Humanist Discussion Group

Humanist Archives: July 25, 2022, 9 a.m. Humanist 36.108 - scale, continued

				
              Humanist Discussion Group, Vol. 36, No. 108.
        Department of Digital Humanities, University of Cologne
                      Hosted by DH-Cologne
                       www.dhhumanist.org
                Submit to: humanist@dhhumanist.org


    [1]    From: Tim Smithers <tim.smithers@cantab.net>
           Subject: Re: [Humanist] 36.102: scale, again (164)

    [2]    From: David Zeitlyn <david.zeitlyn@anthro.ox.ac.uk>
           Subject: Re: [Humanist] 36.102: scale, again (26)


--[1]------------------------------------------------------------------------
        Date: 2022-07-14 09:36:06+00:00
        From: Tim Smithers <tim.smithers@cantab.net>
        Subject: Re: [Humanist] 36.102: scale, again

Dear Willard,

From the things I know about and have worked on -- which is
not a lot -- different scales need different levels, with one
level per scale. The levels are what give the relationship
between things at different scales, and coherence to the
different things we "see" at the different scales.

For example, Allen Newell, in "The Knowledge Level" [1],
presented what he took to be the generally accepted levels of
a [digital] computer system to be.

     Program (Symbol) Level

     Register-Transfer Level

     Logic Circuit Sublevel

     Circuit Level

     [Electronic] Device Level

And argued that on top of this system of layers we have the
Knowledge Level, where knowledge is characterised as a
capacity for rational action, and not, as was more usual in AI
back then, as some kind of represented stuff.

You could, with appropriate equipment, observe the behaviour
of some particular (Symbol Level) program at the Device Level,
and the electronic currents that flow between the devices, as
the program is executed. This electronic behaviour completely
determines the behaviour of the program, given the way we
design and build digital computers, but hardly helps us
understand what the program is doing, and how.

At the different layers above this, and thus at the different
scales we might look at the working of our program, we get a
different view, and a different understanding of what's going
on. But, though the views we have from these different levels,
or scales, are quite different, based, as they are, upon
different ontologies -- things that exist at the Device Level
don't exist at the Circuit Level, circuits do, and so on up
the Levels -- there is a well specified mapping from one level
to the next, above, and below. Mostly, this mapping is built
from encapsulation and abstraction: particular combinations of
electronic devices are encapsulated by a particular type of
electronic circuit, and combinations of particular circuits
are encapsulated by a particular type of [binary] logic
circuit (we need for building a computer.)

This system of coherent levels gives us a ladder we can use to
move up and down the different scales at which we may properly
look at, and understand, the behaviour of some program, all
the way up, according to Newell, to the Knowledge Level, at
which we see what the executing program knows -- performs
rational actions on.

A different, but related, system of levels was proposed by
David Marr in an attempt to present a computational theory of
vision, see Wikipedia: David Marr (neuroscientist). Marr
proposed that this was composed of three levels:

    The Computational Level: what the system does

    The Algorithmic Level: how the system does what it does

    The Implementational/Physical Level: how the system is
     physically realised

And, there are, across different disciplines, other systems of
levels, for Natural things, such as subatomic particles,
atoms, molecules, etc, and artificial systems, like computers
data base systems, buildings, cars, televisions, etc,
including (so called) Large Language Models.

So, for me, to ask about different scales is to ask about the
system of different levels that properly separate out the
scales at which we can look at something we are interested in,
and which provides the coherent way of relating what we see,
and can understand, at one scale to what we see, and can
understand, at another level.

Take, as you did, texts. If (Moretti-like) "distant reading"
is at one scale, and "close reading" is at another scale, in
the same system of levels, what, I would ask, is the ontology
at each of these levels, and what is the mapping between them?
And, are these two scales immediate neighbours -- "distant
reading" is next up from "close reading" -- or are there other
levels in between them? If so, what are these other levels?

