WARNING: Long post, 230 line, 1800 words
Dear Terry,
Defining designing was not in my head as I responded to
Keith, but your point is a good one, and it has helped me
see something I did not have so clear before: designing can
not be properly understood (known) simply in terms of the
activities and actions of the designers doing it.
Despite that, as Klaus rightly put it, "design[ing] resists
being theorized" [Mon, 16 Oct 2000 12:18:06: Subject: Re:
Bicycle Knowledge (coffee topic?)] (or perhaps because of
it) I do, like you, have an interest in trying to develop
a theory of designing. I think our ideas of what any
theory might look like are, however, quite different.
Which, for me, makes it more interesting and useful. I
also do not believe that there can be or will be just one
(grand unified) theory of designing.
Where you seek to identify and thus define designing in
terms of 'micro-elemental acts', I seek an abstraction
away form the designing activities and acts, to a level
which has its own characteristics and properties not seen
in any of the activities or acts of designing, either
individually or collectively.
My starting point is that designing is a single and
distinctive kind of phenomenon. In other words, talk of
dress designing, fuel injection pump designing, investment
portfolio designing, font designing, and other things we
don't conventionally or usually call designing, are all
examples of the same phenomenon; a phenomenon distinct
from other kinds of things people do. (I realise that
this involves a strong assumption, and one many people are
not happy with. For them, dress designing and electronic
circuit designing, for example, are quite different
things.)
Secondly, I take designing, in its many guises and
disguises to be ubiquitous. As Sid Newton put it, "Design
may be something we all do as part of being human, ..."
[Thu, 12 Oct 2000 11:06:39: Subject: Re: Bicycle Knowledge
(coffee topic?)]. In other words, when I talk of
designing I don't just mean professional designing.' I
want to mean all designing.
Designing usually involves using tools, aids, support
systems and other external resources. (I accept the
possibility that designing can be a purely mental
activity, making no use of anything external and resulting
in nothing external, but I prefer to treat this as a
limiting (and limited) case which is best left until we
understand designing better.) So, what we see when we
watch someone or some people designing is a lot of
different activities and actions which depend on the
particular circumstances of the designing, the tools,
aids, systems etc used, and other details of the time,
place, situation, and context.
There is an enormous amount of natural and accepted
variation in the details and organisation of all this
activity. Even when we watch two designers each designing
the same kind of thing, we can see a lot of differences in
what they do, how, when etc. Indeed, it would be very
hard, and very unnatural, to try to make even a small
amount of this variation go away---by trying to force
designers to do things in particular ways or in particular
orders.
All this normal variation in the acts and activities is
not peculiar to designing. We see the same kind of thing
in all other human activities: in cooking, dancing,
presenting talks, writing books, making films, to name
just a very few. Now, I do not believe any theory of
designing is ever going to explain or account for all this
kind of contingent variation. And I don't think we should
be trying to form theories of designing which would seem to
have to. By which I mean cognitive or psychological
theories of designing. The biggest reason for why not,
and why I think designing seems to resist theorising, at
least at this level, is that none of the activities or
acts that we can see in any designing are unique to
designing: we can see the same activities and acts in
other kinds of human behaviour. Nor do I see much hope in
an appeal to the spatio-temporal organisation of the
activities and actions, as a basis for uniquely
distinguishing and defining designing, since this
organisation too is subject to a lot of contingent
variation.
We must also take care not to fall into the trap of
thinking, because we see some particular activity or
combination of activities every time we watch some
designing, that this activity or combination of activities
are essential or necessary for designing. We cannot
observe enough designing to be sufficiently sure of this,
and we would only need one counter example to finish off
such a proto-theory. The only way to be sure would be to
have a theory that says it is a necessary activity. But
as I say, I don't think we can have such a theory. What
designers do and how they do it is often very easily
changed, sometimes quite radically, by the introduction of
a new tool, aid, or support system.
All this natural, and inevitable, contingent variation in
the activities and actions of designing is one reason why
it is so difficult to understand what understanding
(knowing) of designing we get from watching people doing
designing.
The way out of this mess, to some kind of stable
theoretical understanding of designing, is, I think, to
abstract away from all the activities and actions that
constitute the designing when it is done. My particular
attempt at doing this (but I don't wish to suggest that
this is the only or the best approach) is to use a concept
of knowledge we have from Allen Newell and Knowledge
Engineering methods: to develop what I call a Knowledge
Level theory of designing.
