Counting Things and Interpreting Ideas:
  Anthropological Conventions in the Use of
"Hard" Versus "Soft" Models

  Michael Fischer, University of Kent

  One cannot move from the informal to the formal by formal means alone.
  fortune, UNIX BSD 4.x


Past anthropological forays into the natural sciences have been motivated by a
search for tools with which to describe ethnographic settings in more
"rigorous" terms. The typical result of these forays have not always been
satisfactory: good descriptions of natural resource management systems were
produced that contained but pale reflections of the principal subject, people
and their activities.

I suggest  that these failures are due to implicit conventions in anthropology -
in particular notions pertaining to numerical representation - regarding what
to observe and how. This is seen most clearly in anthropology's use of
computer science. There were early successes with simulation models in
anthropology, in which (e.g.) the constructive semantics of kinship were
played out over whole populations, food procurement strategies were
reconciled with local ecology, and tabla improvisations were related to the
constituent "formulae" used by musicians.

These successes were not followed up, however, because the nature of early
computer representations was not appropriate for anthropology. The story
with attempts to incorporate natural science theories, methods and models have
followed much the same pattern.

Depending on the contstituency these failures are typically attributed to one of
the following reasons:

a) matters involving people are intrinsically "too complex" for natural science
derived methods
b) social scientists are too imprecise in their "measurements", too lax in their
methods, and lack any serious form of testable theory

I shall stand to the side and suggest another reason: Natural science methods
were developed to solve problems which are different from those social
scientists pursue.

Without qualification we could argue that this is a variant of either a) or b)
preceeding. That is, either a) social phenomena is intrincially too complex for
natural science derived methods to apply or b) social scientists are asking the
wrong questions, biting off more than they can chew.


 

  Michael Fischer     2

As a scientifically trained anthropologist, who is also a computer scientist, I
have a third reading:
Social phenomena is of a different order from the phenomena which natural
science has been successful in addressing,  but that the failure to successfully
adopt NS methods has largely been because of a failure to address these
differences rather than an intrinsic barrier to adoption.

Natural science has been most successful at describing phenomena which are a)
recursive, b) structural, c)  not self-referential, and d) similar to a high degree  
with respect to different projections (manifestations in different contexts).

By recursive, I mean that the phenomena can be described as the same object
over and over again, and where higher order organisation resembles lower-
order organisation. The same properties, forces and consequences recur. By
structural, I indicate static or time duration of limited extent. Self-referential
indicates an ability to alter relationships based on a moment-to-moment
examination of a process. Projection refers to the playing out or expression of
a model in a given context.

Social scientists rarely find these conditions. The level of recursion almost
always is 1. The different levels of the situation are driven by different
principles and act on different objects. Societies are highly self-referential in
other than the very short term, and individuals on a scale measured in minutes
or hours.

This does not mean that intrinsically the fundamental ideas that apply to NS
research cannot be applied to SS, but it does suggest these will have to be
formulated in an entirely different way.

The other dimension here, of course, is why should SS adopt methods from an
NS vantage at all. If present methods from NS are not largely applicable to SS,
why not develop new methods from the ground up. I believe that we have to
do a little of both.

Authority

That this issue arises at all is not for intrinsic reasons but because of how NS
and SS are applied, and thus perceived, in the world. NS have a proud
tradition of being perceived to "work". They were the driving force behind
technological progress and getting things done. SS was thought to be capable
of working, but largely didn't. NS could happily assume this was because they
were doing something wrong, but it really didn't matter much to them. SS had
little impact on their world. The opposite was not necessarily true.

The reason there is an issue now is that this situation has changed. NS see SS as
competitors, both because there is direct competition in some applied areas,


 

  Michael Fischer     3

and because some SS are attempting to undermine the position of NS as a
purveyor of influence or truth.

Both of these conditions came about because of two factors: greater ability to
examine information due to computers, and the partial or wholly failed
attempts to transfer technology to areas different from the development of
technology or unsuccessfully competing with indigenous technology when
transferred.

Both of these provided indications that NS-based methods did not always
"work" at an optimum level. These were the NS at their most exposed:
embedded in the "real world" of people, processes and knowledge.

With this new "power" SS responded with two apparently divergent tracks,
scientism, and what has come to be known as postmodernism.

