Romulo Krafta said:
> What space syntax does is neither one nor other thing, in fact SS
> actually model anything, yet modeling is about change and, of course,
> In this way, correlations between accessibility and flows are
> definition, yet change and time are not considered, let alone trip
> which are at the heart of urban flows. This is made quite clear in
> evolution studies that frequently display the disarray between ss
> and distribution of flows in different moments in time of a same city.
I agree with Romulo about this. One of the aspects of this discussion
depends on people's use of the word 'model' and assumptions about what
it is that SS does. We do not 'model' in the conventional engineering
sense of trying to build a 'model' of a system that behaves as that
system would under some given set of conditions (and possibly changes).
SS really started out with the Social Logic of Space as a theory of
society and its active relationship to the spatial environment - a
theory of society I which space figures. Remember that in many social
theories space is either not considered at all, or relegated to being
either a context for social action (and so passive) or a product of
social processes (and so also passive). SLS proposed that space is
active in its relations to social production and reproduction, and went
into detail about the ways that this seems to happen. In order to
investigate this various methods for representing spatial environments
were developed - the axial map is one. And various measures of the
networks that could be derived from these representations were developed
- Relative Asymmetry (or integration) was one. It turned out (actually
subsequent to the SLS going to press) that pedestrian flows in urban
space correlated with axial integration - I think this made it to a
paragraph in the introduction (p23) at galley proof stage.
Of course this finding was of fundamental importance to the social
theory proposed in SLS. It gave a mechanism by which spatial morphology
could in principle affect the probabilistic likelihood that people would
meet, communicate, transact and generate or reproduce social forms. What
the methodology was NOT about was trying to create a mechanistic 'model'
of movement flows or any other aspect of the function of the city.
This then is how SS should be thought of - an explanatory theory of
society with an at least plausible mechanism embedded in it. The status
of the axial map (and all the other descriptive maps we have tried,
including maps that are very similar to the traffic modeller's node and
link planar graph) is that they are simplified descriptions of the real
world configuration of open space. Beyond that it seems to me that 'the
proof of the pudding is in the eating' - the value of one description
over another lies not in the convention that 'this is the description
everyone has used before', but in whether or not that representation
discovers any regularities in the real world phenomena that are of more
than passing interest. It turns out that the axial map does so, not just
in its correlations with movement flows, but also in the structural
patterns that it highlights in a diverse range of urban forms. That is
itself a matter for theorizing about. In my view it is linearity itself,
or the minimisation of angular deviation in human route choices that may
account for this.
One further point. Of course SS methods and SLS theories are used in a
'predictive' mode all the time. Surely this is 'modeling'? Well, no, I
don't think it is in the same sense. The reasons people use these
methods are 1. that they are engaged in designing new or modifying old
morphology. That is, the variable they are changing is the design of the
spatial layout. And 2. they are interested in the social and economic
consequences of these changes over a relatively long term future. For
this they need a theory about the relation between the spatial
environment and social and economic function. They also need an analytic
methodology that is theory based and capable of working where the
morphology changes radically or is entirely new. Traffic models tend to
be used where the physical morphology stays pretty much the same, but
where traffic management or policy changes. Here a relatively mechanical
model with many different variable 'knobs' to tweak is a great asset.
The traffic manager has many different tools at their disposal, and by
and large, changing the configuration of the network radically is not
really one of them.
This is why in a previous email I said that SS and traffic modeling are
almost entirely complementary. We work alongside traffic engineers on
almost every urban project, and both sets of tools are used for their
own specific purposes. The clients for these projects don't have two
sets of people (and two sets of fees) for the sake of it. They do it
because the tools and theories answer different questions and they need
the answer to both of them.