Diapositiva 1

1
PRIME Winter School on Emerging
nanotechnologies Grenoble, 4-8 February
2008 The dynamics of science in the nano field
and the changing relation between discovery and
invention Andrea Bonaccorsi, University of Pisa
2

• Outline
• Part 1
• Stylized facts about the dynamics of science and
  technology in nano
• Very high rate of growth
• Sustained generation of novelty, or proliferation
• Institutional complementarity
• 2. Possible explanations
• Part 2
• 3. How scientific discovery influences the
  emergence and consolidation of industry a
  journey into the function space

3
Source Bonaccorsi and Thoma (2007)
4
Composition of keywords used in nano scientific
publications
5
Ratio Between Number of New Keywords Entered and
Total Number of Keywords Used by Year
6

• Measuring the dynamics of proliferation
• Experimental work
• New measure of variety derived from bipartite
  graph theory
• Dataset
• gt100,000 publications in Nano ST
• query from ISI Fraunhofer Karlsruhe expert
  selection
• part of a larger data construction exercise
  (PRIME project)
• period 1990-2001
• extraction of all keywords
• isolation of new keywords per year of birth
• References
• Bonaccorsi Thoma, Research Policy, 2007
• Bonaccorsi Vargas, in progress

7

• New keywords
• Even limited to a field, new keywords have
  multiple meanings.
• True scientific novelty
• Scientists feel the need to introduce new verbal
  forms, new definitions or new abbreviations in
  order to describe the object of their discovery.
• 2. Strategic differentiation by scientists
• They try to establish their own terminology in
  order to gain visibility and recognition.
• 3. Labelling process
• Definitions already existing in other disciplines
  are relabelled when a migration to a new
  discipline is carried out.
• Solution let new keywords spring, let the
  scientific community select.
• We consider only surviving new keywords, with
  occurrence gt 2.

8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
Limit case 1 all new articles share the same set
of keywords gt no correlation between number of
articles and degree. Limit case 2 all articles
share only one keyword perfect correlation
between number of articles and degree.
17
Class A
18
(No Transcript)
19

• Proliferation pattern
• Large cognitive distance between senior
  scientists and junior scientists- doctoral
  education based on competitive allocation of
  resources and proposals
• Impossibility to decide research directions in a
  centralized way- multilayer governance and
  funding systems
• Need to finance competing research projects-
  variety of funding sources (but mainly
  grant-like)
• Need to mobilize research projects in parallel-
  well developed post doc system with possibility
  to apply as principal investigators
• Strong epistemic uncertainty- premium given to
  top quality universities (signaling effect)

20
Composition of communities of inventors in
nanotechnology
21
Entry of individuals by community
22
ST complementarities at the inventor level (a)
8706 Inventors
8706 Co-inventors
23
ST complementarities at the inventor level (b)
24
Distribution of patents by assignee type in
relation to community
25
Authors-inventors

• a combination of the other two communities.
• inventors benefit from at least one who has
  experience in publishing (currently or in the
  early career), while the others may well be pure
  industrial researchers.
• stronger complementarities in use of scientific
  and design knowledge with respect to only-authors
• highest level of information heterogeneity
  (Granovetter 1975 1985)
• presence of hubs individuals and snow ball
  effects in their carriers

26
Research hypotheses

• Proposition 1 The quality of Author-Inventors
  inventive activity in patents should be higher
  than in the other two communities.
• Proposition 2 The productivity of
  Author-Inventors inventive activity should be
  higher if counted by the number of the patents
  produced in the top percentiles of the
  distribution of patents/inventors.
• Proposition 3 Given their higher technological
  performance and combinatorial capabilities of
  Author-Inventors, we should observe more of them
  as founder of companies.

27
Research hypotheses

• Proposition 1 The quality of Author-Inventors
  inventive activity in patents should be higher
  than in the other two communities.
• Proposition 2 The productivity of
  Author-Inventors inventive activity should be
  higher if counted by the number of the patents
  produced in the top percentiles of the
  distribution of patents/inventors.
• Proposition 3 Given their higher technological
  performance and combinatorial capabilities of
  Author-Inventors, we should observe more of them
  as founder of companies.

