Xref: utzoo news.misc:5575 comp.groupware:322
Path: utzoo!attcan!uunet!zaphod.mps.ohio-state.edu!rpi!uupsi!sunic!dkuug!daimi!dsstodol
From: dsstodol@daimi.aau.dk (David S. Stodolsky)
Newsgroups: news.misc,comp.groupware
Subject: How to reduce "noise" on the net? (really) (repost)
Summary: Scientific peer review shows a way to reduce noise
Keywords: peer review priority refereeing
Message-ID: <1990Oct24.185636.13754@daimi.aau.dk>
Date: 24 Oct 90 18:56:36 GMT
Sender: dsstodol@daimi.aau.dk (David S. Stodolsky)
Organization: DAIMI: Computer Science Department, Aarhus University, Denmark
Lines: 54

In Message-ID: <1990Jul27.201053.18550@cunixf.cc.columbia.edu>
ezk@cunixe.cc.columbia.edu (Erez "HWank1" Zadok) asked:
>Are there any people who are actively implementing any new methods that
would reduce "noise" on the net?

Comp.groupware was created with the idea of doing this, and it was originally
proposed that two groups be created, one being a follow-up group so review
messages for "mutual moderation" could be transmitted separately. Creation of
the second group was rejected by the "newgroup" committee. My access to the net
has been a little shaky since, so I have not been posting directly. Some
discussion has and will appear in sci.psychology.digest relating to an
electronic publishing initiative by the APA.

The basic idea is that readers post review messages. An appropriate structure
would mean a better process than achieved by peer reviewed scientific journals.
The approach that I am thinking about now would use the new features in the
latest version of nn. When someone read an article, they could invoke a macro
that would ask a series of questions such as:

: is this post relevant here?
: is it original?
: is it correct?
: etc.
: etc. 
 .
 .
 .
 
responses could be on a scale, say 0 to 9, that would be presented as word
choices (i.e., very relevant ... totally irrelevant).

The resulting number (990 = very relevant and original, totally wrong) would be
appended to original title and posted. Nn has an option that allows such
articles to be collected (they would be easily identifiable to others, since the
body of the message would have zero lines) and treated as a unit. Thus reviews
and the original article could be processed by some macro that would look at the
authors and review ratings given to decide how important a given article was to
the reader. The articles would then be presented in priority order. The reader's
own rating of the article would be used to do a reputation database update after
the review was posted. This could also be done with HyperCard or some other
front end.

I will be posting the stuff that has built up on my machine in comp.groupware
and the first tests will likely be made there. We need to agree on a list of
standard questions for the header ratings, optional ratings could be placed as
additional header lines. If someone has a specific suggestion such as "this
belongs in newsgroup X," it could be the body of the message. Such comments
would probably best be mailed to the author, but old habits die hard.

--
David S. Stodolsky                  Office: + 45 46 75 77 11 x 21 38
Department of Computer Science                Home: + 45 31 55 53 50
Bldg. 20.2, Roskilde University Center        Internet: david@ruc.dk
Post Box 260, DK-4000 Roskilde, Denmark        Fax: + 45 46 75 74 01