5.0033 Citations in the Humanities (3/126)

Elaine Brennan & Allen Renear (EDITORS@BROWNVM.BITNET)
Mon, 13 May 91 21:48:45 EDT

Humanist Discussion Group, Vol. 5, No. 0033. Monday, 13 May 1991.

(1) Date: 13 May 91 16:00:00 EDT (22 lines)
From: "Mary Dee (faculty" <mdharris@guvax.georgetown.edu>
Subject: "Short Citedness"

(2) Date: Mon, 13 May 91 14:14:17 EDT (72 lines)
From: amsler@starbase.MITRE.ORG (Robert A. Amsler)
Subject: 98% uncited Arts and Humanities literature

(3) Date: Mon, 13 May 91 08:25:30 EDT (32 lines)
From: Stevan Harnad <harnad@phoenix.Princeton.EDU>
Subject: Re: 5.0029 Troubling Citation Study

(1) --------------------------------------------------------------------
Date: 13 May 91 16:00:00 EDT
From: "Mary Dee (faculty" <mdharris@guvax.georgetown.edu>
Subject: "Short Citedness"

In Peter Denning's article, as well as Steve DeRose's commentary on
it, there is not a clear explanation of what is meant by "cited". Does it
mean that these journals under consideration were cross-indexed in order to
determine whether other articles in this set of journals referenced original
articles in the set? Or does it mean than of all printed material anywhere,
only these percentages of the articles in this set of journals wer cited?
There's a big difference. I would guess that one should count citations from
textbooks in the field as well as citations from journals.

Another fallacy in counting citations has to do with how the importance
of an article is to judged? If many folks read an article but no one cites,
does that make it unimportant? I also suspect that there certain articles,
oft cited, that few have read, such as Jay Earley's "An Efficient Context-free
Parsing Algorithm," in _CommACM_13(2), 1970, 94-102, which seems to be in
every disucssion on parsing I've seen (almost).

Mary Dee

(2) --------------------------------------------------------------86----
Date: Mon, 13 May 91 14:14:17 EDT
From: amsler@starbase.MITRE.ORG (Robert A. Amsler)
Subject: 98% uncited Arts and Humanities literature

Probably the meaning is cultural rather than reflecting any underlying
flaw. What citations mean in the arts and humanities and the sciences
are somewhat different. In the sciences one cites works to first
establish that one is aware of the relevant literature, i.e. knows
who else has been working in the area. One could be completely
discredited if one proposed something what someone else had proposed
beforehand and didn't cite them for it. One only secondarily cites
works to indicate that one agrees with them. Agreement is in fact
not a big factor in citations--one cites works to indicate that their
contents are being assumed to be true, regardless of whether one
likes the conclusions or not.

In the humanities, I'd assume one cites "classics" but by and large
tries NOT to cite works of rival schools at all, so as to not give
them any further publicity? Given a large amount of "subjective"
judgements one also would avoid citing works with which one didn't
agree, so as to not taint oneself with what in one's mind are
incorrect views. Finally, if one is attempting to establish one's OWN
reputation, one would be careful to deliberately only cite those from
whom one was trying to align oneself. Only more powerful people than
oneself and probably only people one had met so as to be assured of
their politically correct views. Thus, citation in the humanities is a
bit more like endorsement (or attack); rather than "observation".

However, the problem is what should we be doing, sciences or
humanities, to improve this situation. Citation itself is an artifact
of a 19th century printing technology. It is clearly a system in need
of a 20th century equivalent (so when we get to 2001 we'll then only
be a century behind again :-).

Minimally, I suppose I'd suggest the practice of
citation-equalling-direct-communication ought to be adopted; that is,
if one cites someone, the practice of that citation resulting in a
communication to the person cited ought to be achieved. In an
electronic text this would merely mean that publication would result
in electronic copies of the text itself going to the cited person. Why
should YOU have to track down who is citing your work? What an
inefficient system for the cross-fertilization of knowledge and work.

Second, I'd suppose citations ought to be accompanied by a reserved
spot for a response from the cited party. Thus, papers once published
would establish a set of openings for further comments back from the
cited parties and the whole should become "the literature". Thus, when
one came across a paper some years after its publication, one would
be able to find out what the cited parties thought of the use, etc.
being made of their work. Most probably wouldn't have responded, but
some of the replies might have been at least as valuable as the
orginal paper (or more so?).

Finally, the original author ought to have the option of
post-commentary on their own paper. Thus, if they publish more
material later, and their earlier paper becomes obsolete, they ought
to be able to annotate the earlier paper with comments. This should be
an ongoing thing, such that in effect an author could throughout their
life add future thoughts to their earlier works.

This message is long anough now, so I won't speculate about the
electronic means of achieving all this. Xanadu could probably handle
it all, but the existing on-line databases ought to be able to do most
of it without that much difficulty. What is clear though is that the
paper medium is poor at this. One could I suppose create a looseleaf
journal, for which one would receive new update pages to insert into
the text; but in a bound publication there seems no timely way to
handle these back-references from the future.

Robert A. Amsler
Washington, DC
(3) --------------------------------------------------------------43----
Date: Mon, 13 May 91 08:25:30 EDT
From: Stevan Harnad <harnad@phoenix.Princeton.EDU>
Subject: Re: 5.0029 Troubling Citation Study (1/60)

Signal, Noise and the Net

The low citation rate probably correctly reflects the high
forgettability of most published work (and the fact that most
work eventually gets published somewhere, despite peer review).
Does this mean standards are too low? Probably, but probably
also that all areas of human endeavor -- from height and weight
to athletic and intellectual performance -- conform to the
familiar gaussian distribution (bell curve), with the majority
of it straddling the average. In scholarship a certain fixed
and unflattering wheat-to-chaff ratio is inevitable, and the
only way to reduce the absolute quantity of chaff may be to
(China-like) hold down the entire enterprise.

So what we really need is a way of filtering out the quality at the
absolute level we happen to require. In my view, the electronic medium
(with suitable "gating" software, calibrated to the topics, authors,
institutions and levels in the "prestige" hierarchy that each of us
personally elects to hew to) will turn out to be the most rational and
efficient (if not the optimal) way of monitoring, filtering as well as
searching this gaussian sea of noise for the pearls of signal we
each seek.

Stevan Harnad
Department of Psychology
Princeton University