Strategic Management of Knowledge in Big Science Agust Canals KIMO - - PowerPoint PPT Presentation

Strategic Management of Knowledge in Big Science

Agustí Canals KIMO Research Group Universitat Oberta de Catalunya - Barcelona

1

Agenda

• 1. Big Science organizations
• 2. Strategic knowledge mapping in big science projects: a

methodology to identify and develop key strategic knowledge assets and explore their characteristics and relationships

• 3. Structure of interorganizational collaboration in

scientific projects: analysis of collaboration networks

• 4. The role of simulations as a coordination mechanism in

a big science project: simulations as dynamic boundary

• bjects

2

Big Science Organizations

3

Big science

n In many areas (genomics, high energy physics, climate sciences,

ecology, astronomy, nuclear fusion,…) scientific research has moved in the last decades from small or medium-sized experiments to large and complex collaborations (Galison 1992)

n The idea of ‘big science’ put forward in the 1960’s by Weinberg

(1961) and Price (1963) has become commonplace (Hicks & Katz 1996, Knorr-Cetina 1999, Etzkowitz & Kemelgor 1999)

n Big science is taking an important part of research funding and it is

worth looking at its different aspects

n Big science experiments provide very interesting management and

• rganizational insights

n A good example: CERN experiments

4

The Large Hadron Collider (LHC)

5

ATLAS: One of the LHC detectors

6

The ATLAS detector

7

The ATLAS Collaboration

8

A complex organization

9

3000 physicists 174 universities and labs 38 countries

New kinds of organizations

n New virtual collaborations fostered by globalization and ICTs n But managed in a traditional way: organizational authority systems

and clear boundaries

n Some recent developments challenge this: distributed, non-

hierarchical networks such as Linux

n Questions:

n How is coordination actually achieved? n What happens when the task is complex and boundaries are fuzzy? n What level of complexity such networks can manage?

n The ATLAS case: bottom-up culture and very limited use of

managerial authority

10

Three Questions

¡ ATLAS is an exceptional knowledge-based organization! How does it work?

n What are the critical knowledge assets that allow ATLAS

to perform at such high levels?

n How is the structure of internal collaboration? n How is coordination achieved in this complex, non-

hierarchical knowledge system?

11

Strategic Knowledge Mapping

12

Three kinds of knowledge

13

Experiential Knowledge Ø What can I sense? Abstract Symbolic Knowledge Ø What can I extract from it which is stable or durable? Narrative Knowledge Ø What can I say about it?

Structured (Codified and/or Abstract) Unstructured (Uncodified and/or Concrete)

13

Diffused Undiffused

Structuring Information Sharing Information

The I-Space

14

Structured Unstructured

Zen Master Bond Traders

Boisot (1998). Knowledge Assets. OUP.

14

Undiffused Diffused

Knowledge in the I-Space

15

Public knowledge Conventional Wisdom Personal Knowledge Proprietary Knowledge

Structured Unstructured

15

Diffusion Absorption Scanning Problem-solving

The Social Learning Cycle (SLC)

16

Undiffused Diffused Structured Unstructured

Public knowledge Conventional Wisdom Personal Knowledge Proprietary Knowledge

16

Portfolio of knowledge assets

17

UTILITY

Structured Unstructured Diffused Undiffused

SCARCITY

17

Mapping the ATLAS knowledge

18

? ? ? ? ? ? ? UTILITY

Structured Unstructured Diffused Undiffused

SCARCITY ? ?

18

Strategic Knowledge Mapping Process

• 1. What are the organization’s critical performance

dimensions?

• 2. What are the knowledge assets that support

those performance dimensions?

• 3. Where are the knowledge assets located in the 


I-Space?

• 4. What are the strategic implications of the

knowledge map?

• 5. How can the knowledge system develop?

19

Selec&ng ¡knowledge ¡assets ¡

TDAQ Questionnaire: Basic Statistics

GENERAL SURVEY COMPARISON STATISTICS ¡ ¡ ¡ ¡ First Round ¡ Second Round ¡ Both Rounds ¡ Number of people approached ¡ ¡ 74 ¡ ¡ 101 ¡ ¡ 175 ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ Questionnaire hits ¡ ¡ 43 ¡ 58.11% ¡ 89 ¡ 88.12% ¡ 132 ¡ 75.43% ¡ Responses ¡ ¡ 41 ¡ 55.41% ¡ 49 ¡ 48.51% ¡ 90 ¡ 51.43% ¡ Complete responses ¡ 36 ¡ 48.65% ¡ 38 ¡ 37.62% ¡ 74 ¡ 42.29% ¡ Knowledge ¡responses ¡ 82 ¡ 81 163 ¡

21

TDAQ ¡Knowledge ¡Map ¡

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 Codification Diffusion ¡within ¡ATLAS

