectivity.  And the content of any archive about this issue would 
offer the ease of access of the list form.

The assignment emphasized by IETF, of improving effectiveness of 
language and terminology, is a different matter.  Previous posters have 
noted that may not be feasible, and I with a strong background in the 
fields which mashed together give sociolinguistics, would agree.  The 
problem is that "principles for assessing when language is effective" 
are elusive.

"Effective" is tricky to define in operational terms, and may end up 
seeming to be vacuous or full of special cases.  It is not like eg the 
definition of the effectiveness of an attempt to make a sports "goal".  
That can be based on 100% sampling of attempts; the achieving of a goal, 
and the responsibility for achieving it, is obvious to viewers; and 
there are no carry-over effects like there are in language use, ie an 
attempt to achieve one goal instance now is not affected by earlier 
attempt instances.

However, I can suggest the use of a tool which has a long history of 
acceptance as contributing to effective language use, the famous (in 
literary circles) Strunk & White, _The Elements of Style_, a small, 
efficient book.  My copy at hand is 2000, 4th Edition.  So the 
expression in a RFC of "general principles for assessing when language 
is effective" would consist in saying that writings should cohere with 
the rules in Strunk & White.  If they cohere, they can be treated as 
ceteris paribus assumed to be effective.

And internally, the effectiveness of what will be the TERM RFC, might be 
measured crudely with usual frequency and density measures.  Change over 
time of number of complaints of this sort, proportions over time of 
these complaints compared to all complaints for all RFCs, etc.  But 
these are far from direct measures of language in general.