Frontier research and new institutions for European science
Andrea Bonaccorsi
University of Pisa
Member of the High Level Expert Group onMaximizing the wider benefit of basic research and the
European Research CouncilEuropean Commission, DG Research
CRUI- University of Genua ConferenceApril 19, 2005
Outline
The new scientific landscape: the emergence of new leading sciences
The performance of European science in new leading sciences
The European Research Council (ERC) as a major step in institution building
The new scientific landscape
A new scientific landscape has taken shape in the last 20 years or so. It results from the combination of several revolutionary advances:
- the molecular biology revolution, particularly after the recombinant DNA discovery and the development of PCR and more generally life sciences and bioengineering;
- the pervasive information technology revolution, resulting from advances in algorithms, computer science, microelectronics, and more recently from the convergence with telecommunication;
- new advances in materials science;- new opportunities in nanotechnology,
particularly after the invention of STM.
The new scientific landscape
These new fields and disciplines share some intrinsic characteristics:
- are based on reductionist explanation strategies, but deal with complex systems at various levels of resolution;- evolve through a complex interaction between scientific understanding and engineering manipulation, i.e. between sciences of nature and sciences of artificial;- include many general purpose technologies;- cut across disciplinary boundaries and actively promote overlapping.
Search regimes
Rate of growth
- fast growing vs slow growing
Degree of diversity
- convergent dynamics vs divergent dynamics
Level of complementarity- physical infrastructure vs. human capital and institutional complementarity
New leading sciences (materials science, life sciences, computer science, incl. biotech and nanotech): fast growing, divergent dynamics, human capital and institutional complementarityHLEG: definition of frontier research.
Rate of growth
How do new fields of research are generated within disciplines?
Which is the post-entry dynamics of growth? Which is the steady state rate of growth?
We study the entry of new words in scientific publications:
- post-entry dynamics
- arrival process within the scientific discipline and turnover ratio (new words/existing words)
Evidence from Nanopublications
0
20000
40000
60000
80000
100000
120000
140000
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Cumulate arrivial of publications Cumulate entry of authors Cumulate entry of aff iliations
Source: Bonaccorsi and Thoma (2005)
Genetic Algorithm
0
100
200
300
400
500
600
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Number of occurrences of the word “Genetic algorithm” in the publications of the top 1000 scientists in computer science
Neural Network
0
200
400
600
800
1000
1200
1400
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Number of occurrences of the word “Neural network” in the publications of the top 1000 scientists in computer science
Wireless
0
50
100
150
200
250
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Number of occurrences of the word “Wireless” in the publications of the top 1000 scientists in computer science
Atomic Force Microscopy
0
50
100
150
200
250
300
350
400
450
500
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Number of occurrences of the word “Atomic force microscope” in the publications of the top 1000 scientists in high energy physics
Hadron Collider
0
50
100
150
200
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Number of occurrences of the word “Hadron collider” in the publications of the top 1000 scientists in high energy physics
0
5000
10000
15000
20000
25000
30000
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Total number New born
Total number of keywords and number of newly-appearing keywords in publications of top 1000 high energy physicists
0
1000
2000
3000
4000
5000
6000
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Total number New born
Total number of keywords and number of newly-appearing keywords in publications of top 1000 computer scientists
0,00
10,00
20,00
30,00
40,00
50,00
60,00
70,00
80,00
90,00
100,00
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Computer science High energy physics
Ratio between newly appearing keywords and total number of keywords in high energy physics and computer science
Stylized evidence on rates of growth
First, scientific fields grow at very different rates after entry. As a first broad distinction, there are fields that grow extremely rapidly and fields characterized by slow growth after entry. Post-entry growth rates sharply differ.
Second, disciplines largely differ in the composition of fields characterized by different rates of growth. In some disciplines it seems that new fields are generated continuously, so that the turnover ratio is extremely high, while in other disciplines the turnover is much lower.
Degree of diversity
How many different directions does search take?
Even within the same paradigm and theory, research programmes may differ by:- specific hypotheses- experimental technique- object or locus of observation
A dynamics of increasing diversity may be defined divergent.Divergence may be:- strong (competing, non compatible hypotheses)- weak or complementary (mutually compatible hypotheses but diverging search strategies)
Convergent dynamicsdiversity is stable or tends to decreasemost research programmes follow the same set of specific hypotheses and use the same tools
e.g. high energy physics, nuclear physics, astronomy, traditional chemistry, nuclear technology, aerospace, TLC, conventional engineering
Divergent dynamicsdiversity explodesthe same theory or paradigm (e.g. molecular biology) gives origin to many competing or just diverse programmes
e.g. HIV, Alzheimer, molecular oncology, nanotechnology, computer languages, computational chemistry
We study the concentration of keywords used in scientific publications
- highly concentrated disciplines: a few keywords absorb a large share of publications- research programmes tend to converge along the same directions
- highly fragmented disciplines: there are many directions of research, no dominant pattern.
