Evolutionary Patterns in Technology Ecosystems

Transcription

1 Evolutionary Patterns in Technology Ecosystems 4 th Intelligent Storage Workshop May 10, 2006 Jesse C. Bockstedt (with Gedas Adomavicius)

2 General Research Problem Explain the evolution of a technology and account for complex system of factors that influences its development Model related technologies and their evolution with respect to the analysis of a given context Provide insights for technology development decision making AGENDA 1. Theoretical background and conceptual model 2. Patterns of technology evolution and evolutionary cycles 3. Wi-Fi technology empirical study 4. Limitations, conclusions and future work

3 Theoretical Background According to Ziman (2000), technological innovations in an industry are so interrelated that one might describe them as occurring in an ecological system of co-evolving artifacts Accepting a system and population view of technology evolution: Iansiti and Levien (2004), Rosenkopf and Tushman (1994, 1998), Saviotti (1996) Defining the structure of the system with theories of hierarchies and roles in technology evolution: Rosenkopf and Nerkar (1999), Tushman and Rosenkopf (1992), Baldwin and Clark (2000), Podolny and Stuart (1995) Defining change in the system with theories of technology trajectories and patterns of innovation: Campbell (1990, 1994), Dosi (1982), Sahal (1985)

4 Ecosystem Model of Technology Evolution Product/Application Component Support/Infrastructure Specific vs. general ecosystem view Three specific roles technologies can play: 1) Product/Application, 2) Component, and 3) Support/Infrastructure Relationships among technologies are defined by the roles Influences among technologies shape patterns of evolution

5 Digital Music State Diagram Paths of influence in digital music ecosystem

6 WiFi Technology: Data 2425 WiFi technology certifications issued by the WiFi Alliance ( between Mar 2000 & Dec 2005 WiFi Alliance Technology Categories: Compact Flash (Component) External Card (Component) Internal Card (Component) USB Client Device (Component) Cellular Convergence (Product) Embedded Client (Product) PDA (Product) Access Point (Infrastructure) Ethernet Client Device (Infrastructure) Wireless Printers and Print Servers (Infrastructure) 3 types of technical functionality classifications: IEEE Standards (e.g b, g) Security (e.g. WPA, EAP) Multimedia (e.g. WMM, WMM PowerSave)

7 Data b Certifications by Category Category # Wifi Alliance Category Compact Flash 1 External Card 2 Internal Card December 6, 1999 August 12, 2000 April 19, 2001 December 25, 2001 Septembe r 1, 2002 May 9, 2003 Date January 14, 2004 Septembe r 20, 2004 May 28, 2005 February 2, 2006 October 10, 2006 USB Client Device 4 Access Point 5 WiFi Alliance Category g Certifications by Category Ethernet Client Device 6 Wireless Printers and Print Servers 7 Cellular Convergence October 1, 2000 April 19, 2001 November 5, 2001 May 24, 2002 Decembe r 10, 2002 June 28, 2003 January 14, 2004 August 1, 2004 February 17, 2005 Septembe March 24, r 5, Embedded Client 9 PDA 10 Date

8 Q2 05 Q b Certifications b Certification (Percentage) Percentage of Total Certifications 0 Q2 00 Q4 00 Q2 01 Q4 01 Q2 02 Q4 02 Q2 03 Q4 03 Q2 04 Quarter Component Products Infrastructure

9 802.11b & g Certifications b&g Certification (Percentage) Percentage of Total Certifications Q2 00 Q4 00 Q2 01 Q4 01 Q2 02 Q4 02 Q2 03 Quarter Q4 03 Q2 04 Q4 04 Q2 05 Comp-b Prod-b Infra-b Comp-g Prod-g Infra-g

10 WiFi Technology State Diagram Different cycle - infrastructure plays enabling role in the ecosystem (as opposed to supporting)

11 Empirical Problem Are there pockets of technology evolution activity within an ecosystem? And how can we identify them? Can we clearly map the temporal changes within an ecosystem? Approach Data Mining Data mining allows the data to speak for itself Identify naturally occurring patterns in technology attribute data

12 Proposed Technique Technology Attribute Clustering in Overlapping Time Periods Divide WiFi data into two year sets with one year overlap (e.g., , ) Use hierarchical clustering technique (Ward s method with Euclidean distance metric) to reduce data into clusters of similar technologies Define edges between clusters based on number of shared technologies Determine cluster centers based on average attribute values

13 Evolutionary Graph

14 Finer Granularity

15 Insights Allows researchers to generate evolutionary maps of technology Patterns of evolution Relationships between technological activities Complexity of ecosystem Identifies key evolutionary steps and trajectories within an ecosystem Birth, death Convergence (integration), divergence (separation)

16 Limitations and Challenges Defining the boundaries of an ecosystem Data collection is difficult Comprehensive storage-related data??? What is the appropriate level of analysis? First release or patent of a technology Frequency or number of products using a new technology

17 Conclusions and Future Work This work represents one piece of a new research stream on information technology evolution Developing new data analysis methods and tools for representing technology evolution Refining data mining and time series data analysis Software tools for analyzing and visualizing temporal data Incorporating other elements into the model Firms and economic factors Consumers, adoption, and social factors