Dear all,
Using the content of Jakes' last mail, I like to bring a few elements into
the discussion about the axial line:
> Wouldn't it be more interesting to present some evidence on issues of
> reliability with axial maps instead of debating it in the abstract? There
> is obviously ~some~ error margin and a question about reproducibility when
> using axial maps, so how significant are these issues? Sheep asserts that
> hand-drawn axial maps are robust. Maybe they are, ...
... I think they are, not because of the reproducibility of the lines, but
because of the number of nodes between two places (that is of course the guide
line for producing fewest line maps). Even when different and even quiet
un-trained people draw maps, depth in their proposals is very comparable, not
the precise places of the lines. The more connectivity is high, the more lines
are precisely located. Maybe "what is a line?" can also be formulated as "what
is a node?". A node has no extension, it is sharper then a line when I try to
experience it in the real world. To be in the node is to be in two realities
(two lines, two horizons, two isovists). If the line is the step, the node is
the step stone. A line is ambigue; in a complex structure, a line belongs to
more than one path. A line is more like: "at least one line". A node is less
ambigue. If I try to SEE one line, I don't know from where I have to look. If
I try to locate a node, and I do that in respect to an particular intention (a
path, a destination), it is very comparable to what other people find. My
axial map is then all the necessary nodes in respect to all the possible
paths. The ambiguities which stays than in the map are merely the dead ends of
the lines, and also lines with connectivity=1.
> but without any evidence
> its just guesswork. What is the error margin involved in a sample of
> 'trained' people drawing the same axial map to any particular
> specification? What is the control procedure for differences in drawing
> style that makes maps comparable and reproducible? How much error can you
> ignore and how much becomes a problem? Until someone publishes this kind of
> study, Tom Dine's original question about arbitrariness will remain
> unanswered.
>
> One of the reasons we chose to develop Visibility Graph Analysis software
> for our work at Intelligent Space is the issue of reliability with
> hand-drawn representations of spatial structure such as the axial map. The
> automated spatial sampling technique provided by VGA resolves this issue
> and makes the whole analysis methodology a lot more open to scrutiny.
Can you explain that, Jake? I don't see well how VGA can deal with topological
distance, expressed by the axial map. There is of course a relation, since a
line is a sort of reduced isovist. Whatever, I think that tomorrow Tom Dine
can ask: What is an isovist?". What is it in the isovist that is important for
the view- or standpoint of the isovist. Is it the extension, the depth, the
symmetry, what I see on the edge, or what the edge tells me about where I
probably am, and where I can probably go, the degree of convexity, or
concavity, closeness, openness, etc. ?
> Whichever techniques are used, I think the aim for all of us involved in
> socio-spatial research should be to develop and use methodologies that are
> non-controversial. Don't we want people to be able to see past the
> methodology so we can get back to talking about what matters- how space may
> or may not influence people's lives? The axial map has been a useful tool
> for investigating spatial questions about society. Visibility Graph
> Analysis offers a step further in the right direction for the development
> of open methodology in this field.
>
> Jake
>
> _____________________________
>
> Dr. Jake Desyllas
> Partner
> Intelligent Space
> 68 Great Eastern Street
> London EC2A 3JT
> t: +44 (0) 20 7739 9729
> f: +44 (0) 20 7739 9547
> e: [log in to unmask]
> w: http://www.intelligentspace.com
Guido
Guido STEGEN
ARSIS archit. & town planning / Space Syntax Brussels
rue Edm. de Grimberghe 47
1080 Brussels
t: +32 (0)2 410 84 39
f: +32 (0)2 410 34 31
e: [log in to unmask]
|