A close reading of Newell's Knowledge Level paper might, I
suggest, we a useful place to start in looking at this
question, in part because I think Newell's notion of knowledge
-- as a capacity for rational action -- might offer a more
practical, and thus more useful, concept of knowledge for
understanding what we can know from a "close reading," and
what we can know from a "distant reading," and how we might
map one way of knowing to the other way, than the classical
notion of knowledge as justified true belief can, which
remains difficult to use in practice.

Just some thought from how AI used to be done :)

Best regards,

Tim


Reference

[1] Allen Newell, 1982. The Knowledge Level, Artificial
    Intelligence vol 18, pp 87-127.
    <https://cs.uns.edu.ar/~grs/InteligenciaArtificial/Allen%20Newell%20-%20The%
20knowledge%20level.pdf>




> On 13 Jul 2022, at 09:06, Humanist <humanist@dhhumanist.org> wrote:
>
>
>              Humanist Discussion Group, Vol. 36, No. 102.
>        Department of Digital Humanities, University of Cologne
>                      Hosted by DH-Cologne
>                       www.dhhumanist.org
>                Submit to: humanist@dhhumanist.org
>
>
>
>
>        Date: 2022-07-13 06:54:32+00:00
>        From: Willard McCarty <willard.mccarty@mccarty.org.uk>
>        Subject: scale, again
>
> Lest Alan Liu's helpful reference to Zach Horton's very interesting
> book, The Cosmic Zoom, end the discussion of scale, allow me to ask
> again, or further, what's going on with scale in digital studies of
> texts, or populations, or pretty much anything. Take texts, for example.
> Is it not the case that both Moretti's "distant reading" and the much
> earlier "close reading" of literary criticism suggest getting to the
> truth of the matter, or at least doing much better than foggy notions,
> say of genre or the meaning of an individual work of literary art,
> formed by the unaided reader? Do mechanisms affording or merely urging
> on macro- or microscopic views tell us what is really going on or open
> up other dimensions of reality? Without these mechanisms, are these
> other dimensions merely inaccessible or are they non-existent?
>
> Perhaps, given NASA's photographs, we should be asking (with reference
> to Horton) about how these visions of the new, or newly seen, are
> generated, or as we have learned to say, mediated. Photographers know all
> about the creative element of their imaging processes, within the camera
> and in the workflow that follows, nowadays in software, formerly in the
> darkroom. I think now, in the opposite mode, of Antonioni's Blow Up
> (1966).
>
> Comments?
>
> Yours,
> WM
> --
> Willard McCarty,
> Professor emeritus, King's College London;
> Editor, Interdisciplinary Science Reviews;  Humanist
> www.mccarty.org.uk

--[2]------------------------------------------------------------------------
        Date: 2022-07-13 08:52:21+00:00
        From: David Zeitlyn <david.zeitlyn@anthro.ox.ac.uk>
        Subject: Re: [Humanist] 36.102: scale, again

Dear all

this discussion reminds me that Steve Woolgar organised a conference in 2009 on
matters of scale - and coined (I think) the idea of scalography

One publication resulting is

Olli Pyyhtinen, "Matters of Scale: Sociology in and for a Complex World". 
The Canadian Review of Sociology, 2017, Vol.54 (3), p.297-308

[This is behind a paywall but accessible through many institutions; the following 
is for those that have access to the Bodleian at Oxford--WM]

<https://solo.bodleian.ox.ac.uk/primo-explore/fulldisplay?docid=TN_cdi_proq
uest_miscellaneous_1927834551&context=PC&vid=SOLO&lang=en_US&search_scope=LSCOP_
ALL&adaptor=primo_central_multiple_fe&tab=local&query=any,contains,scalography&o
ffset=0&pcAvailability=true>

best wishes
davidz

--
Professor David Zeitlyn  ORCID: 0000-0001-5853-7351
Institute of Social and Cultural Anthropology (ISCA), University of Oxford, 51
Banbury Road, Oxford, OX2 6PF, UK.
https://linktr.ee/mambila


_______________________________________________
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