Scientism

Conventionally, anthropologists tend to adopt NS methods directly from NS
disciplines. In some cases this is appropriate, for instance evolution and
genetics apply as well to people as to all other living things. But there is also a
great deal of importation into areas not so well suited. One of the earliest
failures in importation in anthropology was applying standard evolutionary
theory to socieites and cultures, languages, and technological development.
This is not to say that there is not some intrinsic merit to the notions that
societies evolve or change over time, but that the model applied was a bare
modification of one which applied to living things, which have entirely
different conditions for reproduction and survival.

There has been considerable comment on the relation between 'scientism',
scientific knowledge and modes of inquiry, and other 'modes' of knowledge.
There is bound to be some disagreement between anthropologists, in both their
scientific and humananist poses, and other scientists. Standing back and looking
at different kinds of knowledge, although the question of equivalence is a
difficult one, we should have to say that there is an existence proof that most
indigenous bodies of knowledge are successful with respect to the particular
way of life with which these are associated, and that these same bodies of
knowledge will appear defective to most participants in other bodies of
knowledge. However, this is not the case with all possible bodies of
knowledge. One of the problem that many people (anthropologists included)
appear to suffer from is to assume that any idea that they may have is valid.
'New' knowledge requires a means of integration and reproduction, and
potential death, wherever it is to persist.

About 20 years ago multicultural matters assumed more importance, and
gradually much of anthropology has been appropriated from a genre of


 

  Michael Fischer     4

generalising situational knowledge to translating situational knowledge from
other cultures into situational knowlege for western socieities. This
'application' of ethnography and anthropological matters is welcome in many
respects. The baseline of a successful locus of knowledge is the ability to
incorporate with and between other locii.

Postmodernism in anthropology

Much as anthropologists adopted scientific philosophy after it had be shaped by
natural scientists, anthropologists adopted postmodernism after it had been
shaped by literary critics. Note that there is no necessary relationship between
postmodernism and literary criticism, other than both deal with language to
some extent, Derrida in particular sometimes dealt with writing, and PM and
LC deal with complex semantics and pragmatics. This may make post-
structural philosophy applicable to literary criticism, but does not make LC a
necessary stage.

To accomodate this importation, early adopters recast the data of anthropology
as a text,  in the literal sense of an ethnography as a text, conversations as oral
texts, and, by extension, the claim that lived lives are like texts. There appears
to be little motivation for this other than the need to shape the data to pre-
existing applications of post-structural theory in literary criticism.

The only possible contender for an independent motivation was the
examination of written ethnography as a specific genre, although as a specific
movement this came rather late in the day in anthropology, probably
introduced by G. Marcus in 1983 or 1984, although one could credit Boone in
the late 1970s with this innovation. This also marked the beginning of the PM
examination of anthropology as a science, drawing it into the PM History of
Science movement, which also appears to have been heavily influenced by LC.

As it is practised in anthropology  PM now seems to focus on three basic
principles. First, that the form of anthropology for the past century or so is
attributed to the power wielded by a few key men and women in the discipline
and their relationship to other wielders of power outside the discipline.
Second,  that there are no factual, essential or 'authoritative' positions. Third,
that social phenomena is a illusion non-reductively constructed by the
interactions of unique individuals, in a process often described as negotiation.

Only the first emerges from classical post-structuralist philosophy (e.g.
Derrida, Foucoult, Lacan). The second is a consequence of the first in
conjunction with the third, which itself perhaps serves as the basis for making
the people in a society the authors of texts that anthropological PMs can
discuss.


 

  Michael Fischer     5



In short, APM, via LC, transformed a philosophical theory questioning the
linkage between representation and semantic analysis of symbols and favouring
linkage to political pragmatics, into a practice which dealt with texts, and
fictional ones at that. One could almost suggest, if one were a PM, that simply
transforming social productions into texts was not sufficent, these had to be
transformed to literature (fiction) as well to suit the imported methodology of
LC.

Convergence

Thus, if some anthropologists have erred in attempting to adopt NS methods
unmodified, I claim that so have others in their adoption of post-structuralist
theory and methodology.

From a personal view, I am facinated by the divergence between what I saw in
Derrida some 25 years ago when I accidently ran across his Critique of
Hussurl's "On Phenomenology", in which his essay "Differance" appears.

