Platform Leaders, True Believers, and Coordinated Innovation: The Role of Key Architects In Influenc

1
Platform Leaders, True Believers, and Coordinated
InnovationThe Role of Key Architects In
Influencing Technology Trajectories

• Erica R.H. Fuchs
• Assistant Professor
• Department of Engineering and Public Policy
• Carnegie Mellon University

2
Moores Law

• The complexity for minimum component costs has
  increased at a rate of roughly a factor of two
  per year Certainly over the short term this rate
  can be expected to continue (Moore 1965)
• or double every year (Moore 1975, 1995)
• Somewhere along the way
• Doubling of number of components / circuit every
  18 months
• Doubling of processing power / chip every 18
  months
• Doubling of computing power every 18 months
• Price of computing power falling by half every 18
  months
• Enormous Impact on Information Economy
• 50 growth in GDP in last decade due to
  technology progress in information technology
  (Jorgenson 2001)

3
Critical Technological Challenges

• A multicore trajectory
• IBM 2001, AMD 2005, Intel 2006
• New challenges in
• Processor Architecture (Hennessey and Patterson
  ACM Queue 2007)
• Interconnects on and off chip
• Processor to processor
• Processor to memory
• Processor to accelerator
• Programming of chip
• Requires radical architectural innovation
  (Nelson and Winter 1982, Abernathy and Clark
  1985, Henderson and Clark 1990)
• Integration of fringe technological communities
  optoelectronics, parallel programming

4
What are the institutions and processes
coordinating innovation to meet Moores Law?
How well-suited are these to continuing Moores
Law in the upcoming decades?
5
Moores Law The Metronome for the Computing and
Communications Industry

• I think of Moores Law like a drummer on a
  slave ship. He pounds down this hammer and goes,
  you know, 0.35 microns. Boom. 0.25 microns.
  Then 0.18. And on that slave ship, the guys
  that own disk drive, the disk drive makers and
  the processor designers, and the software guys,
  and the fiber optic guys, currently all those
  guys have oars. And if they pull all at the same
  time, the ship goes forward, really good. But if
  any one of them does this, and screws up, then
  none of the oars work. And the ship wont go
  anywhere.
• (Frank Levison, former CTO Finisar Corp.)

but what determines the direction of the
technology?
6
Background A New Game?

• Significant changes over the last decade in
• Industrial and Government Funding
• Shift away from corporate RD labs (Mowery 2000,
  NAS 2006)
• Many U.S. firms externalize portion of RD
  (Mowery 2000)
• Industry Structure
• Initially few pioneering firms supplied
  computers, today hundreds loosely linked
  suppliers (Breshnan 2000)

7
Background A New Game?

• Significant changes over the last decade in
• Industrial and Government Funding
• Shift away from corporate RD labs (Mowery 2000,
  NAS 2006)
• Many U.S. firms externalize portion of RD
  (Mowery 2000)
• Shift to venture-funded small and medium sized
  firms, from research to development (NAS 2006)
• Shift in DARPA funding from universities to
  industry (Lazowska and Patterson 2005, Markoff
  2005, Joint Statement 2005)
• Industry Structure
• Initially few pioneering firms supplied
  computers, today hundreds loosely linked
  suppliers (Breshnan 2000)

8
Changes in Innovation Environment

• Earlier part of 20th Century, RD in corporate
  laboratories
• RD shifted to venture-funded small and medium
  sized enterprises (NAS 2006)
• Small firms considered more flexible, innovative
  (Piore 1984, Pavitt 1987, Powell 1990, Acs 1991,
  Rothwell 1994)
• Large firms encouraged to source innovation needs
  to smaller firms through technology alliances and
  acquisitions (Cohen 1990, Lamb 1997, Chesbrough
  2003)

9
New Model Networks of Innovators

• Complex networks of firms, universities
  government labs, esp. in industries with rapid
  technological progress (Powell and Grodal 2005)
• Increased reliance on external sources of RD
  (NRC 1998, Mowery 19997, Powell and Grodal 2005)
• Interdependency of Innovation trajectories across
  products and services (Breshnahan 2000, Macher,
  Mower, and Hodges 2000, Powell and Grodal 2005)
• Marked increase by mid-1990s in alliances linking
  U.S. firms to their domestic competitors (NSF
  1999)
• Who are the main actors?
• How do they coordinate to influence technology
  trajectories?

