COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	∂15-Dec-80  0153	DUFFEY at MIT-AI (Roger D. Duffey, II) 	HUMAN-NETS Digest   V2 #170   
C00012 ENDMK
C⊗;
∂15-Dec-80  0153	DUFFEY at MIT-AI (Roger D. Duffey, II) 	HUMAN-NETS Digest   V2 #170   

Date:  12/13/80 1741-est
From:  Pool at MIT-Multics
Subject:  Reactions to HUMAN-NETS

For a year or so now I have been a passive reader of HUMAN-NETS.
It has been a very useful but frustrating experience.  It is
owing to you colleagues from whom I have learned so much to
give you some reactions.

Aside from HUMAN-NETS, I read two other daily papers: The
New York Times and Stanford's SAIL.  The New York Times wins
hands down.  It isn't specialized on my personal professional
interests (which HUMAN-NETS is), and it misses some stories
that are in the more comprehensive SAIL, but it is edited to
help the reader find what he wants.  After years of reading
I know exactly where I will find what I want, and the editor
helps me with positioning and headlines.  I know that an
expert editor has flagged important stories that I must
know about for the front page, that the foreign news is
in the next few pages, followed by national news, and then
the Op-Ed and editorial page.  I then read the business news
looking for stories on the communications industry, skip the
sports, autos, classified ads, society; glance briefly at
the metropolitan section because I grew up in New York.  My
only frustration is that the obit page in the present Times
format moves around and the index doesn't tell you where
it is.  Being in my 60's I often find friends there.  Even
there the editor helps me.  I don't read the paid notices;
experience has shown the incidence of useful items is too
low.  But I do read the names that the editor singled out
for stories.  Headlines give them fast.

Headlines and the standard format for news stories of putting
the key facts in the first couple of sentences are other great
aids. They enable the reader to be active rather than passive.
You choose what you want to read.  Reading HUMAN-NETS is more
like watching the TV news, which I don't do.  You have to wade
through the things you aren't interested in.  Granted I don't
read all or even a quarter of HUMAN-NETS words, but it takes
a lot more digging to decide whether a story in HUMAN-NETS
contains some gold that I am interested in than it does with
the NYT.

I used to have all the same sorts of problems with SAIL,
but now I no longer access it directly.  I have a colleague,
Richard Solomon, who follows it closely and has interests
much like my own.  Whenever he notes a story that I would
have missed and which he thinks I would like to see he sends
it on to me.  So I have an expert editor; I certainly am
grateful.

Obviously electronic newspapers and message groups are not
going to mechanically imitate the solutions found for older
and different technologies, but if they can't serve the same
functions better in new ways then they will be relegated to
specialized supplementary functions.  So the question that
needs to be addressed is whether and how future services
like HUMAN-NETS can achieve what a newspaper achieves by
positioning of stories, headlines, editing and typography.

Several solutions are obvious though partial and in some
cases ambivalent in effect.  Other approaches may occur
to others on the net.  I'd like to hear.

1. The user can have a key-word based algorithm for screening
   stories (as on SAIL).  As recent net messages have noted
   they work only so-so now.

2. The key words or other markers that resulted in selection
   can be printed bold to give the reader a chance quickly
   to check the criteria for mechanical selection, or to find
   the relevant passages in a long item.

3. Provide easy adaptive algorithms that allow the search
   procedures to be modified with experience, e.g. changing
   the exclusions as one reads undesired stories.

4. Repeats and updates can be suppressed by a program that
   compares sufficiently large hunks of text; rather earlier
   versions can be suppressed, printing out the latest version
   at the time of reference.

5. Stories or messages queued for reading can be shifted
   around under some algorithm so that all the ones on the
   same topics occur in a bunch.

6. Automatic abstracts can be attached as headers.

7. The Today's Topics that Duffey kindly provides could be
   automated and expanded with a message saying that today
   you have N messages that meet your requirements out of M
   total messages. The table of contents of your messages is
   as follows: ...

8. As some recent messages have suggested, we can wait till AI
   enables us to have the computer understand the message and
   select it (or even edit it) for us intelligently.  I am a
   sceptic.  I'm sure it will happen, but not with adequate
   efficiency for the handling of unstructured, unconstrained
   text on systems of the eighties.  (Of course, its not an
   all-or-none matter; progress will start.)

9. Procedural discipline can be imposed on contributers requiring
   them to provide header information or organize their message
   under set categories.  This may be a bad idea, or maybe in
   moderation a good idea; I am not trying to judge these ideas,
   only list them.

10. Hierarchy can be imposed as to who can contribute what, or
    there can be recognized identifiers of degree of expertise
    or role of contributer.  I don't like the sound of that,
    but it does need to be thought about.  For example, in a
    newspaper we do read news stories, adds, editorials, and
    letters to the editor differently on the basis of our
    knowledge of the source.  Stories do say "a high official
    said", or "some persons in Washington believe"; it makes a
    difference.  Professional associations do have their
    reporting criteria.

11. Editors can be hired.  That is expensive.

12. ...

13. ...

14. ...

What else can be added?

I will be away from access to HUMAN-NETS for two or three weeks,
so if I don't reply soon to the various comments that this may
generate, its not impoliteness or disinterest.  I'll certainly
read them with interest and reply when I'm back.

[ Please see Communicating via Network - Organizing Presentation
  from [HNT V2 #107-113,119] for earlier discussion on the use of
  keywords mailing list applications.                      -- RDD ]