1 Standard Model 2 Beyond Standard Model 3 General P-P Collision 4 Overall view of the state of the art of electronics 5 FPGA and DSP 6 Hardware 7 Operating Systems 8 Software analysis and design 9 Programming 10 Database technologies 11 Networking and Point-to-point links 12 Project Management 13 People Management 14 Interpersonal communication skills 15 Detector readout and instrumentation 16 LHC machine parameters 17 MC simulation 18 Overview of the ATLAS experiment 22

What is the most salient knowledge? ¡

Programming (37) Standard ¡Model (11) Software ¡analysis ¡ and ¡design ¡(23) Detector ¡readout ¡ and ¡instrumentation (9) Overview ¡of ¡the ¡ ATLAS ¡experiment (15) Project ¡Management (8) Interpersonal ¡ communication ¡skills (10) Operating ¡Systems (8) 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00 6.50 7.00 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 Codification ¡ Diffusion ¡within ¡ATLAS

23

So9/Management ¡Skills ¡in ¡ATLAS

Programming Standard ¡Model Software ¡analysis ¡ and ¡design Detector ¡readout ¡ and ¡instrumentation Overview ¡of ¡the ¡ ATLAS ¡experiment Project ¡Management Interpersonal ¡ communication ¡skills Operating ¡Systems 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00 6.50 7.00 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 Codification ¡ Diffusion ¡within ¡ATLAS

24

Challenges ¡for ¡ATLAS ¡ ¡

n Strategically developing value (competitive advantage)

• Understanding the nature of one’s core competences
• Over and beyond the ATLAS project cycle (15 years)

n Fostering the further development of soft skills in ATLAS?

• Manpower development in High Energy Physics
• Formal courses (upper I-Space)
• Apprenticeships (lower I-Space)
• Correlation between position and choice of soft-skills?

n Managing the flow of people in and out of projects and

between home institutions and ATLAS

• Knowledge walking out of the door

25

Structure of Interorganizational Collaboration

26

Scientific collaboration

n Scientific collaboration has a direct effect on the impact of the

resulting publications (Benavent-Pérez et al. 2012), accentuated in the case of international collaboration (Kronegger et al. 2011)

n Important public funding is applied to scientific collaboration n It can be analyzed from different perspectives: authors, institutions,

countries (Sonnenwald 2007)

n In order to analyze it, scientific collaboration must be

contextualized: by discipline, by geographical area, by type of research, … (Gzani, Sugimoto & Didegah 2012)

n We are interested in understanding collaboration patterns in ‘big

science’

27

Studying scientific collaboration

n Usual methodology: co-authorship networks

(Sonnenwald 2007)

n … but in big science co-authorship networks of

published papers might be misleading

28

In big science: genomics

29

In big science: H.E.Physics

30

Collaboration in Physics

n Most of studies look at the institutional level n High degree of inter-institutional (~ 50%) and international (~ 30%)

collaboration (Gazni et al. 2012, Benavent-Pérez et al. 2012)

n Higher degree of international collaboration (especially in Europe)

and influence of geographical distance

n In a longitudinal analysis, Lorigo & Pellacini (2007) observe:

n An increase in the number of inter-institutional collaborations n An increase in the strength of inter-institutional collaborations (number

• f papers)

n An increase in the percentatge of nodes belonging to the largest

connected component

n Loss of centrality of CERN as an institutional node

n As Huang et al. (2012) suggest, collaboration networks like CERN

need to be studied in depth

31

Research design

n Access to internal ATLAS data

n Preprints database of the physical analysis phase

(with editors)

n Authors list with institutions

n Data (until 31/12/2012):

n 371 papers n 1543 authors n 217 institutes

n Co-authorship network analysis at the

institutional level

32

Co-authorship network

33

140 vertices 1073 edges density = 0.11 avg degree = 15.33

• clust. coef. = 0.39

1 2 5 10 20 50 0.01 0.05 0.50 degree k P(x) >= k

Community analysis

34

Adelaide Albany Alberta Annecy LAPP Argonne Arizona Arlington UT Athens Austin Barcelona Beijing Bergen Berkeley LBNL Berlin HU Birmingham Bologna Bonn Boston Brandeis Bratislava Brookhaven BNL Buenos Aires CERN Cambridge Cape Town Carleton Chicago Clermont - Ferrand Columbia Copenhagen NBI Cosenza Cracow AGH-UST Cracow IFJ PAN DESY Dallas SMU Dallas UT Dresden Duke Edinburgh Freiburg Geneva Genova Glasgow Goettingen Grenoble LPSC Harvard Hefei Heidelberg KIP Heidelberg PI Indiana Iowa State JINR Dubna KEK LPNHE-Paris La Plata Lancaster Liverpool Ljubljana London QMUL London RHBNC London UC Louisiana Tech Lund Madrid UA Mainz Manchester Marseille CPPM Massachusetts McGill Melbourne Michigan Michigan SU Milano Minsk AC Montreal Moscow MEPhI Moscow SU Munich LMU Munich MPI NYU New York Nagoya Napoli New Mexico Nijmegen Nikhef Northern Illinois Novosibirsk BINP Ohio SU Oklahoma Oklahoma SU Oregon Orsay LAL Osaka Oxford Pavia Pennsylvania Pisa Pittsburgh Portugal LIP Prague AS Prague CU Protvino IHEP RAL Roma I Roma II Roma Tre SLAC Saclay CEA Santa Cruz UC Seattle Washington Sheffield Siegen Simon Fraser Burnaby Stockholm Stockholm KTH Stony Brook Sussex TRIUMF Taipei AS Technion Haifa Tel-Aviv Thessaloniki Tokyo ICEPP Toronto Trieste ICTP Tufts UC Irvine UI Urbana Udine Uppsala Valencia Valparaiso Vancouver UBC Victoria Weizmann Rehovot Wisconsin Witwatersrand Wuppertal Yale York

