Dear Bill

I think people like me get misled by the idea that you are measuring
the 'geometry' of a place.  Axial lines, convex spaces and isovists
measure the geometry of co-visibility.  The geometry of streets and
buildings has only a limiting effect on this, it is not the relevant
geometry.  If an axial line passes along the middle of a street it doesn't
matter what the facades are like.

To go back to your succinct definition of space syntax as "Extrinsic
measures of relations between geometrical elements", wouldn't it be
more exact to say that a space syntax tool is 'Any method that makes
extrinsic measurements of co-visibility and/or co-accessibility.'

The only physical facts we need to know about a place are the limits it
sets on co-visibility or co-accessibility.  Co-visibility and
co-accessibility  are the two independent variables on which we
suppose certain social phenomena to be dependant, and you have
found the link to be in the pattern of configuration as shown by the
j-graph.

I don't see why it is necessary to talk in such an abstract way about
"Spatial elements which, prima facie, seem to have some degree of
embedding in or relation to human behaviour" when you are actually
working with concrete facts about whether a person in one location
can see a person in another location.   This is an inherently social fact,
with an overwhelmingly strong relation to human behaviour.  It
sometimes seems as if you don't see what a powerful case you have.

Of course other investigations of configuration are interesting as well.
Sanjay's suggestion of another name for other tools might help define
another distinct field of study.  In fact, the significant difference seems
to be that space syntax deals with the potential spatial relations
between people within a space, whereas other methods measure
spatial relations within shapes or between people and shapes.
Wouldn't you call this Shape Syntax?

And then you open up a possible whole new world of purely Social
Syntax . . . !

Regards,  Tom

On Wed, 28 Jan 2004 13:31:49 +0000, Professor Bill Hillier
<[log in to unmask]> wrote:

>Dear Jake - I agree with both Alasdair and Tom (and also with Alan's
latest
>whch I ave just seen). Space Syntax is a research programme with a
>particular theoretical approach, not just a set of tools. But there is a
>link. The theoretical approach is based on two key ideas which are
>reflected in - in fact the very basis of -  the tools. One idea is that of
>representing space in terms of spatial elements which, prima facie,
seemm
>to have some degree of embedding in or relation to human
behaviour e.g
>movement is linear both itself in a local sense, but also in the sense
that
>in navigating, say, an street network, we try to approximate a line
between
>where we think the destination is in relation to the origin - so
movement
>from one to the other is likely to reflect how we conceptualise the
pattern
>of lines available between the two. I discussed this a way of knowing
urban
>systems in my second paper to the fourth symposium, and it seems
likely
>that non-linear information is not involved in this picture. Likewise
human
>interaction in real space is normally convex, so again behaviour has
a
>geometrical implication. Again, our visual experience at any moment
is
>something like a directed isovist. So in all these senses spatial
elements
>in space syntax embody some degree of functional potentiality. One
task of
>the researcher is to decide what representation is most likely to
capture
>the logic, spatial and functional, of the system being investigated.
This
>is not to ignore geometry, as some of our critics argue, but, on the
>contrary, to try to embed the natural geometry of human behaviour in
formal
>spatial analysis. It may even me that one reason space syntax
seems to work
>is that the 'natural geometry' of human behaviour is, as one would
expect,
>already built into the systems of space we are studying, both at the
level
>of the space and the level of the configuration.
>
>The second idea is to use graphs to assign values to spatial
element
>(defined as above) which reflect their relations to some or all other
>elements in the system. This idea is embodied in the j-graph, which
seems
>to me central to both the theory and methods of space syntax.
>Conceptualising elements in a system in terms of the shape of their
>relations to all or some others is perhaps the paradigm element in
space
>syntax. We can see a system, for example, not just as 'elements and
>relations', but as its set of j-graphs, so that the system is in effect
>made up of the points from which the whole system can be viewed.
This
>immediately shows, for example, that systems of space have a
different
>shape when considered from different points within them, even
though they
>are the same system. I haves suggested that it might be useful to
see other
>kinds of system in this way, such a social systems (see for example
my
>'Society seen through the prism of space' at the Atlanta Symposium).
>
>It may be worth adding that in most space syntax studies the other
elements
>are of the same kind, but not necessarily so. For example, as you will
>remember, in our early studies of the Tate, we  tried different ways of
>representing space, but found that by far the most successful in
predicting
>movement was using convex elements linked by lines - which of
course is
>what the Tate feels like to a visitor. But whatever spatial elements we
>use, syntax focuses, as Alasdair says, on the extrinsic rather than
>intrinsic attributes of spaces, that is on their external embedding in
the
>system as a whole rather than their intrinsic properties - though you
can,
>by using points as your elements, apply extrinsic measures to
'intrinsic'
>properties of spatial elements such as shape (see for example the
same
>paper referred above).
>
>These are the defining characteristics of all of the techniques of
spatial
>analysis that have been developed by space syntax and and I am not
aware of
>other methods of spatial analysis which set out deliberately to
combine
>these two ideas. In doing so, space syntax is in a sense trying to
create
>methdods of analysis which, while being formal and rigorous,
embody the
>human subject in a sense that even some phenomenologists
approve of (for
>example Seaman). - Bill
>
>