Kenedict Innovation Analytics

How Google Connects Technologies and Inventors: A Network View

Google has been making headlines with its myriad of inventions over the past few years. Google Glass is one of the most anticipated products in tech currently, and driverless cars should be becoming reality anytime now. At the same time, Google still focuses its innovation efforts relentlessly on established services as well. How does the company manage to connect its internal R&D expertise when it focuses on such divergent technologies? A network perspective provides an answer.

Written by André Vermeij, Kenedict Innovation Analytics

Google’s Research & Development expenditures are soaring. According to a recent study by Strategy& (the former Booz & Company), Google made its first appearance in the top 20 global R&D spenders in 2013. Much in line with this significant rise in R&D expenditures, Google received just under 2,000 patent grants in 2013 alone, amounting to almost double the amount of all grants received during all other years of the company’s existence. Many of these patents are directly relatable to established products and services such as Android, Google+, advertisement services and search engine technology, but many others trace back to exotic emerging technologies such as driverless cars, contact lenses with built-in cameras and balloon-based positioning systems.

Many of these technology areas seem largely unrelated at first sight. How does Google connect its diverse range of technologies and inventors across these areas? Network analytics provides an answer to this question.

To find out how Google’s technologies and inventor population are connected, this post presents an analysis of the company’s most recent patent applications based on the first four months of this year. The analysis that follows is thus based on a total of 1,287 patent applications by Google filed from January to April 2014. Google’s R&D network is analysed from two perspectives: (1) as networks of inventors and (2) as networks of patents and technology areas.

Google’s network of inventors

The inventors listed on each of the patent applications serve as the basis for constructing the inventor network. The full network is made up of 2039 inventors and 4816 connections between them. In line with the key question of this post, the visualization below shows the largest interconnected cluster of inventors in Google’s R&D network, consisting of 786 inventors with 2857 connections between them. Each node represents an inventor; each connection represents co-authorship on at least one patent application. Colours represent sub-clusters or communities of inventors.

The visualisation is fully interactive, so please feel free to zoom and click individual inventors to see their personal networks.

About 40% of all inventors in the full sample reside in the above cluster, together making up approximately 60% of all connections. What does this tell us about inventor connectivity within Google? Even though the sample is based on just the first four months of this year, it already shows that (coloured) communities of inventors are clearly interconnected, especially within the core of the cluster. Diving deeper, the communities consist of teams of inventors (the closely knit sub-clusters, as seen in the visualisation), which are together responsible for one or more patents. When taking a larger sample (for example, all patent applications over the past two years), an even higher connectivity is to be expected.

Of course, some communities are connected by just a few connections. The inventors responsible for these connections essentially serve as knowledge conduits within the network, effectively connecting diverse expertise areas and ensuring knowledge flow across Google’s various research teams. From the visualisation, we can distill who these knowledge brokers are – here’s a list (non-exhaustive):

Rich Gossweiler – a senior research scientist working in the area of human-computer interaction, user experience and user interaction design;
Hrishikesh Aradhye – Head of Search, Discovery and Infrastructure at Google Play;
Paul Buchheit – Gmail Creator and Lead Developer;
Douglas Eck – Research Scientist working at the intersection of music and machine learning;
Alex Faaborg – a Designer who has worked on Android, Google Now and Google Glass.

This shows us that Google’s social network of inventors based on patent co-authorship is relatively well-connected, with various expertise areas in the core of the cluster, and multiple areas in the ‘periphery’ of the cluster. However, this doesn’t tell us anything yet about how technology areas are connected within Google. Let’s turn to a different perspective to find out if this is the case as well.

Google’s technology network

Using slightly different assumptions, it is possible to construct a network of patents and associated technology areas based on the same dataset. Here, (groups of) patents are connected whenever these patents share one or more inventors. This allows for grouping of patents and technologies in the network, and provides a very clear picture of how technologies are connected within Google based on shared expertise. The visualisation shows the largest cluster of interconnected patents, consisting of 454 nodes/patents and 1428 connections between them based on shared inventor expertise. The full network (not shown) consists of 1287 patents and 2428 connections.

Similar to the inventor network, the largest cluster above consists of about 35% of all patents and approximately 60% of all connections in the full network. So, which technologies are present in this cluster? Zooming in at the top of the network, we encounter a green cluster with patents such as ‘Traffic Signal Mapping and Detection’, ‘Construction Zone Sign Detection’ and ‘Controlling Vehicle Lateral Lane Positioning’. Clearly a cluster of patents associated with the much-discussed driverless car Google is working on! The orange cluster just to the right of the car cluster mainly focuses on mobile location and positioning systems, while the blue cluster further to the right consists of various types of contact lens functionality. Google Lens, perhaps?

The dark orange area just below the car cluster contains patents on notifications, lockscreens and mobile devices, and can thus be directly related to Android. The purple and bright yellow clusters just below the core are another interesting example of connectivity between complementary technologies. Most of these patents relate to social network and sharing technologies, and are thus most likely associated with Google+. In contrast, the orange area just below these clusters mainly focuses on advertising, product placement and machine learning, yet is still connected to various other clusters.

As was done above for inventors, let’s also find out which are the key connecting technologies in the above network. Which patents and associated technology areas serve as brokers here?

User Interfaces for Head-Mounted Devices: connects the driverless car and Android clusters;
Serving Advertisements based on Content:
connects various sub-clusters in the bottom orange cluster;
Mobile Sitemaps:
connects the Google+ cluster with speech/text recognition and search technology;
Location-Based Responses to Telephone Requests:
connects the bottom orange advertisements cluster with the driverless car and Android clusters;
Indoor Localization of Mobile Devices
: connects the car cluster with the orange clusters to its right (location/positioning systems).

Of course, these are just some examples of the connected clusters and broker technologies which can be found in the network. For example, other clusters include (multi)media technology and technology related to applications & software recommendations. Many more interesting links between technologies can be found when further examining the visualisation. Please do let me know if you find any other interesting insights!

Conclusion

The above analysis shows that Google manages to utilise its in-house R&D expertise across its various technology areas, thereby leveraging the diverse knowledge of its inventor population. Seemingly unrelated technology areas are connected to each other based on individual inventor expertise spanning multiple areas. Many of Google’s key inventors have multi-disciplinary knowledge, enabling them to traverse different research areas and connect research teams across a wide variety of both established and emerging technologies.

Network analytics such as the above can thus provide us with previously hidden insights concerning Google’s internal R&D structure and collaboration networks. Stay tuned for future posts with similar analyses for other organisations and technologies!

 

Leave a Reply

Kenedict’s mission

To significantly improve Research & Development organisations’ decision-making by ensuring day-to-day usage of a social network perspective in the areas of Open Innovation, Competitive Intelligence, Human Resource Management and R&D Policy.

Contact Kenedict

Phone: +31 61 55 00 888

E-mail: info@kenedict.com