Humanist Discussion Group, Vol. 36, No. 35.
Department of Digital Humanities, University of Cologne
Hosted by DH-Cologne
www.dhhumanist.org
Submit to: humanist@dhhumanist.org
Date: 2022-05-26 08:44:25+00:00
From: Manfred Thaller <manfred.thaller@uni-koeln.de>
Subject: Re: [Humanist] 36.32: in the dark
Dear Willard,
in the very late seventies I used the computing center available the
Max-Planck-Society institutes at Göttingen, dedicated to sub-fields of
physics, chemistry and medicine, besides the history institute I worked for.
Once I had a conversation with one of the physics people who asked what
software we actually used. Mentioning that besides the data base
software I developed myself I point to SPSS for statistical analysis.
The reaction was: "Well, that may be fine for you. We as physicists have
to know exactly how our equipment works. We cannot accept an analytical
tool which might have unknown bugs introduced by somebody else." I
should point out, that that was the institute where in 1948 magnetic
drum storage was invented and which in the early fifties built its own
"electronic computers" from wires, bulbs and transistors, so the claim
as such had a certain plausibility.
I have thought of him quite frequently over the years, particularly
about ca. 2010, when I was involved in a funding proposal for a super
computer center. I really wonder if any of the physicists working there
had any clue what glitches in the software controlling distribution of
algorithms about many thousand processors might actually influence the
outcome.
> but (I hear the objections quickly forming) the machine is
> largely a black box.
Well, the objections you hear quickly forming, include probably this one:
In the computational part of your work the process can "in principle" be
controlled at an extremely fine resolution. How serious you take that is
up to you, though.
If you rely on intuitive insights, you never get a chance.
The question is, of course, how serious you are about the control over
the research process.
Anecdote 1: A researcher in whose MA thesis in medieval history I was
involved (who in the meantime holds a chair in Digital Humanities),
submitted his thesis at the history department, where his supervisor
insisted that he drop all descriptions of the computing procedures from
the text of the thesis "as these are just irrelevant details".
Anecdote 2: At a recent one day introductory course to a tool supporting
Latent Dirichlet Analysis the author of the tool and teacher of the
course is quoted as saying. "Do not worry, that is very complex. But as
I have produced the tool, you as literary scholars do not have to
understand how it works."
What the two have in common, is the assumption that you safely can close
your eyes; but that assumption is one you make explicitly - or should,
or at the very least can.
[ By the way, I remember my brother (amateur photographer) to work with
one of the bags you describe. His results improved noticeably, after he
found ways and means to fit a small room in our house as a dark room,
replacing the bag. Is it really a virtue intentionally to cripple
yourself? ]
> Are we not REALLY in the dark with computers -- and so necessarily
> writing close to "degree zero"? So much for 'objectivity', the less so the
> greater the amount of data, the closer to complexity the process gets?
I think you mix two issues here. At the level of computation nothing is
hidden "in principle" (see below). The data ...
When in the earlier seventies I was involved in those SPSS based
statistical analyses "reliability of codes" was a big issue. In big
projects: How could you be sure, that two or more people encoding, say,
occupational terms, arrive at the same data? At the individual level:
How could you be sure, that your own encoding remained consistent over a
few months of data acquisition? I remember engaging in quite laborious
tests trying to ensure this sort of reliability.
Hm. How many tests have there actually been, how consistent the
application of a slightly fancier encoding, say CIDOC CRM, is within
some of the projects currently employing it?
But, of course, that again could be done "in principle".
"In principle": The question is how much control DO you want to have?
For me computation provides a much higher potential degree of control
than any non-formalized research process.
How much of that degree of control you realize depends on many things:
Your institutional setting: If as a young researcher you have to dazzle
respected traditional literary scholars, who are convinced that
technology is much below their dignity, by a wonderful visualization,
explaining what's behind the dazzle is clearly counter-productive to
your career.
Generalized: How much your discipline values prove of your results -
depends on the discipline, much more than the tools.
> So much for 'objectivity', the less so the
> greater the amount of data, the closer to complexity the process gets?
Well, personal confession, as I have written in another context:
Historians who believe that absolute objectivity is possible are fools,
historians who do not strive to achieve as much objectivity as possible,
are intellectual cowards.
[ Or have to survive living together with cowardly colleagues. ]
Or: How much we are in the dark with computers depends on one thing
only. Us.
Kind regards,
Manfred
Am 26.05.22 um 08:16 schrieb Humanist:
> Humanist Discussion Group, Vol. 36, No. 32.
> Department of Digital Humanities, University of Cologne
> Hosted by DH-Cologne
> www.dhhumanist.org
> Submit to: humanist@dhhumanist.org
>
>
>
>
> Date: 2022-05-26 06:01:05+00:00
> From: Willard McCarty <willard.mccarty@mccarty.org.uk>
> Subject: in the dark
>
> A philosopher of science I know has complained that there are good
> grounds for questioning the veracity of many papers he reads. The
> problem is, he says, that they often are more concerned with justifying
> their conclusions than re-enacting the routes taken to reach them. Even
> when the intent is clearly to report on the steps taken, in a rhetoric-free
> "writing degree zero" (Barthes), the complex processes involved are
> simplified, setbacks left out, confusions and controversies passed over
> and so on. My former colleague at Toronto, Russ Wooldridge, liked to
> point out that in much published research in the humanities which used
> computing, the computer quickly disappeared into the background
> once useful results were obtained, thus obscuring the means, the
> setbacks, the controversial moves -- and perhaps more interesting
> results (positive or negative) than those reported.
>
> In research involving a computer, is not the problem worse in
> principle than in other experimental work? One has the data,
> of course, and the software, and knowledge of how the machine
> works--but (I hear the objections quickly forming) the machine is
> largely a black box.
>
> Let me use an analogy. If you're old enough to remember chemical
> photography, you may recall the photographer's developing bag, black
> inside and absolutely light-proof, with elasticated holes for the arms
> so that film and developing apparatus could be manipulated without
> exposure to light when a darkroom was not handy. One got quite good
> 'seeing' with one's hands. (Those with impared vision get to be very good
> at navigating the world without visual help, of course. They are the
> experts here.)
>
> Are we not REALLY in the dark with computers -- and so necessarily
> writing close to "degree zero"? So much for 'objectivity', the less so the
> greater the amount of data, the closer to complexity the process gets?
>
> Comments?
>
> Yours,
> WM
> --
> Willard McCarty,
> Professor emeritus, King's College London;
> Editor, Interdisciplinary Science Reviews; Humanist
> www.mccarty.org.uk
--
Prof.em.Dr. Manfred Thaller
formerly University at Cologne /
zuletzt Universität zu Köln
_______________________________________________
Unsubscribe at: http://dhhumanist.org/Restricted
List posts to: humanist@dhhumanist.org
List info and archives at at: http://dhhumanist.org
Listmember interface at: http://dhhumanist.org/Restricted/
Subscribe at: http://dhhumanist.org/membership_form.php
