Human Services, Crisp Programs, Fuzzy Need

Ekadanta (ekadanta@melbourne.dialix.oz.au)
Fri, 26 Jan 1996 13:14:32 +0100


A little challenge! Below is a comment I recently sent to the organisers of
the forthcoming HUSITA4 Conference in Finland. This is a Computers in Human
Services Conference. I had asked whether they were expecting any fuzzy
related papers. Guess what? They are not!

I have asked in this newsgroup before if there is anyone working on the
whole area of case management, case work, and monitoring mechanisms, from a
fuzzy perspective. A number of people, including Professor Bart Kosko,
responded in an attempt to make useful suggestions. But, as yet, I have
found no examples.

This is now starting to puzzle me. To the extent that fuzzy is about
systems, controllers that cope with complexity and vagueness and so on, it
would seem tailor-made to work on HUMAN SYSTEMS. Human services workers are
involved with extraordinarily complex and vague factors in responding to
the vast array of human need. And this is not just liberal humanitarianism.
It has huge costs to provide, not to provide, and to get it wrong when
providing it. Under economic rationalism (which I'm not going to debate for
or against), huge systems in public health and welfare, have been subjected
to measurement regimes intended to monitor and manage efficiency and
effectiveness. Output Based Funding approaches ostensibly reward providers
for effort, and punish for failure. ALL of these systems are built on crisp
taxonomies, ameliorated by probability weightings (in the most
sophisticated, such as the US and Australian casemix public hospital
funding systems). Yet these mechanistic structures are based on modified
certainty when the reality they overlay is fundamentally uncertain, and
marked by highly imperfect knowledge.

In the note below, I say that it is 'unfortunate' that fuzzy seems to be
dominated by engineering and linguistic interests. The 'misfortune' in the
former case is not intended as a slur on engineers (I'm conscious of my
audience here!). It's simply a notation on the apparent composition of the
fuzzy interests population. And, engineers tend naturally to me more
interested in pursuing 'engineering' subjects, and designing machines or
physical processing systems. But, hey, theres a whole world of PEOPLE out
here, who are not just users and potential users of new machines or
commodities, but who themselves are the SUBJECTS of systems which fuzzy
logic could improve. Indeed, fuzzy logic and it's cousins, could provide
for interaction in such systems, changing the whole dynamic that currently
exists between 'practitioner' and 'client'. So heres the message I sent to
HUSITA4. I describe it as a challenge, so I'm putting it up here as a
challenge too, directly to the fuzzy community:

------------------------------------------------

I had referred to fuzzy logic. Unfortunately I am not a programmer or even
computer expert, just a reasonably well-informed human services computer
user/consumer. I am very frustrated by the inadequacies of current case
management SOFTWARE, and output and outcome measurement tools and models.
It seems to me that, if it is at all true that the 'crisp set' paradigm of
traditional computing is inadequate, then it is particularly inadequate in
the human services. Human beings are not categorisable through crisp set
models. Human beings are vague. Needs and desires, as well as responses to
phenomena such as case interventions, are not precise or predictably
certain. This is true for any one person, or any practitioner:client
relationship. How much more true is it when the person or client is
recognised as existing in a dynamic context of 'real world'. Then the
imprecise needs, the uncertain 'therapeutic actions', are also subjected to
forces over which neither practitioner or client has a great deal of
control. These may be immediate, such as the influence of other people on
both the practitioner and the client. Or they may be super-structural, such
as changes in the housing market, the economy more broadly, or the
political environment.

Human service software and models which treat 'cases' as simple, contained,
and certain, fail to allow for this uncertainty, complexity, and lack of
control. This poses fundamental problems for practitioners forced into
simplistic accountability regimes by statisticians and programmers. It
poses insurmountable ethical and integrity problems for evaluation
methodology in terms of the 'attribution of program effect'. It establishes
or reinforces practice paradigms and modes of delivery which only account
for a narrow middle range of clients, a lowest common denominator. This in
turn leads to 'mainstream' services which exclude the more complex cases,
the ethnically diverse clients, or those who simply are not comfortable
with the particular method or style of delivery. In other words, it
produces a crisp practice paradigm, which itself creates marginalisation
and exclusion. In this sense, it is people who fall foul of Aristotles 'law
of the excluded middle'!

The work of Lotfi Zadeh, Bart Kosko, and others, on fuzzy logic, adaptive
fuzzy systems, and fuzzy cognitive maps, seems to me to offer a way IN. It
is the way that the fuzzy paradigm values vagueness and uncertainty, along
with balancing complex arrays of competing forces. It allows for grey, and
in this, more nearly replicates actual human thought structures and
behaviour. Unfortunately, most of the work going on in this area is either
in engineering or linguistics. And there are very few people it seems
(no-one?) in the human services arena that is even questioning the
FUNDAMENTAL assumptions of a crisp set based taxonomy and programming
environment.

This is a challenge to programmers, statistical experts and so on. You
never know, sometimes such challenges may be taken up by just one person,
and yet a major breakthrough may occur!
-----------------------------------------------
Thanks, Quentin
(Anybody with any bright ideas, leads etc. can e-mail me direct:
ekadanta@connexus.apana.org.au)