Dear Sven,

Sorry for the delay in my response.

True, there seems to be a problem in the formula. Instead of calculating
NC radius 2/ Connectivity, it is calculating connectivity/ NC. I don't
know whether this was intentional. While waiting for the correct version
you may try these two steps;

- do a VGA analysis for both local and global measures on radius n and
radius 2.
- Create a new column for controllability and update it using the
following formula;
value("Visual Node Count R2")/value("Connectivity")

You will then get a more convincing controllability measure following
Alasdair's definition in Depthmap manual.

In the article I have attached in my previous email, Philip Steadman has
presented some controversies with regards to the design of the panopticon
and the notion of surveillance as it was materialised through that design.
Alasdair has also addressed a caveat regarding the VGA measures and
suggested that one might consider the central space as both controlled and
controllable. It all depends on whether you consider the occluding barrier
surrounding the central space and the multilevel complexity. However,
Alasdair suggested in the manual that these measures might be helpful in
finding spaces that are more vulnerable to crime. And indeed,
connectivity, the common variable between control and controllability was
found to correlate to crime in some studies (Hillier and Sahbaz, 2005),
(Nubani and Wineman, 2005),(Jones & Fanek, 1997). The latter study
considered axial control. Yet, it depends on the context, the type of
crime and other dependant variables. Landuses seem to play part in
defining different associations between high/low connectivity and crime
locations. Taking that into account, it would make sense if there was a
relationship between control or controllability and crime.

All the studies I have mentioned looked at axial/segment representation.
Students have occasionally used Control and controllability in their
research, but I don't know if there is any published work on that. There
might be studies that looked at VGA analysis that I am not aware of. I
will leave it to other researchers on the list to advise you on that.



Kinda


> Dear Kinda,
>
> thanks for the interesting article.
>
> The problem of calculating controllability in depthmap still exists: the
> formula, as alasdair wrote in the tutorial, is to divide what you see on a
> second view (number of nodes until step depth 2) by what you see at the
> first view (connectivity). In the cell of a panopticon the first view
> (connectivity) is low and the second view (nodes at stepdepth 2) is high:
> means controllability must be high (high/low). This formula makes sense
> from my point of view, but the opposite seems to be the case when
> calculating controllability in depthmap: the center has value 1 and the
> cells tend to 0. the formula seems to be turned around: first view/second
> view.
>
> by the way: is there any serious research, using the
> control/controllability values, or are they just theoretical constructs?
>
> best
> sven
>