10
Networks, Knowledge Flow, Coordination

• Strong vs. weak ties (Granovetter 1973)
• Knowledge flows across ties (Argote, McEvily,
  Reagans 2003)
• Properties units individual, organization
• Properties knowledge tacit (Nonnaka 1991),
  codifiability (Zander and Kogut 1995), stickiness
  (Von Hippel 1994), ambiguity (Szulanski 1996),
  complexity (Hansen 1999 Sorenson, Rivkin, and
  Fleming 2002)
• Properties relationships between units
• Relational proximity (common identity, language)
  similarity between scientists (Song, Alemeida, Wu
  2003), communities of practice (Brown and Duguid
  2001)
• Geographic proximity (Allen 1970, Teece 1977,
  Mansfield 1982, Saxenian 1994, Porter 2001)
• Labor markets (Azoulay 2003)
• Coordinate?
• Platform leadership complementary innovations
  (Gawer and Cusumano 2002)
• Engineers share knowledge among rivals (Allen
  1983, Von Hippel 1987)
• Game Theory
• Cooperative Technical Organizations (Rosenkopf
  and Tushman 1998)
• Task forces, standards bodies, technical
  committees
• Small collection experts, span multiple
  organizations, limit group size

11
Co-Evolution Networks and Technology

• Cooperative Technical Organizations
• Small collections of experts spanning multiple
  organizations
• Leaders limit group size and who is invited to
  attend
• Ex. task forces, technical committees, standards
  bodies
• Networks linked to cycle of industry
• Eras of incremental change CTO membership
  relatively constant
• Eras of ferment movement of new members into CTO
• enable reformation of interorganizational
  networks
• select among competing technological alternatives
• Interdependency of innovation trajectories
• Limited to formal network ties
• How? By what processes?

12
Platform Leaders Complementary Innovation

• Interdependency of Innovation trajectories across
  products and services (Breshnahan 2000, Macher,
  Mower, and Hodges 2000, Powell and Grodal 2005)
• Platform (Gawer and Cusumano 2000, Gawer and
  Henderson)
• one component or subsystem of an evolving
  technological system is strongly interdependent
  with most of the other components of this system
• Platform Leader (Gawer and Cusumano 2000)
• Firm ensure innovations in complementary products
  on which the evolution of the firms own products
  is dependent

13
Knowledge Sharing Among Rivals

• Tendency of engineers to share information with
  rivals (Allen 1983, Von Hippel 1987, Saxenian
  1994) informal
• Game theory
• Saxenian? -- geographical proximity not enough
• Strong ties -- relational proximity
• Communities of practice (Brown and Duguid 2001)
• Coopetition

14
Methods

• Grounded Theory-Building (Glasner and Strauss
  1967)
• Two starting propositions Intel, DARPA, (ITRS)
• Snowball effect (Rosenkopf and Tushman 1998)
• Five leading computing firms (Intel, AMD, IBM,
  HP, Sun)
• Start-ups, Universities, Government Institutions
• 35 semi-structured interviews (Miles and Huberman
  1984)
• Personal history, network of architects (75
  references)
• Participant observations
• Conferences, within firms
• Four-day ethnography elite, invitation-only
  workshop
• Who is influencing technology directions?
• How? -- i.e. by what processes?

15
A few key architects

• (Not Intel, not DARPA)
• Context Matters --
• Two groups different identities, incentives,
  characteristics

16
A Few Key Architects

• Establishment
• theres a small number of 15-20 people who
  actually know what all of these I-O technologies
  are going to be, and where theyre going to go,
  and how they fit together. And, we could tell
  you where theyre all going, and we could tell
  you, in fact, what each of our companies is going
  to do.
• Fringe
• We were part of a little parallel programming
  mafia. Once we realized we both were working in
  the same area, we probably met once a week.

17
A Few Key Architects

• theres a small number of 15-20 people who
  actually know what all of these I-O technologies
  are going to be, and where theyre going to go,
  and how they fit together. And, we could tell
  you where theyre all going, and we could tell
  you, in fact, what each of our companies is going
  to do.
• So you really have inside that processor realm,
  theres probably a dozen guys, between those
  three companies, who are actually setting where
  those things are going to go. In I-O, theres
  probably another dozen people who talk about
  what it would look like. And Im sure theres
  probably a similar thing if you pick optics, if
  you pick memory technologies, or whatever.

18
Established ArchitectsCoordinating Innovation
-- Why

• Some of your closest competitors you actually
  share more information sometimes than with your
  suppliers, simply because youre trying tofigure
  out a solution or an environment thats a win-win
  for all. And not get into this really tweaky,
  fractured stuff that just throws everything off
  kilter.
• You just cant make anything happen in industry
  on your own because its completely impossible.
  You have to find a partner, you have to convince
  your competition this is the right thing to do.
• Youre guiding people to do the most common
  sense things, and they ask, why are you helping
  me with this? And the fact is you give them
  information so the suppliers are in the right
  place for you.

