Hi – if you came here looking for the draft of ‘cities and centralities’, I’ve taken it down to rework it some more. I’m not at all happy with it, so I’m going to start over. Of course, I’m sure you could probably retrieve a copy of it from Google’s cache or other similar place, but I’d appreciate if you didn’t. (See also Alun’s comments on permanence.)
Electric Archaeology 
The Pompeii study sounds like Eric Poehler’s work. He’s at UMass at the moment. There’s a version of “Romans on the Right” at Athanor http://www.fsu.edu/~arh/images/athanor/athxxi/AthanorXXI_poehler.pdf and more versions listed at http://www.pompeiana.org/Research/Streets_Research/Streets_Research.htm I don’t know if that’s helpful.
How sensitive is the methodology to unknown streets? I remember Hillier saying that in his method a street or two could make major changes. This means that the grid has to be very well known in his system. If your method is more robust, then that’s a big advantage for sites like Falerii Nova.
Thanks for the links!
Your question re sensitivity… that’s a good question! If we think of these networks in terms of their degree distributions – how many intersections have two streets, how many have three, how many have four, etc, then we’re going to get something resembling a power law – for every intersection (public space, market, forum, crossroad, whatever) with, for instance, 10 streets, we’ll get 10 intersections with only 2 streets, as it were. So I should think that even if a plan derived from geophys or excavation or whatever misses a few streets, they won’t make too much difference: unless we miss a major intersecting space.
Hmm.
Even if it did miss a major space, I think the betweeness calculation would still throw up places worth checking out. I guess the way to sort this out would be to run the analysis again, but block out portions of the street grid to see what emerges.
So I think the method is reasonably robust – a few streets here and there are not going to make much difference… of course, I don’t think you would want to use it anyway unless you had the majority of the grid in the first place?
On a related note, I was trying a slightly different approach here: http://home.cc.umanitoba.ca/~grahams/pompeii.html Using an ABM, I pour artificial romans onto the street scape, and measure how long it takes for information to diffuse throughout the city, to every artificial Roman. Modifying it somewhat, you can make the wee Romans leave traces on the city map (or you can ask the map itself to record how many Romans passed that way) to see which areas naturally ‘funnel’ traffic through them.
BTW, my ABMS can be found at http://graeworks.net