28
OLS regression of the inventor type on the
multidimensional quality index
29
(No Transcript)
30
Research hypotheses

• Proposition 1 The quality of Author-Inventors
  inventive activity in patents should be higher
  than in the other two communities.
• Proposition 2 The productivity of
  Author-Inventors inventive activity should be
  higher if counted by the number of the patents
  produced in the top percentiles of the
  distribution of patents/inventors.
• Proposition 3 Given their higher technological
  performance and combinatorial capabilities of
  Author-Inventors, we should observe more of them
  as founder of companies.

31
Sources own elaboration
32
Research hypotheses

• Proposition 1 The quality of Author-Inventors
  inventive activity in patents should be higher
  than in the other two communities.
• Proposition 2 The productivity of
  Author-Inventors inventive activity should be
  higher if counted by the number of the patents
  produced in the top percentiles of the
  distribution of patents/inventors.
• Proposition 3 Given their higher technological
  performance and combinatorial capabilities of
  Author-Inventors, we should observe more of them
  as founder of companies.

33
Notes The normalization procedure takes into the
account the fact that the Authors-Inventor
community is larger than the other two
communities. The normalization ratio adjusts for
the size of the community.
34
In search of explanations for the observed
dynamics of growth Theoretical background on
variety/ diversity in science Professional norms
of science (Merton) Reduction of variety as a
result of internal professional dynamics
(Callon) Variety and diversity in science mainly
comes from paradigmatic uncertainty (Kuhn)
Within normal science there is limited variety
convergence towards a common framework
35
Dynamics of growth Notion of divergent dynamics
(or proliferation) Large intra-paradigmatic
diversity Research programmes sharing fundamental
explanations but diverging on lower level
hypotheses or experimental techniques/
objects Main reasons A. reductionism in
explanation vs systemic integration B.
observation vs manipulation C. new forms of
combination between science and engineering
36
Reductionism in explanation vs systemic
integration Experimental advancements make it
possible to explain phenomena by making reference
to variables at lower level of resolution of
matter (i.e. molecular and atomic
level). Reductionist approach explaining higher
levels of organization of matter using knowledge
of lower levels. One gene, one disease
programme. Interestingly, when applied to
complex objects or systems (i.e. proteins, or
cells, tissues, organs) and their behavior (e.g.
disease) the reductionist approach does not lead
to complete explanation- it does not reduce but
rather increases epistemic uncertainty.
37
HIV-1
38

• The case of HIV research
• The search for a causal explanation of the AIDS
  disease was solved with the discovery of HIV
  virus.
• However, this fundamental explanation (the cause
  of the diseases lies in the agent HIV) over
  which the scientific consensus was almost
  universal, did not reduce the uncertainty over
• the specific biochemical mechanisms of
  interaction of the virus with the cell
• the entry points of the virus in the cell
• the patterns of mutation of HIV, etc.
• The reductionist approach did not produce a
  reduction ad unum, but rather opened the way for
  a proliferation of diverse (even competing)
  sub-hypotheses.

39
A schematic representation of the degree of
uncertainty that exists in the underlying
mathematical equations describing various
phenomena relative to the intrinsic complexity of
the phenomena
Uncertainty about basic equations
Source Barrow (1998) after Ruelle
40

• Observation vs manipulation
• In the development of science there has always
  been separation between the level of resolution
  at which it could be possible at any date
• make predictions
• observe
• manipulate
• New experimental technologies, e.g.
• scanning tunneling microscope (1981)
• polymerase chain reaction (1985)
• atomic force microscope (1986)
• These technologies make it possible to manipulate
  before observing, or to observe before making
  predictions

41

• Science-driven engineering
• New sciences make it possible new combinations
  between scientific explanation (knowing the
  properties of nature) and engineering
  (manipulating nature for a purpose)
• New relations between natural and artificial,
  discovery and invention
• the fundamental properties of matter cannot be
  discovered unless a specific configuration is
  designed
• design is an artificial activity oriented
  towards goals
• design goals can be achieved following many
  possible directions (design theory)
• scientists become engineers
• Large scale exploration through manipulation

42

• Summing up
• Dynamics of science
• Nano ST characterized by a strong proliferation
  pattern
• New themes and issues (appr. by keywords) appear
  at a high rate
• The entry of new issues does not diminish over a
  decade
• Large part of entrant new keywords do not survive
• Surviving new keywords may grow very large
• The growth of knowledge takes place via a
  continuous creation of novelty, rather than
  mainly by problem solving and cumulative learning
  on already established problems
• What is the impact on industrial dynamics
  (emergence, growth, consolidation, competition)?