Modularity method (Greedy algorithm) 9 communities

Discussion and conclusions

n Interesting findings

n High degree of collaboration n Not a scale free network, as opposite to the co-authorship

network of published articles (Newman 2001)

n Apparently no effect of geographical distance

n Conclusions

n Big science collaborations have an internal structure,

sometimes different from the rest

n In spite of the “one case” limitation, we may conclude that in

disciplines where big science has become important, traditional co-authorship analysis should be taken with care when studying scientific collaboration

35

Simulations as Boundary Objects

36

Coordination

n Cooperation needs transactions, that present some problems

(bounded rationality, information asymmetries)

n Traditional solution: Management (“visible hand”) through

hierarchical control

n Alternatives:

n Routines and rules: only effective under conditions of repetition n When all members agree upon the goals of the organization

and the techniques for achieving these goals are within the ability of all members, few or no rules are required: small

• rganizations oriented around expressive needs

n Under certain circumstances, the latter can apply to fairly large and

geographically scattered organizations like ATLAS

37

The ATLAS Puzzle

n A complex task n A project-oriented structure n A complex organization

n 3000 physicists n 175 universities and laboratories n 38 countries

n A non-hierarchical organization

n Held together by Memoranda of Understanding n Decision making is bottom-up n Decision making is distributed

Group 2 Group 6 Group 3 Group 1 Group 7 Group 4 Group 5

38

ATLAS in the I-Space

Diffused Undiffused Codified and Abstract

Bureaucracies

Clans

Fiefs

Markets

Uncodified and Concrete

ATLAS

Adhocracy

39

Boundary objects

n Boundary objects (Star 1989, Carlile 2002, 2004) act as a

scaffolding that enables people to:

n Gradually build up a shared understanding of common tasks facilitating

knowledge flows

n Provide coherence across intersecting social groups

n Examples of boundary objects: blueprints, maps, common

interests, rules, plans, conceptual frameworks. Feynman diagrams

40

Research design

n Case study developed between March and December of 2009 n Part of a wider investigation about different aspects of knowledge

creation, transfer and use within the ATLAS Collaboration

n Data collected through 30 semi-structured interviews to members

• f the Collaboration (9 senior members and 21 group leaders)

n Complemented with archival information from the ATLAS

Collaboration and participating observation

41

Key role of simulations

n Monte Carlo simulation

techniques

n Co-evolution of prototypes

and simulation in the design phase

n Necessary to interpret the

results in the operation phase

42

Simulations as boundary objects

The beginning of this experiment was a simulation. You simulate the whole experiment first until you’re confident that all the bits and pieces which you imagine… all the different things you imagine you put them into the simulation and see how they perform. So you’re really evolving two objects. You’re evolving a virtual object and you’re evolving a real object. […] And both are equally complex. The one

• n the computer may even be more complex because it contains all the detail.

[That core simulation is an object…] Not only to co-ordinate but to feed everybody with all the necessary information that the person needs in

• rder to perform within a complex…

[…in bio-technology you’ve got lots of prejudices that compete with each other with people having different ways of doing…] Yeah, yeah. Well here also, but here you use the simulation to iron them out.

43

Some insights

n Clans are governed by the intangible hand of trust and mutual

esteem, what requires personalized interaction and, therefore, are limited in size, but the ATLAS case suggests that clans can be expanded through the use of external scaffolding acting as boundary

• bjects

n Simulation absorbs complexity by capturing it in a “black box” and

behaves as a boundary object that facilitates alignment between groups

n The needed coordination is provided by culture and boundary objects:

the “intangible hand”

n Main implication: In cases of task complexity, boundary objects

together with clan or adhocratic cultures may substitute for the traditional coordination mechanisms

44

Conclusions and implications

n Is the ATLAS case unique? n The ATLAS culture produced an ‘organized anarchy’ that

works

n In The rise of the creative class, Florida (2002) suggests

that this kind of organizations are set to grow

n The ATLAS case suggests that they may be not

necessarily small scale organizations with few coordination problems, but also larger and more focused

• rganizations

45

Thank you!

46

Agustí Canals

acanalsp@uoc.edu

Questions?