* Many of the other major CS/EE research universities are in cow towns like Ann Arbor, Chambana, and Madison.
* Many of the other non-Valley tech centers happen to be in places, like Atlanta, Seattle, and Austin, where a research university coincides with a thriving metro area.
* The few places that don't have startup scenes proportional to their population size and University caliber are, like NYC and LA, prohibitively expensive.
I like the work you're doing. I'm just pointing out that there's a huge biasing effect with ISS. It's the first successful IT security startup (the Netscape of its field), which sets you up for post-hoc-ergo-propter-hoc. Even its GATech roots are easy to overstate (much of its original talent came from Arizona and California).
ISS is just a starting point. As a writer, its easier to approach a complicated topic with a hook - ISS is the hook. But I don't think its coincidence that ISS was the netscape of its field, and the fact that we have a sustaining cluster in security.
I agree that bringing in talent from afar happened - that seems to have been key to many of the offspring.
(If you can't tell, I'm somewhat closer to the history of ISS than most HN readers, and not just because I'm in security).
I think that all adds up to a good fit. I think Atlanta wanted to birth a security cluster. I think there were vectors in place, pushing Atlanta the whole way.
Noonan is also definitely a product of Atlanta.
I haven't said so - but I take it for granted that Georgia Tech is behind all of this. It is widely acknowledged that a solid research university is a prerequisite for these clusters to form.
A founder map is just one way to look at this data. If you look at all employment overlap, funding, etc. the web gets very dense, and you see many more connections between these companies. The executive overlap in particular is very strong. When I started this, Linkedin's API was not open, and so I did this one manually because it is at a scale I could manually aggregate. Now that its open, other kinds of maps are possible.
This is just an introduction, I'm going to be digging in deeper in later articles.