43

• Dynamics of science and industry evolution
• Nano ST is a field formed by several sub-fields
  characterized by different search regimes and
  prospective industrial dynamics (Nanoelectronics
  Nanobio
• Nanomaterials)
• In these areas knowledge proliferation is linked
  to the function space in different ways.
• Uncertainty on the function space depends mainly
  on
• physical properties of the object itself
• interaction between the nano-object and the
  context (e.g. human body, or conventional
  product)
• industrial procedures to obtain the desired
  properties
• Nanomaterials and nanoparticles
• i) Pure process technology (catalysis)
• ii) Enlarge functionalities of existing
  materials (glass, automotive, construction,
  cosmetics) via better performance of known
  activities (e.g. coatings, surface treatment)
• iii) Create totally new functionalities (with
  high uncertainty)
• Incumbents dominate in i) and ii) new entrants
  in iii)

44
(b) Nanobio i) Manipulation of biological
processes may lead to totally new
functionalities, which for the state being are
only expected/predicted at the level of the
interaction between the nano-object and the
body Vertical division of innovative labour New
entrants up to the stage of clinical testing
incumbents afterwards (c) Nanoeletronics i)
Pure process technology (litography) down to 32
nm/ 10 nm ii) Unknown interaction between
product design (physical phenomena at quantum
level) and production processes below 32 nm/10
nm Incumbents dominate in i) with entry and
survival based on investment strategic readiness
(IBM vs Motorola or Siemens) and strategic
alliances (epistemic continuity) Incumbents
still dominate in ii) but entry and survival will
be based on knowledge base. Division of labour
between fabless and foundries
45

• Dynamics of science and market shaping
• There are several mechanisms through which the
  enormous creation of novelty (proliferation
  pattern) is managed, uncertainty reduced, product
  design managed and production plans made
  feasible
• Roadmap (nanoelectronics)
• Regulation of product attributes, or standard
  (nanomaterials?)
• Regulation of product entry (pharma, bio)
• The way in which market shaping will be regulated
  will influence market structure and competition.
• The crucial point is the articulation between
  discovery of properties of matter and discovery
  of functions

46

• A formal definition of functions
• FUNCTIONS are expressed by verb-object forms,
  where the ACTIONS correspond to the verbs, and
  the FLOWS to the objects.
• An ACTION is an entity that evolves in time a
  given aspect (preferably measurable) of a flow
  (material, energy, signal).
• Example TO ABSORB
• - remove liquid (when intended as absorbing a
  liquid from a surface to clean it)
• - import liquid (when intended as absorbing a
  liquid in a porous material to wet it)
• - stabilize solid (to absorb a shock)
• - import energy (when an equipment absorbs
  electrical energy in order to function)
• - block energy/signal (to absorb a radiation or a
  sound wave)
• - capture gas (in the meaning of adsorbing
  chemically)

47

• FUNCTIONS are expressed by verb-object forms,
  where the ACTIONS correspond to the verbs, and
  the FLOWS to the objects.
• An ACTION is an entity that evolves in time a
  given aspect (preferably measurable) of a flow
  (material, energy, signal).
• Examples of broad classes
• store/supply (store, empty, supply, receive,
  hold, stop, release )
• connect (mix, switch, compare, arithmetic
  operations....)
• branch (separate, cut, count)
• channel (transmit, transport, convey)
• convert (convert, change state, sense,
  integrate, differentiate, process)

48
(No Transcript)
49
(No Transcript)
50
Levels of description of artifacts
Values Needs Functions
Attributes Specifications
Lifestyle Comfort Stay afloat Lenght 20
mt Drawings Autonomy Reliability Control
direction Weight 10 tons Calculations
Status Safety Get horizontal Composite stabil
ity
51
Levels of description of artifacts World of
users Values Needs Functions Attributes Technical
specifications World of technology
52

• Functional description (Polanyi, 1958)
• The function of heart is to pump blood.
• The function of a hammer is to push a nail in
  the wall.
• Functional descriptions cannot be reduced to
  structural descriptions.
• Functions are not needs. The latter are defined
  in discursive terms by human users of objects.
  Functions can be defined in physical terms, as
  behaviors under specified conditions, which are
  connected to a desired effect.

