[Humanist] 28.452 Big Data no boondoggle

Humanist Discussion Group willard.mccarty at mccarty.org.uk
Thu Oct 30 09:06:05 CET 2014


                 Humanist Discussion Group, Vol. 28, No. 452.
            Department of Digital Humanities, King's College London
                       www.digitalhumanities.org/humanist
                Submit to: humanist at lists.digitalhumanities.org

  [1]   From:    Paul Fishwick <metaphorz at gmail.com>                      (147)
        Subject: Re:  28.447 Big Data no boondoggle

  [2]   From:    Joris van Zundert <joris.van.zundert at huygens.knaw.nl>    (116)
        Subject: 28.447 Big Data no boondoggle


--[1]------------------------------------------------------------------------
        Date: Wed, 29 Oct 2014 11:10:26 -0500
        From: Paul Fishwick <metaphorz at gmail.com>
        Subject: Re:  28.447 Big Data no boondoggle
        In-Reply-To: <20141029080130.8B7A48F92 at digitalhumanities.org>


Willard

  Your analogy to microscopes is a good one, and I am enjoying reading
Hacking’s article. In prior digests, there has been discussion on modeling,
and mathematics by way of computing (which is built on mathematics). Models 
may operate similarly to microscopes in that they create new abstractions that can 
alter our epistemology. Jay Forrester’s System Dynamics is a way of seeing (via
a hydraulic analogy), as is using the method of diffraction for generating new
knowledge of matter.

-p


Paul Fishwick, PhD
Chair, ACM SIGSIM
Distinguished University Chair of Arts & Technology 
   and Professor of Computer Science
Director, Creative Automata Laboratory
The University of Texas at Dallas
Arts & Technology
800 West Campbell Road, AT10
Richardson, TX 75080-3021
Home: utdallas.edu/atec/fishwick
Blog: creative-automata.com



--[2]------------------------------------------------------------------------
        Date: Wed, 29 Oct 2014 21:57:22 +0100
        From: Joris van Zundert <joris.van.zundert at huygens.knaw.nl>
        Subject: 28.447 Big Data no boondoggle
        In-Reply-To: <20141029080130.8B7A48F92 at digitalhumanities.org>


Dear Willard,

Just to make sure: I wasn't wanting to be flippant, I sincerely was looking
for the questions behind your question.

To recapture and really get this correct let me try to paraphrase… The
epistemological problem here is that we know ever better there is more that
we do not know, but we also know that the more we uncover that it is always
with caveats of imprecision and false negatives. And most importantly we
have little clue how the digital layers between us and the data affects our
epistemology.

You write: "A threshold has been reached, I have crossed it and nothing
will ever be the same again." Does this mean that for you our episteme has
been 'amplified' but at inconvenient epistemic ramifications? I am now very
much intrigued about the nature of these epistemic uncertainties, if they
are that.

To humanists that tend to reason abductively, building syntheses based on
narrative logic, the greater visibility of the fact of 'too many data, too
many texts' could very well be unsettling, because we cannot delude
ourselves anymore with the hubris of 'having read it all'. This implies
more than just the platitude that we are getting to know better how much we
do not know. The problem is: how do we argue the validity of our narratives
knowing that many data may not have been covered by it? Still coherence of
argument goes a long way, I would suggest. But indeed it feels as if we are
getting in an epistemic unbalance. How will we argue and referee once we
find an article primarily based on text mined data diametrically opposed to
the conclusions of a well argued, well written, fully bibliographically
referenced coherent hermeneutically styled reasoning?

Another uncertainty could be that increasingly we will find narratives that
will be augmented with analytic results from 'big data' analysis. Most
certainly our tools are imprecise in recall, so we know we are getting
sub-optimal answers–and even if they were perfect they would miss what is
not or can not be digitized. But more importantly I think: the queries and
the query algorithms of these tools are always (certainly still at this
time) a highly imprecise translation of the questions that we want the
recalled documents to answer to. The theory and practice of building
software and computational algorithms strikes me as far more blurred and
multiform even than those underpinning what we can reason about seeing with
a microscope. That is where I think some urgent epistemological questions
are waiting for us: how do we evaluate the inner hermeneutics of
algorithms? They are not right because they work, is the point.

These are indeed important epistemological problems. My guess is we
actually agree on that. That would not further the discussion very much. On
the other hand, I may hope I'm still missing half of the problem.

All the best
--Joris

On Wednesday, October 29, 2014, Humanist Discussion Group <
willard.mccarty at mccarty.org.uk
>
>                  Humanist Discussion Group, Vol. 28, No. 447.
>             Department of Digital Humanities, King's College London
>                        www.digitalhumanities.org/humanist
>                 Submit to: humanist at lists.digitalhumanities.org
>
>         Date: Tue, 28 Oct 2014 08:23:14 +0000
>         From: Willard McCarty <willard.mccarty at mccarty.org.uk>
>         Subject: Big Data thresholds and tolerances
>
> Dear Joris,
>
> You say that,
>
> > .... If you are talking just about how
> > humanities faces big data because of things like JSTOR, then the
> > epistemological problem is not new, right? We always knew there were far
> > more journal articles, monographs, information, and data out there than
> we
> > would ever be able to find and to gauge. Maybe digital archives just put
> > that problem more clearly in our face. That again is an effect of what we
> > call tongue-in-cheek Daniel O'Donnel's first law of computing: problems
> are
> > not so much created through computing, but they are magnified manifold by
> > it.
>
> We could say, and I would agree, that *in principle* the problem is very
> old. But that's not how we used to think. When I wrote my dissertation
> on Milton's Paradise Lost in its relation to biblical and classical
> literature (late 1970s-early 1980s), it was still assumed that I would
> read everything that had been written on that topic, e.g. all criticism
> up to that time, all the major works of Greek and Latin literature, all
> of Augustine and so on. I did read quite a bit but not all. I did
> actually finish the thing, though it took me 8 years. I would assume
> that nowadays if anyone at all works on Milton no such assumption is made.
>
> There are thresholds past which different things happen. The problem I
> was really thinking of was, however, not merely the known or estimable
> volume of relevant literature but the ease with which I can find out
> about and obtain items. The failure of mechanisms for retrieving items
> with the best precision/recall ratio in combination with natural
> curiosity in combination with that ease is the difference that has made
> a difference. A threshold has been reached, I have crossed it and
> nothing will ever be the same again.
>
> The microscope (or I should say all kinds of microscopes) only magnify.
> You could say all that stuff has always been there, so what's the big
> deal? I'd say, go read Hacking's "Do we see through a microscope?" and
> then think again. So I'd argue that yes, we do have a new
> epistemological problem, at least in practice.
>
> Yours,
> W
>
> --
> Willard McCarty (www.mccarty.org.uk/), Professor, Department of Digital
> Humanities, King's College London, and Digital Humanities Research
> Group, University of Western Sydney


-- 
Drs. Joris J. van Zundert

*Researcher & Developer Digital and Computational Humanities*
Huygens Institute for the History of the Netherlands

*Royal Netherlands Academy of Arts and Sciences*
 http://www.huygens.knaw.nl/vanzundert/
http://www.huygens.knaw.nl/vanzundert/
 http://www.huygens.knaw.nl/vanzundert/?lang=en

-------

*Jack Sparrow: I thought you were supposed to keep to the code.Mr. Gibbs:
We figured they were more actual guidelines.*





More information about the Humanist mailing list