Extensible standards and impact on technology switching costs Mark Pecen, Sr. VP, BlackBerry Research and Advanced Technology September 2013
Agenda Why standardize? Impact of standardization on adoption of cryptographic solutions Technology switching costs Why so much legacy infrastructure still exists Extensible standards Start from the beginning Program framework BlackBerry Z10 smartphone
Why Standardize? We can learn from our experience in ETSI wireless
Equipment Interoperability The first mobile radio systems were proprietary (e.g. an Ericsson radio only talked to another Ericsson radio and only in areas having infrastructure) Single-source supplier has large monopoly power Proprietary technology tends to create smaller and more tightly segmented markets Economies of Scale Standard technology enables large scale adoption, which fuels the learning curve
Economies of scope Re-use of platform for multiple products no need for special country-specific technologies Large-scale adoption potential For example, 3GPP wireless technologies (GSM/GPRS/EDGE/UMTS) have 87% of global subscriber market share
A wireless example: 3GPP technologies [Source: 4G Americas, 2012]
Standardization can create huge markets Positive network externalities can be created The more a certain technology is in use, the greater further adoption potential Reasonable trade-offs possible Give up certain proprietary advantages in exchange for the creation of large global markets
A double-edged sword Future innovation can be severely constrained Innovation produces Ricardian economic rents You may lose some competency-based rents as the price you pay for creation of large markets The large installed base of customers then becomes a constraint in itself
What does this mean for cryptography? Impact of standardization on adoption of cryptographic solutions
Cryptographic solutions are highly dependent on network externalities (i.e. the ability for others to use the same solution) Must be a mechanism for key exchange, an authority to authenticate identities, etc. A proprietary cryptographic solution may be appropriate in some cases, but is likely to occupy a small and specialized market segment where scaling isn t a problem. Such proprietary techniques generally have difficulty in scaling in deployment size, across national borders, etc.
Standardized cryptographic solutions anticipate market scale, scope and deployment scenarios. Standards tend to be defined that are simple to implement and deploy the details of which are designed to mean the same thing to system designers in Beijing as they do to those in Paris.
Consider technology switching costs Why so much legacy infrastructure still exists
What are Technology Switching Costs?
More Practical View
Example: Electric Power Generation in the United States General Direct Current (DC) electric power distribution has been outdated in the United States since 1903. Nevertheless, the last Consolidated Edison DC power generation and distribution center in the state of New York was turned off in 2007 customers still using DC were supplied with AC to DC rectifier units to ease switching costs.
Meanwhile in wireless Over 5 billion GSM subscribers Estimated that there are more than 2.8 million GSM base stations deployed worldwide 4 th Generation Long-Term Evolution Advanced (LTE-A) is the latest, most spectrally efficient and fastest cellular wireless technology available for mass deployment today but it s hard to suddenly disconnect millions of base stations, and to give up the roaming revenue they provide, even if they cost much more to operate than LTE base stations. Fortunately, there are extensibility features in the ETSI 3GPP standard that ease some of these migration aspects
Extensible Standards Start from the beginning
Extensible Standards Anticipate migration from one technology to future technology Ease or eliminate technology switching costs Relatively simple, if considered up-front
Extensible Crypto Standards Architecture, message structure to handle multiple crypto standards Not difficult, if defined early in standardization cycle Classical Processing #1 Classical Processing #2 Which Classical version? Quantum Processing #1 Which Quantum version? Quantum Processing #2 Classical or Quantum? Common Message Stream
Simple to add extensibility up front Difficult to impossible to implement after deployment
Way forward? I would urge my industry colleagues to consider the creation of an ETSI Technical Report (TR) to be drafted over the next year to consider, e.g. what might an extensible QKD architecture look like?