53

• Functional analysis
• The level of functions makes it possible to
  connect rigorously the narratives on the world of
  users (user analysis, etnographic research,
  cognitive ergonomy..) to the world of product
  design and technology
• So far, functional analysis has mainly been
  adopted to support incremental innovation (e.g.
  House of Quality, Quality Function Deployment,
  conjoint analysis).
• Recent developments allow the extension to the
  more fundamental problem of generating radically
  new solutions.

54

• Function space
• Let F be the function space, or the set of all
  possible functions. This space contains all
  possible behaviors of objects under any
  conceivable condition, subject to the constraints
  that they can be implemented in the physical
  world.
• This means that the function moving at speed
  higher than light is not (so far) included in
  the function space.
• Two main questions
• How is the function space structured?
• What is the relation between functions and
  structures?

55

• Structure of the function space
• The fundamental property of the function space
  is hierarchy.
• Functions can be hierarchically decomposed, down
  to elementary units. A functional hierarchy is an
  iterative decomposition of a high level or
  macro-function into more elementary
  sub-functions.
• The achievement of lower level functions is a
  requisite for the achievement of higher level
  ones.
• Sub-functions are either intrinsically related
  to the main function as necessary elements, or
  are generated by the extension of conditions for
  behavior.

56
A functional hierarchy of a power plant (portion).
From Kitamura, Kasai and Mizoguchi (2001)
57

• Relation between functions and structure
• Let S be the structure space, or the set of all
  possible structures.
• This includes all natural structures and all
  conceivable artifacts, the latter subject to the
  constraints that they do not violate physical
  laws. This means that the structure chocolate
  bar at 300 C or the structure an engine
  exhibiting motum perpetuum are not included in
  the structure space.
• Human beings have the distinctive ability to
  represent functions (perhaps large classes of
  functions) as independent from objects.
• Design is a many-to-many correspondence between
  the function and the structure space.

58
Function Structure space space
59
A design is a correspondenceor mapping from a
function onto a structure
60
Function Structure space space
F0
S0
61
How many different objects can implement the
function F0?How many different functions can be
implemented by the object, or structure, S0?
62
All functions can be implemented by a variety of
structures
63
All structures can implement a variety of
functions
64
Function Structure space space
Design as many-to-many correspondance
65

• Innovation as abduction
• The mapping does not have the property of
  isomorphism (Polya, 1954).
• Design problem solving is a form of abductive
  thinking, that does not move from antecedents to
  consequences, but makes the reverse path, from
  desired consequences (functions) to possible
  causal factors (structures).
• There is no logical necessity in abduction.
• In this perspective, design is not the object of
  logics, but rather of informal logic or
  semiotics. In fact, the way in which functions
  are projected onto physical structures resembles
  the process of attribution of meanings to signs,
  or semiosis (Eco, 1990).

66
 
Flow-block diagrams in materials design (steel)
From Olson (2002)
67
A few conjectures of the dynamics of the space of
functions C1. Mutual redundance Artifacts have
embedded more functions than intended. Functions
have more solutions than the currently used. The
mutual redundance acts as a dynamic generation of
novelty C2. Discovery of functions Functions are
discovered in artifacts through social processes
of use, in non-existing artifacts through
intentional search C3. Cycles Discovery of
functions through social use follows a cyclical
pattern of shift between satiation of functional
needs vs non functional needs C4. General
constraints Removing or relaxing constraints that
are high in the functional hierarchy generates a
cascade of innovations C5. Scaling Scale is a
general constraint due to law of gravity