I had already gained a healty respect for philosophy as a source of theoretical
and analytic ideas through the work of Richard Montague, who demonstrated
by example that there were no significant differences in the semantics of
formal and natural languages, laying any additional complexity of natural
language to pragmatics (meta-intensionality). He began to pursue pragmatics,
but was unfortunately murdered (some suggested by  agents of the MIT
linguistics department which he constantly taunted) before that work was
complete. Perhaps most significant to me though was his demonstration that an
identical syntactic form could have a very large number of entirely different
derivations for a single corresponding semantic form. That is, formally
speaking, one did not require a deterministic one-to-one correspondance
between the process of deriving one form from the other, and that this was not
a defect, as the converse, multiple semantic forms for a single syntactic form
(ambiguity), was often seen.

I had also just discovered the work of Rene Thom on the use of topology, in
the form of Catastrophe Theory, to formalise irregular, discontinous
qualitative systems.  The major application focus of Catastrophe theory at the
time was representing discontinuous phenomena, such as stock market crashes,
and agression,  ignoring Thom's examples, which were too void of content to
be truly useful scientifically, but demonstrated the possible scope - the origins
of language, culture and the morphogenisis of the genome into an individual.
Extraordinary stuff, for the moment lost on the trashheap of ideas because of
the triviality of his followers. Perhaps the most interesting result of Thom's


 

  Michael Fischer     6

work was his 'proof' that any real processural phenomena could be described
in at most 7 dimensions, and that any phenomena, real or imaginary, in at
most 11 (which resonates well with the hypothesised 10 dimensions of super-
string theory in physics). In other words he proved there was a finite limit of
complexity to any process (for Thom worked in processes or structural
unfoldings). It could be very complex, but there were limits.

I saw Derrida much in the same vein, teasing out the implications of features
of coded representation, providing possible motivation for tools to describe in
more precise form those things that had been intractable. In particular I was
taken by his gloss for differance, as an idea/structure that was the same but not
identical. But again, what I saw as the most striking feature of Derrida's
framework has gone largely unnoticed, and given the development of critical
theory, contravened. Effectively in this work Derrida more or less demolishes
the philosophical position of phenomenology, and in a later work "Of
Grammatology", destroys semotics and semiology, and attacks the structure of
the symbol in general.  

To me, this work represented a basis for reducing the scope of semantic
meaning, and  venturing forth into the territory of pragmatics, or the playing
out of ideas in "real life" (Thom's dynamical structural unfolding, Montague's
pragmatics or meta-intensionality). From these ideas I assembled a montage
for formalising ethngraphic description which I have been adapting to
research material ever since, most recently in the semantic dennotation of
ethnographic multi-media representations. However, I seem to represent an N
of 1 in travelling along this particular vector, always a quality indicator of the
common crank.

Of these three, Montague and Thom have had virtually no impact at all in
anthropology, though Thom has had some influence in biology and Montague
in linguistics. Derrida, of course, serves as much of the base for motivating
postmodernism.

Annealing

Restating the converse of the kinds of phenomena that NS is "good at", what I
see as the main characteristics of phenomena that SS engage are:

a) not generally recursive or limited recursion,
b) processural (structural unfolding),
c)  self-referential (meta-intensional or pragmatic),
d) dissimilar to a high degree with respect to different "points of view"
  (projections or perspectutual unfoldings).


 

  Michael Fischer     7

However, this is not really true, any more than the NS version of this is. One
of the points of the critque of science is that it tends to ignore that when these
are unfolding at the level of technology or science embedded in the world of
people. Both kinds of phenomena must be considered, and "successful"
analyses will integrate both. Likewise, social scientists, of whatever
persuasion, must not imagine that they can ignore the impact of physical
phenomena on the social.

This indeed affects social scientists rather more than NS, because there are
many phenomena that NS can descibe where this embedding is not a factor. It
becomes a factor in NS in its unfolding into the world of living things,
especially people. However SS are never isolated from the world of physical
forces, because SS are always looking at unfoldings or the results of
unfoldings. SS sometimes try to get around this by keeping discussion rather
philosophical, or by inventing conceptual forces analogous to physical forces,
but this is simply evading the problem.

Integration

I can only speak for anthropology here, but one road to integration that is
compatible with both scientistic and PM approaches might be the adoption of
two techniques developed for representing complex information on computers,
declarative resolution logic and production systems.