Newell's concept of knowledge is a competence notion:
knowledge is a capacity or a potential for rational action.
This is quite different from the more Classical concept of
knowledge we have from Epistemology, where knowledge is
conceived of (indeed, defined) as justifiable true belief.
It is also different from the Logisist approach within AI,
which takes knowledge to be (logically) verifiable truth.
This concept of knowledge, as a capacity to act rationally,
like the Classical concept, needs an agent---to do the
rational acting. (In the Classical case, the agent is
needed to hold the belief that is true and which it can
justify, i.e., has evidence for.) But the abstractive
power and practicality of Newell's concept of knowledge
lies in the fact that an agent can be said to know
something because it is observed to do something, but not
necessarily observed to do it is some particular way. So,
two agents which are observed to be capable of the same
thing, but in quite different ways, can be said to know
the same thing.
Newell originally presented the Knowledge Level as a
natural level of abstraction placed immediately above the
Symbol Level---the level at which we describe and define
the necessary and sufficient mechanism for intelligent
action, in machines and humans (for Newell and Simon and
other symbol processing believers).
Modern Knowledge Engineering approaches have taken Newell's
original proposal of the Knowledge Level and done four
things with it.
1) They adopted the concept of knowledge it is built upon:
knowledge as a capacity or potential for rational
action.
2) They detached it from the Symbol Level and gave it a
separate and independent existence as an abstraction
level over the behaviour of agents and agencies.
3) They dropped all cognitive connotations, thus removing
any claim for or specification of how the capacity for
rational action is implemented.
4) They introduced structure. For Newell, the knowledge
at the Knowledge Level was simply one amorphous
capacity. In Knowledge Engineering we have different
kinds of knowledge which can play different roles and have
different relations to other kinds of knowledge.
In developing my Knowledge Level theory of designing I
have taken these Knowledge Engineering developments and
tried to define kinds of knowledge, roles and relations
that go together to define what designing is in terms of
the knowledge (necessarily and sufficiently) used and
generated in any designing.
It is, therefore, a very abstract theory. It only says
what kinds of knowledge are used and generated. It does
not say, and cannot say, what knowledge is used and
generated. Nor can it be used in a normative way, not
directly at least: it cannot be used to say how designing
should be done. What it is supposed to do is to offer
theoretical support to the development of Knowledge Level
models of particular kinds of designing, which, in turn,
can be used to design and develop effective
(knowledge-based) design support systems. It can also (I
speculate) be used to support aspects of the Knowledge
Management of designing.
As I said at the beginning. This is only one type of
theory of designing that I can see could be developed, and
it may be not even the most useful. It is, however, very
important not to mistake this for a theory like we see in
science. The concept of knowledge that it is based upon
is not an independent or absolute (Platonistic-like)
concept, as we are supposed to have in science. Knowledge
can only exist if we have agents that can act rationally:
it needs agents to carry it. But agents, and in the
context of designing we are here talking about people, not
(yet, at least) any kind of artificial agent, are the only
things that can have the knowledge sufficient to have and
use this concept of knowledge. This means that however
objective all this might look, and no matter how objective
we might tends to see and take all this, it is not an
objective thing that we are using here. It is an
abstraction, hopefully a useful one, that is forever tide
into the subject matter we are abstracting over. This
concept of knowledge is thus a kind of second-order thing,
in the sense of second-order cybernetics (the cybernetics
of cybernetics).
So, Terry, as I also mentioned at the beginning, we share
an interest in theorising designing, but have quite
different ways of trying to do this. I have set out my
approach at some length, though not in much detail, so
that you and others might see the sense and/or nonsense in
it, and comment on this. I'd simply add that it is always
good to have the encouragement that comes with other's
identification of the sense they see in this, and not just
the discouragement that come with the identification of
the nonsense they see. I also think that having both helps
others, particularly the PhD students amongst us, to
assess and evaluate all this.
Best regards,
Tim
CEIT, Donostia / San Sebastián
P.S. A meta question, to you Terry and to others: should
we try to move this thread to the DRS list, or just tell
them that its here, or neither?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|