File /Humanist.vol22.txt, message 425

From: Humanist Discussion Group <>
Date: Thu,  8 Jan 2009 06:32:12 +0000 (GMT)
Subject: [Humanist] 22.430 theory and experiment

Humanist Discussion Group, Vol. 22, No. 430.
Centre for Computing in the Humanities, King's College London
Submit to:

[1]   From:    Willard McCarty <>           (9)
Subject: "theory"

[2]   From:    Willard McCarty <>          (75)
Subject: experiment to theory

Date: Wed, 07 Jan 2009 06:50:00 +0000
From: Willard McCarty <>
Subject: "theory"

Those here still interested in use of the word "theory" may find useful
Andrew Chesterman's "On the idea of a theory", Across Languages and
Cultures 8.1 (2007): 1-16.

Willard McCarty, Professor of Humanities Computing,
King's College London,;
Editor, Humanist,;
Interdisciplinary Science Reviews,

Date: Wed, 07 Jan 2009 11:33:20 +0000
From: Willard McCarty <>
Subject: experiment to theory

Following the discussion on prototypes and theories with respect to
computing, I've found a good example in an old article I've mentioned
before, Margaret Masterman, "The Intellect's New Eye" (1962). Making a
case for a truly revolutionary outcome of computing, Masterman cites the
work in classification that was then coming out of combinatorics. Here
is what she says:

> As an example of this new power, and following up Leibniz's thought,
> consider in general the activity of classification. Long before the
> seventeenth century, and long after, thinkers ranging from Aristotle
> to Nelson Goodman, interested in exploring the general nature of
> classifying, had not yet succeeded in obtaining any new fundamental
> knowledge of it. Now, however, using digital computers, a new and
> elegant mathematical theory of classification is being developed all
> over the world. To get the feel of this, consider the following
> problem: You have discovered 150 characteristics, characterizing 100
> so-called "species" (say of tapioca plants, or diseases, or sorts of
> words, or what-have-you). The problem is to find out how these newly
> discovered characteristics re-group up the 100 antecedently named
> species. Seen in terms of the groupings of these 150 characteristics,
> for instance, have we really got 100 species of tapioca plant? If
> not, how tnany have we? Human intuition, unaided, fails here; for we
> have too much classifying material. By making a 100 X 150 array,
> however, into the squares of which we put a 1 wherever a species has
> one of the characteristics, we can provide basic long-range data for
> a computer. And by adopting some agreed-to-be-satisfactory criterion
> of similarity between any new species, we can provide it also with
> its immediate fodder. (Tanimoto's criterion of such similarity, for
> instance, is the number of properties in common between the two
> species divided by the number shown by at least one of themso that,
> if the two species had three properties in common, the first having
> five altogether and the second six, the coefficient would be 318.)
> This preparation achieved, a whole new classificatory countryside
> opens out; for, using the computer, the measures of similarity thus
> obtained can be compared, ordered and clustered, and the resulting
> clusters or clumps recompared with the original data, which may then
> suddenly appear in a new light. The range of application of this new
> kind of analysis is clearly enormous; already it is being, or has
> been, applied to classificatory problems in information retrieval,
> linguistics, medicine and anthropology. But that's only the start of
> the potential range of application; and that's only the start of the
> development of the theory. For, once you have begun to think in this
> new kind of way, it becomes clear that other nonobvious criteria of
> similarity can bc tried out with sometimes the most unexpected
> effects.

Now what Masterman describes depends for its newness on a new "order of
power" (her earlier phrase) beyond what would be practical for a human
being to undertake. Hence what she points to isn't new with respect to
thought -- if it were it could not be programmed -- rather new with
respect to action. Afterward, when the action has been carried out and
results obtained, the new thinking comes in the attempt to theorize what
has been discovered. Thus experiment leads to theory, or at least to
new ideas.

When we in our nervousness to justify ourselves reach for something new
to pull out of our collective hat, we tend to dismiss whatever *could
have been done before*, i.e. in theory. But what we find in the hat are
not new theories, or not primarily, but ways of acting extensible beyond
first what one would do, then what one might be inclined to do, then
what one could do, then what an old fashioned German academic project
could do in more than a century and so forth. In other words the
criterion is wrong. A sign of this is in our turning to broad
socio-intellectual consequences of computing, which are very important
indeed but hard to argue beyond the anecdotal at this early stage. And
besides, we need the social scientists for that sort of thing.

Whether one cites Genesis or John, action and word are one in that domain,
as is certainly not the case with us here below. Does not "Im Anfang war die
Tat" betray the confusion of a thoroughly fallen mind?



Willard McCarty, Professor of Humanities Computing,
King's College London,;
Editor, Humanist,;
Interdisciplinary Science Reviews,

List posts to:
List info and archives at at:
Listmember interface at:
Subscribe at:


Humanist Main Page


Display software: ArchTracker © Malgosia Askanas, 2000-2005