Technological and Organizational Dynamics (a problem solving perspective)

Transcription

1 April 11-12, 2011 DIMETIC, Strasbourg Technological and Organizational Dynamics (a problem solving perspective) Stefano Brusoni

2 Table of Contents (today and tomorrow) Motivation Part II - The field, and its gaps From modularity to modularization The role of knowledge integration capabilities Part II - The method, and a bit of gap filling Qualitative research methods Sampling strategies Validity and bias

3 Motivation Explain the (very) unlikely New trajectories and paradigms Breakthrough innovations Change in business models New architectures modularity and (de)modularization

4 Reading list Session 1 Nickerson J. and T. R. Zanger (2004) A knowledge-based theory of the firm The problem solving perspective. Organization Science 15 (6): Sanchez R. and J. T. Mahoney Modularity, Flexibility, and Knowledge Management in Product and Organisation Design. Strategic Management Journal, 17, Winter Special Issue: Chesbrough, H., and K. Kusunoki The Modularity Trap: innovation, technology phase-shifts, and the resulting limits of virtual organisations. In Nonaka, I. and D. Teece, (eds.) Knowledge and the Firm, Russell Sage Press.

5 A long and distinguished history No way I m listing names here Scott W. R. and G. F. Davis (2007) Technology and Structure. Chapter 6 in Organizations and Organizing Rational, natural and open systems perspectives. Prentice Hall Three main building blocks 1. Macro-framework for comparing things Problems, of different sorts 2. Micro-processes to make sense of things Search, of different scope 3. Empirically observable relationships Interdependencies, of different strength

6 1. A framework for comparisons Starting point: opportunism vs. knowledge generation Zingales vs. Simon. Do organizations emerge when markets fail? Or do organizations exist because they can do something better than markets?? And what is that something, btw? Key managerial problem is not about monitoring opportunistic individuals, but rather the selection of the problem which is most likely to generate desirable and appropriable knowledge and capabilities. After the problem is chosen, the manager must organized employees in order to solve it. Issue here is identifying the criteria to match the right problem with the right type of institutional set up. Hence, the need to compare things!

7 What is a problem? Hora System 1 System 10 Subsystem 1 Subsystem 10 Component 1 Component 10 TEMPUS Component 1 Component 1,000

8 High interaction problem Moderate interaction problem Low interaction problem

9 Hora and Tempus, and beyond Complex systems Hierarchical & decomposable systems Problem solving vs. problem framing What do economic institutions do? They choose which problem to solve and how to frame it decomposability They evaluate the solution they ve found aspiration levels If satisfied stop; else they activate search processes Different types of search processes Local search Heuristic search

10 2. Search

11 2. Search

12 NK (N= 3, K=0) 000 f x f.y. 0.6 f..z 0.8 f xyz (0.63) 110 (0.80) (0.53) 111 (0.70) (0.53) 100 (0.70) (0.43) 101 (0.60)

13 NK (N= 3, K=2) 000 f x f.y. 0.3 f..z 0.5 f xyz (0.70) 110 (0.63) (0.53) 111 (0.43) (0.47) 100 (0.83) (0.50) 101 (0.40)

14 NK for real ELECTRICAL ENGINE ICE ELECTRIC TRANSMISSION SYSTEM BATTERIES PACK FUEL TANK CONTROL SYSTEM Hybrid propulsion system N=6, low K Control Box Data Gathering. ELECTRICAL ENGINE ICE Cables Converter BATTERIES PACK FUEL TANK Control box Sensor System Inverter HMI Digital Signal Processor Commutation System Hybrid propulsion system N=14, higher K Data Gathering System Data Transmission System Display Source: Vaccaro, Brusoni and Veloso, JMS 2010

15 NK and search for real TECHNICAL COMPETENCIES REPLICATION IN PROJECT M ARCHITECTURAL REPLICATION COMPLEMENTARY ASSETS GRANULARIZATION OF THE DESIGN SPACE PRODUCT INNOVATION REPLICATION IN PROJECT T STRATEGIC OBJECTIVES ARCHITECTURAL REPLICATION FUNCTIONAL REPLICATION Source: Vaccaro, Brusoni and Veloso, JMS, 2010

16 Two types of search Search Aspiration levels activate search (or not) Local or directional search Guided only by prior experience Experiments run in isolation and independently from each other Each is evaluated against the impact on overall performance. Ideal for low interaction problems E.g. DELL assembling PCs designed and developed by others, Sony on batteries, etc. etc. Heuristic or cognitive search Guided by cognitive representations (models of the world) which simplify the problem space Used for high interaction problems Technological trajectories, dominant designs!!! Semiconductors: smaller, faster Chemical reactions: reaction efficiency, larger scale Dominant design: cars 4 wheels, closed body, ICE