Computers are ideal systems for representing unfoldings. Computer programs
always consist of some initial state of information and a method for
transforming that information to another state.

Declarative resolution logic is an approach to expressing this information
structurally, and then unfolding these declarations to their consequences. A
statement is represented by a goal and a declarative defintion in terms of either
other goals (which themselves must be "proved") or given data. In practice,
one can either

a) specify the goal, and then the goal is "proved" if it happens in the process of
  unfolding, inconclusive if it doesn't,
b) find all the possible goals for a given definition for a give set of intial
  information (e.g. all the goals that are "proved"), or
c) partially specify the goal, and get some results of type a. and b).

Goal1 IS  PROVED  IF Goal2 AND Goal3 And Goal4
example:

harvest crayfish  if  correct season  and  recent rainfall


 

  Michael Fischer     8


Production systems are a conceptualisation of deriving global results from an
arbitrary number of local productions. It can be implemented based on DRL.
In a production system there is a global area where information, including
relationships is represented. Production rules are simple declarative rule that
can be of the DRL sort. A production system is used much less "intensionally"
than a DRL system. That is, you initialise the information, insert the rules,
start the system going, and then monitor the global area to see what is
happening in the midst of all these unfolding rules.  The propositions search
the global area to see if their goals apply, and if so post the consequence goals
to the global area, thus creating the possibility that other propositions can
"fire". One of the interesting things about a production system is that the rules
can be of different orders and levels.

For example:

If we want to represent pearl farming in Manihiki with a production system
we might do the following.

Manihiki is a small atoll 3° south of the equator, more or less south of Hawaii.
Recently cultivated black pearls have been introduced to the lagoon, which is
well suited for this purpose. As such this is a new industry, so there is no
indigenous system to replace, though it obviously replaces other indigenous
uses of resources, in particular copra making, which is currently lapsed. To
represent the overall process of pearl cultivation from a social and natural
resource point of view we have to examine a number of different orders.

To indicate a few of these:

1) social organisation
2) population flows
3) atoll politics
4) historical linkages to other atolls
5) resource explotation skills
6) knowledge resources and their distribution
7) new skills relating to pearl cultivation and their distribution
8) external technicans with "secret" knowledge
9) markets, local and external
10) national politics
11) the lagoon and lagoon ecology
12) the atoll and atoll ecology
13) transport
14) supplies


 

  Michael Fischer     9

None of these can be adequtely discussed without reference to at least one or
two others. The resulting network of dependencies of these systems tends to
relate them all, in good old fashioned holistic style for anthropology.

In the production system we can have propositions (goals and consequences of
the goals) of the following sorts:

a) propositions whose condition goals and conseqence goals are of the same
order, eg operate within the same domain.
b) propositions whose condition goals and conseqence goals are of different
orders, eg they translate results between domains.

It is not necessary to have any type b) propositions to have a useful
representation, though in practice there are many. For example:

pay surcharge if finance is inadequate or surcharge < 1 for 2 shells

check shell health immediately if did not pay surcharge

seek advise from extension if shell health poor

seek advise from lawyer if shell health poor and did not pay surcharge

In this way representations can be created where some productions are in
terms of formulae, eg the volumn of a particular area of the lagoon which
partially determines carrying capacity, or a set of rules for maintaining
oysters, together which might activate propositions about scheduling
maintenance, planning trips to Rarotonga for supplies (and fun), or deciding to
get married.

In this way our model includes both reductive and non-reductive aspects, eg
the entire unfolding is available for view, or just the states of information at a
given time. Other rules can generate gossip, fights, bring up images of people
places and things, select video footage, or effectively activate any kind of
representation at all. Any, all, or no viewpoints need be priviliged except in
specific individual goals (eg all goals assume some kind of privilige, but only
for that goal).

Production system offer real prospects for a method of representation of both
traditional NS and SS conceptions of a problem, as well as incorporating new
concepetions.

For further information see some of the other papers at this site.
http://lucy.ukc.ac.uk/Simulate/
also:
Chip Morningstar on "How to deconstruct almost anything: My postmodern Adventure"
Derrida on facts, truth and justice
Derrida on deconstruction and differance in a letter to his Japanese translater
Lewontin on scientific truth


   Contents

Return to the CSAC Ethnographics Gallery

Contact Michael Fischer