Thursday, September 27, 2012

There Will Be Graphs: Part Five of a Series

By Evan D. Robertson. 

I’ve long believed that, at the end of the day, innovation occurs in place. Whether it’s a bar, R&D lab, college classroom, or tech incubator the ideation, the exchange, and the pursuit of an innovative idea must become existent at a specific place. However, one cannot escape the fact that the speed and ease at which we transmit ideas across a web of interconnecting fiber optic networks (endearingly referred to as “tubes” by some) has allowed individuals to exchange complex information with relative ease, potentially untying innovation from place.

For today’s blog, I sought to explore whether innovation remains attached to place, and what better business sector to explore this idea than telecommunications. After all, since this sector is ground zero for email, text, telephonic, and other electronic communications, shouldn’t they also be the early adopters of innovation by electronic communication?

What better region to give us insight into this question than Georgia’s Metro Atlanta region where we are currently working on a holistic, regional economic development strategy. As indicated during numerous public meetings, stakeholders frequently pointed to the region’s strength in the telecom sector and its strong fiber optic infrastructure as major assets to the community. However, Metro Atlanta’s strength in telecommunications transcends infrastructure, its telecom inventor network is second to none.

Metro Atlanta’s telecommunications inventors have produced 431 telephonic communication patents since 1979, more than any other metro area in the United States. Patent titles range from thrilling topics such as “call screening method and apparatus” to “methods, systems, and computer program products for rule-based direction of customer service calls.” As you may have guessed, these patents almost exclusively deal with the telephone. While this technology maybe on its way out, the question remains, are these telecommunications innovations researched and created in a place? Or, does the telephonic communications sector utilize a much diffuse network of inventors from across the nation, and even the world?

As with most of the initial social network analysis graphs, the first output is a little chaotic. The graph shows the connections (edges) between individual inventors of telephonic patents in Atlanta. The edges represent co-invention, inventors were said to be connected if they produced a patent with another inventor. The general order of the graph is that those inventors closer to the middle of the graph tend to be more “connected” than those towards the edges of the graph. Another way to view the initial graph is to calculate (electronically of course!) metrics that determine the overall structure and connectedness of the network.

As stated in last week’s blog, we can also identify the overall density of the network, that is, how connected inventors in Atlanta’s telephonic communication’s field are to other inventors in the network. The overall density of Atlanta’s telephonic inventors is fairly low (.0086), where 1.00 would equate to a world where every inventor connected to every other, and 0.00 is a network where none of the inventors are connected (i.e. not a social network).

The overall density is extremely low because, inventors of Atlanta’s telephonic communications patents are almost solely connected by where they are employed. That is, there are few connections in which inventors are connected across firm boundaries. An inventor working at AT&T tends not to co-invent patents with individuals from IBM (or, Sprint Communications). Those connections which transcend firm boundaries usually entail either individuals who spun off from the parent firm, or, in the case of Atlanta, was a byproduct of a major corporate merger. Thus, the most numerous inter-firm connections occur between AT&T and BellSouth, the latter of which was acquired by AT&T (formerly SBC Communications) in 2006.

The above graphic groups the individual inventors by their home firm. Since the data transcends a large period of time, home firm is defined as the last firm in which the inventor produced a patent. Two firms, AT&T and BellSouth, accounted for a large portion of the social network. Combined, these two firms produced 305 of the 431 telephonic communications patents between 1979 and August 2012. Or, roughly 71 percent of all patents of this class in the Atlanta metro area. Given the fact that patent production is dominated by two large firms, it should be no surprise that inventors in these firms are highly connected to one another and the broader social network.

The preceding graph filters out inventors with a low betweenness centrality (also described in last week’s blog) leaving those individuals who are highly connected within the social network. Betweenness centrality gives a general impression of how connected an inventor is within the context of the larger network. Filtering for a high betweenness centrality leaves only those individuals who are extremely connected, and serve as an important role of bridging other inventor groups within the social network. The results left only a few individuals from three different firms, BellSouth, AT&T, and Dono Technologies (a company with strong ties to AT&T).

Interesting also, is that AT&T has more individuals serving a network role. That is, there are more inventors acting as bridges between other inventor groups in AT&T than in its previous company BellSouth. This may indicate a shift in the way AT&T manages its inventor groups, pointing to a more fluid innovation structure. But, without actual interviews with AT&T inventors and managers, it’s hard to say whether this is actually the case.

Finally, only a handful of the 550 inventors that produced a telephonic communication patent in Atlanta were from outside the Metro Atlanta area. Those inventors that were from outside the Metro Atlanta area tended to be from non-local firms such as Sprint Communications (with connections to Overland Park, Kansas), and Research In Motion (via California and Toronto, Canada). Inventors who produced patents for AT&T and BellSouth, by and large, were from the Atlanta metro area. Some evidence, at least in Atlanta’s telephonic communications sector, that innovation has remained a regional affair.

To conclude this lengthy blog (bordering on treatise), social network analysis, while potentially time consuming, can give great insight into your local regional innovation system as well as how this innovation system has evolved over time. It can isolate those connections that are critical to a fully functional innovation system along with identifying external ties from outside the region that firms are reliant upon to produce a patent. If your region relies heavily on such ties, than it would be extremely helpful to understand how external forces shape patent production in your local area, and may potentially require you to forge close relationships with those regions that are crucial to the success of your innovation system.