17 Type of problem Type of search Decomposable Nearly decomposable Non decomposable Directional (or local) search Heuristic search

18 Knowledge formation hazards Heuristic search sounds good, but rely necessarily on distributed knowledge, which need to be coordinated. Two hazards in knowledge formation and sharing Knowledge appropriation hazard Derives from Arrow s information paradox. Big issue for heuristic search. Strategic knowledge accumulation hazard Individuals try to shape the search heuristics which influence the development of new knowledge in order to max the importance of their own already accumulatedknowledge NIH syndrome Early cases in AIDS research (or human BSE) Virologists vs. Immunologists in current research on an anti-hiv vaccine Both hazards are less important when problems are decomposable But not every problem is decomposable

19 Governance for mitigating hazards How do we mitigate the hazards that slow down knowledge transfer? By choosing the governance mode which best fit the characteristics of the problem at hand. Markets Hierarchy, authority-based Hierarchy, consensus-based The three proposed governance modes differ in terms of their: Decision rights over the path of solution search Communication channels to support knowledge transfer Incentives to motivate search

20 Markets Hayek (1945) The marvel of the market is that it provides inducements which make the individuals do the desirable thins without anyone having to tell them what to do (p. 520) Markets dispense with the need of conscious control (p. 527) The most significant fact about this system is the economy of knowledge which it operates, or how little the individual participants need to know [about other actors] in order to be able to take the right action (p. 527) The path of search is merely the aggregation of independent search paths, autonomously chosen by individual actors They respond to the price system They are motivated by individual profits They need limited knowledge transfer Ideal for decomposable problems!! Which require local search only, and feedback from direct, local experience.

21 Markets, examples DELL computers and Sony in development of litium batteries Automotive breaking systems tire systems Pharmaceuticals (?) Disaggregation of drug development process (But see Pisano HBR 2006)

22 Hierarchy, authority-based Authority, the centralization of decision making, serves to economize on the transmission and handling of knowledge (Arrow, 1974: p. 69) Direction substitutes for education Demsetz (1988) If the problem is not perfectly decomposable, you need a central figure which directs search toward certain clusters of interdependencies This central figures authoritatively limits the search space It can activate heuristic learning, wider than local search, but still within pre-defined boundaries. It limits horizontal communications, though, in so doing limiting knowledge transfer. But it control the incentives to strategically manipulate the search path to increase the value of specific bodies of knowledge However, managers tend to be quite overconfident about their own skills and abilities. E.g. Steve Jobs in the early 1990s. As problems become more complex, they cannot cope. Direction becomes an inefficient way of coping with the problems of knowledge hazards

23 Hierarchy, authority-based - examples Project management in most engineering intensive industries Aerospace Jet engines Process engineering Universities Curriculum design!!

24 Hierarchy, consensus-based Hierarchy (i.e. the firm) is a device which favours knowledge transfer. The emphasis here is not on authority, the fact you cannot go to court in case of misunderstanding about what one is expected to do the emphasis here is on the fact that the boundaries of the firm defines a space of interaction and communication which support the development of a common language and a commonly accepted way of solving problems. March 1991 on socialization What kind of problems? Complex problems!! Arrow, Simon, March, Nelson and Winter they all talk about common identities, which support heuristic search, the exploration of uncertain alternatives, the development of common criteria to evaluate experiments Danger: Veto power Consensus take time to be achieved and to be maintained death by committee (vs. the adults!!!). Ties that bind may become the ties that blind (Powell & Smith-Doerr 1994: 393) E.g. a research centre

25 Hierarchy, consensus-based -examples Anti-HIV vaccine research Enormous potential for developing countries Anti-AIDS drugs are effective but expensive Enormous complexity of the problem Different strains of virus Rapid mutability Enormous fragmentation in the field Molecular biologists vs. vaccinologists Public institutes vs. private outfits Big pharma vs. specialized biotech Result: no progress whatsover 1996/7: IAVI International AIDS Vaccine Initiative

26 Type of problem Degree of interaction among subproblems Solution landscape characteristi cs Appropriate search mode Need for knowledge transfer Knowledge hazards Predicted governance forms Decomposable Low Few peaks Directional Low Low threat M Nearly decomposable Moderate Intermediat e ruggedness Directional and heuristic Moderate Moderate threat ABH Non decomposable High Many peaks Heuristic High High threat CBH