19
Established ArchitectsCoordinating Innovation
-- How

• You get to know what people are looking for and
  why. And you find, the industry as a whole is
  all actually on the same track.
• you have this connection, and we share
  information. You know, we dont violate our
  confidentiality, because some things are very
  confidential, but we know enough about what is
  going on, that we can talk about problems or
  whatever, and we can pretty much figure out what
  theyre trying to solve.
• I can tell you what youll find. I was there
  (at the workshop), and theyre all presenting to
  each other what theyre going to do. Theyre all
  talking to each other. And theyre all doing the
  same thing.

20
Established ArchitectsCoordinating Innovation
-- Who

• Despite extensive coordination, higher ups say
  choosing the next technology a wet finger test
• The best way to get my company to make the
  product I develop is to leak my idea to
  ltcompetitor companygt, and then go back to my boss
  and say, hey, look what ltcompetitor companygt is
  doing, and hell decide we better do it, too.

21
Key Architects The Front End of the Pipeline

• Going that direction regardless of the funding
• Believers you believe in what youre doing and
  that its the right thing
• Small cliques Talking regularly to 2-6 people
• How about this? What did x paper actually mean
• Need to evangelize, develop communities
• Graduate students, at/create conferences
• Government play a key role in education, funding,
  development communities

22
Fringe ArchitectsCliques, Mafias

• We were part of a little parallel programming
  mafia. Once we realized we both were working in
  the same area, we probably met once a week.
• Its a real clique between MIT, Stanford, and
  Berkeley.

23
Fringe ArchitectsIdentity -- Outsiders

• So its a small group of people. Half of them
  have a visually apparent speech problem.
  Somethings a little bit off about the people
  attracted to this problem.
• We tried to get the government interested in
  funding research in this area, but they werent
  interested. They couldnt see the problem. The
  vendors also arent interested. Its really hard
  to convince people that this is a problem. They
  say, IBM wouldnt build it if they couldnt
  program it. Well, thats nice, but its not
  true.

24
Fringe Architects Who -- Knowledge Complexity,
Limited Talent

• I have the kind of brain that can solve these
  types of problems. Theres sort of this group of
  people who really understand these ideas.
  Its like writing a song. Anyone can write a
  song. You can write a song. I can write a song.
  Anyone can serial program. But writing a
  program for 100,000 processors is like writing a
  symphony. Only a few people can write a
  symphony.
• Theres not a lot of people, and its slow
  finding people to add to the community. You need
  someone with enough passion and talent.

25
Fringe Architects Incentives -- Direction
Regardless of Funding

• Are you familiar with the allegory of the cave?
  Outside (is) a campfire. The people inside the
  cave are trying to deduce what is happening
  outside from the shadows. Its the same thing
  with the funding. The technology direction is a
  separate thing.
• A lot of technology being offered is just being
  renamed. Anant, for example, was already
  developing this technology. When the BAA came
  along, he just repackaged what he was doing to
  look like it fit the BAA The Stanford guys are
  like Anant, in that they already were developing
  technology. Mark Horowitz (at Stanford) was
  already going to do what he was doing, regardless
  of the BAA.

26
Fringe ArchitectsWhy (Incentives) -- True
Believers

• Q So why are you doing this? I mean, youre
  not getting funding, why does someone do this?
• A Because you believe in what youre doing and
  that this is the right thing to do.
• The community was made up of believers.
  Believers that this is a superior way of
  maintaining research. There was only a total
  investment of 5M over 5 years. That this
  technology is actually being used is a big feat.

27
Fringe ArchitectsHow -- Beyond whats published.

• Q So what are the conversations like between
  the five to six people with whom you talk?
• A Hey, did you see this? I noticed this.
  What did you think of that? I saw this persons
  paper, whats really going on there? You get
  the real scoop on stuff.

28
Networks of Innovators Context Matters
29
Implications for Architectural Innovation

• Radical architectural change (Henderson and Clark
  1990)
• Firm-competence destroying (Nelson and Winter
  1982, Abernathy and Clark 1985)
• Need to re-integrate functions (Fine 1998
  Chesborough 2001, 2003)
• Theory hold for so many interlinked, co-dependent
  firms?
• Lessons from/for network theory
• Brokers
• Co-evolution of networks and technology
  (Rosenkopf and Tushman 1998)
• By what processes?
• Does network context matter?
• Network origins, evolution, extinction
• Lessons for other industries
• Inter-firm coordination
• Standards bodies, Roadmapping institutions