27 Key concepts Problem space Problem representation Interdependencies Search Directional/Local search Heuristic/Cognitive search Hierarchy Authority-based Consensus-based

28 3. Empirically observable relationships (aka hypotheses) Modularity A literature which identifies a set of candidate relationships between organization and technology

29 Core concepts Reinforced Overturned Linkages between core concepts and components Unchanged Changed Incremental innovation Architectural innovation Modular innovation Radical innovation Source: Henderson and Clark, 1990: 12

30 Core concepts Reinforced Overturned Linkages between core concepts and components Unchanged Changed Incremental innovation Photolithographi c equipment Litium batteries in laptops Radical innovation Source: Henderson and Clark, 1990: 12

31 Modes of learning Learning about components functions and designs Moderate Significant Learning about component interactions and configurations Moderate Significant Incremental learning at the component level Architectural learning Modular learning at the component level Radical learning at the architectural and component levels Source: Sanchez and Mahoney, SMJ 1996

32 Modular Products

33 and Modular Industries The Old Vertical Computer Industry - Circa 1980 The New Horizontal Computer Industry - Circa 1995 Sales and distribution Sales and distribution Application software Retail Stores Superstores Dealers Mail Order Word Word Perfect Lotus I-net SAP Application software Operating systems DOS and Windows OS/2 Mac UNIX Linux Operating systems Computer Disk drives Packard Compaq Dell HP IBM Etc Bell Seagate Quantum Western Digital Maxtor Computer Chips Intel Architecture Motorola RISC Chips Contract manufacturers Printers Selectron SCI Flextronics HP Jabi l Epson Celestica IBM DEC Sperry Univac Wang Source: Adaptation from Only the Paranoid Survive by Andrew Grove, 1996.

34 Advantages Increase the number of options (e.g. Baldwin and Clark) Increase division of labour and the use of market coordination (e.g. Sanchez and Mahoney; MacKormak) Decouple the development of the architecture from the development of modules (e.g. Tiwana) Increase flexibility Parallel search (speed of experimentation), e.g. Ethiraj and Levinthal Upgradeability, e.g. Langlois and Robertson Economies of substitution (without cannibalization), e.g. Garud and Kumaraswami

35 Disadvantages Very costly architecture to put in place Needs to achieve a thorough understanding of the system Trade off at strategic level Performance vs. variety Hold up problems and TC issues Learning trade off Speed of search vs. breadth of search

36 I am very surprised. I have to think about it (Steve Ballmer on Oracle s take over of Sun, The Economist, April , p. 63) Oracle (software giant) buy Sun (computer maker). IBM and Cisco pre-empted But also: HP buy EDS (service) Cisco to sell servers IBM and SAP? Why? Maturity only way to grow is invading somebody else s niche Integrated solutions people want to buy the whole package Blurring of boundaries cloud computing Back to an integrated structure?

37 Gap Long term viability of modular organizations depend upon the ability of introducing new architectures and platforms BUT: Lack of empirical analysis of processes of modularization, or remodularization, or de-modularization. Modularity literature normally accepts the idea that architectural and component-level knowledge are fully separable Some firms specialize on developing architectures, others focus on components??? The Turing machine-view of industrial evolution: platform- and industry-evolution are themselves modular processes.

38 Issue 1 Modularity as an organisational strategy Key question: how to organise for the development of new modular or not- product architectures? How to overcome the tunnel vision effect? How to avoid so-called modularity traps?

39 Myopia: an example from the HDD industry Clear cut research trajectory Increase recording density Three different technologies (and related shifts) Ferrite-based (1960s and 1970s) Thin film based (1970s and 1980s) Magneto resistive head (1990s --)

40 Technology phase shifts New technologies start with integral architectures (usually) Interdependencies and functionalities are poorly understood Suppliers are not knowledgeable about components and specifications are hard to codify anyway Administrative coordination is required More rapid access to useful information Weaker incentives to holding up and strategizing Appropriability issues

41 Technology phase shifts As time goes by, understanding about how the technology works increase and diffuse Dominant design Standards emerge Modularization of technologies and products Vertical disintegration becomes a viable option Useful information about interfaces embodied in standards Established standards enable competition among suppliers (internal and external) Reduction of appropriability issues

42 From thin films to MR IBM deliver new insights in early 1990s Magneto resistive head!! We don t fully understand the physics [of it], but we are able to replicate it fairly consistently (1992) New problems emerge Electrostatic discharge New heads are very sensitive to this problem Producers did not know how to control for it, though Thermal asperities Heads gets closer and closer to the media (at faster rotating speeds and thus higher temperatures) Casual touch cause heat, which in turn damage the media Leading US suppliers of heads report massive losses.

43 From thin films to MR What about Japanese suppliers? Fujitsu, Toshiba, Hitachi and NEC Toshiba modular strategy In house disk drive designs Outsourced design & production of heads and media MR perceived as improvement along established trajectory

44 Toshiba s modular structure It was very crucial for us to have good outside suppliers of key components in order to achieve efficient product development. In house development [ ] requires heavy investments, taking effort over a long time. We have tried to have at least two suppliers for a particular key component like heads because [we want] stable supply and cost reduction through competition (Kamamura, 1998)

45 Toshiba s modular strategy We viewed HDD competition as purely a matter of speed. [ ] But in the case of MR heads, Toshiba could not be first. We tried to define the specs required for our heads. But we couldn t completely specify them because we were less knowledgeable about MR heads than our suppliers. [ ] it appeared more effective and efficient for us to leave the major part of head-related problem-solving in suppliers hands simply because they were component specialists and knew more than we did (Kamamura, 1998)

46 Toshiba falls in modularity trap (myopia) Problem frame Speed decomposable (modular) problem Exploit specialization and competition among suppliers (parallel experimentation, upgradeability) Search processes (routine-based) Local, along modular lines Toshiba design the system Head suppliers focus on heads Structure Modular network of suppliers Major mistake: innovation in heads require adjustment in production process and its control routines Suppliers do not have a clue on these issues Toshiba know about the process, but not about the heads

47 Fujitsu and the HDD The case of Fujitsu exemplifies the case of a firm that successfully managed the introduction of a new product architecture, stemming from a major technological breakthrough embodied into the magneto-resistive head. During the modular phase, Fujitsu like other firms relied on a decoupled network of external suppliers. Unlike its competitors, Fujitsu did not discontinue its investments in systems knowledge and materials and component technology in its R&D labs. Fujitsu s systems knowledge went well beyond the range of products and components that the company produced in house.

48 Fujitsu s learning processes At that time [early 1990s] we had neither an off-the-job nor onthe-job training programme for mastering MR technology. What we did was on-the-job learning, which includes lots of trial and error. Though we had not been so knowledgeable about MR, we could be rather careful about how to deal with the new technology when using it within the HDD system as a whole (Fujitsu engineer)

49 Fujitsu and the TA problem We saw two avenues to correct this problem, the drive side or the head itself. Seemingly the problem was head-related. However [ ] corrections efforts on the drive system side substantially contributed to solving the problem. This problem-solving required the coordinated efforts of different departments. We were sceptical about relying on only the head suppliers to fix this problem, because they might define the problem too narrowly, which might limit their ability to find the most effective solution to the problem (Sugihara et al 1998)

50 Fujitsu s strategy and structure Coordination among different departments and in house R&D Flexible categorization of design activities and, hence, flexibility in assigning engineers to different units (against strict specialization) Co-location of cross-departmental working groups (maintain diversity and get diverse people in touch!) System-based differentiation: experiment with different architectures (i.e. different frames to limit simplification) Actively engage one key supplier in R&D efforts Competition for internal supplier Acquire process capabilities (head-related) Establish industry de facto standard selling to other firms

51 Fujitsu does not fall in modularity trap Problem frames Speed Exploit specialization and competition among suppliers (parallel experimentation, upgradeability) But also in-house system-wide exploration capabilities Search processes Local, along modular lines (exploitation) Fujitsu design the system while head suppliers focus on heads But also heuristic search: e.g. on interdependencies among head and media (exploration) Structure Modular network of suppliers But also in house flexible, highly interdependent organization Major achievement: innovation in heads require adjustment in production process and its control routines

52 Lessons from Fujitsu Modular product yes Modular organization yes, in terms of the production and engineering activities carried out in house. Modular knowledge bases no! Fujitsu maintained wide capabilities. Similar evidence is emerging from a range of industries as diverse as aero-engines, chemicals, oil, automotive. Systems integration activities Problem (re)framing

53 Issue 1 Conclusions (chunk 2) Conclusion I product modularity has major implications for organisational design, sometimes. Things other than product characteristics are at stake The evolution of specialised bodies of knowledge Appropriability considerations (big gap in the literature, btw) Risk architecture (even bigger gap, possibly) Conclusion II the division of labour does not necessarily match the division of knowledge Hard to decouple completely architectural developments from module-level developments Hence, role for systems integrating firms

54 Later Issue 2 How do new problem frames come into being? Technological frames the case of radical process innovation robotization Organizational frames the case of radical managerial innovation Six Sigma