Future research: mapping/ clustering of keywords over time
Concentration of keywords in publications of top 1000 scientists in Computer science and High energy physics
Computer High
science energyphysics
Number of publications of top 1,000 scientists 9,062 41,770
Number of publications with keywords 6,401 34,379
Publications with keywords (%) 71% 82%
Number of different keywords 18,031 50,952
Average number of keywords per author 5.35 5.44
Concentration ratio (C250)* 26.5% 29.3%
* Cumulative market share of top 250 keywords (Number of occurrences of the top 250 keywords/ total number of occurrences in all publications)
0
20
40
60
80
100
120
Top 250 keywords
Rela
tive f
req
uen
cy (
% o
f to
p k
eyw
ord
)
Highenergyphysics
Computerscience
Relative frequency of top keywords in High energy physics and Computer science
0
50
100
150
200
250
300
0 50 100 150 200 250 300
Rank in period 1991-1995
Ra
nk
in p
eri
od
19
96
-20
00
Plot of rank correlation of top 250 keywords in High energy physics (r=.79)
0
50
100
150
200
250
300
0 50 100 150 200 250 300
Rank in period 1991-1995
Ra
nk
in p
eri
od
19
96
-20
00
Plot of rank correlation of top 250 keywords in Computer science (r= .49)
Level of complementarity
Traditional type of complementarity in science:- physical facilities (e.g. big science)
New forms of complementarity:
- human capital complementarity (different disciplinary background, need for flexibility in education, career, affiliation, organizational setting)
- institutional complementarity (different institutions involved, e.g. university/laboratory/ hospital in molecular medicine)
The performance of Europe in new leading sciences
(a) European science is strong in fields characterized by convergent search regimes and weak in fields characterized by divergent search regimes.
(b European science is strong in fields characterized by high levels of infrastructural complementarities while it is much less prepared in fields characterized by human capital and institutional complementarities.
(c) Consequently, European science is strong in fields characterized by slow growth and weak in fields characterized by turbulent growth.
(d) European science is only quantitatively comparable to US science but is weaker in the overall quality and is severely under-represented in the upper tail of scientific quality.
Specialisation patterns (Revealed Comparative Advantages, 1981-1994)
no European country is specialised in Computer scienceno European country is specialised in Engineering;in biology and biochemistry small European countries (Netherlands, Sweden, Denmark, Norway, Finland) exhibit strong specialisation while large countries have an index lower than unity;in molecular biology several large countries (United Kingdom, Germany and France) and small countries (Netherlands, Finland, in addition to Switzerland) are specialised;Europe as a whole is specialised in a few biomedical areas (pharmacology, immunology, microbiology) and in the large traditional disciplines of chemistry, physics and astronomy.
In materials science EU-15 produce 40,108 papers and receive 83,748 citations, while NAFTA produce 31,620 papers but receive 106,841 citations
In the life sciences EU-15 produce 616,212 papers and US 529,608 in the period 1995-1999, but the citation impact (1993-1999) is 1.35 in USA and only 0.90 in EU-15
In computer science the citation impact (1993-1999) is 1.33 for Israel, 1.17 for US, but only in the range between 0.81 (Germany) and 0.95 (Italy) for the four largest countries
Source: Third European Report on S&T Indicators (2003)
What do these disciplines have in common?
A divergent search regime
a dynamics of proliferating research programmes,often generated within the same paradigm,that increase the diversity of the field in terms of hypotheses, experimental techniques, objects of investigation.
European science is strong in fields characterized by convergent search regimes and weak in fields characterized by divergent search regimes
What do these disciplines have in common?
A search regime characterized by new forms of complementarities
Not much physical infrastructure complementarity (big science)
But: - human capital complementarity
- institutional complementarity
European science is strong in fields characterized by high levels of infrastructural complementarities while it is much less prepared in fields characterized by human capital and institutional complementarities.
European science has developed separate institutions at national, intergovernmental and European level, for dealing with search regimes with strong physical infrastructure complementarities
(e.g. high energy physics, astronomy, space research, oceanography, nuclear technology).
It is much more difficult to provide emerging fields the required complementarities in terms of human capital within the common institutional framework.
There are few rapid growth mechanisms.
Rate of growth of broad disciplines
over the period 1995-1999 the fastest growing area has been computer science with a growth rate of almost 10% earth sciences, engineering and mathematics also show high growth rates, varying between 4.2 and 4.6% biology and agriculture have the lowest growth rates with 1.4 and 1.6% respectively the broad field of life sciences as a whole experienced a growth rate of 2.33% the growth rate for the broad field of engineering was 4.5%, of which 35% was materials science, that grew at 1.9% per year.European science is strong in fields characterized by slow growth and weak in fields characterized by turbulent growth
Upper tail in quality of research. Piece of evidence # 1
Data on the most cited scientists worldwide have been recently made available by ISI on the basis of the analysis of 19 million papers in the period 1981-1999, authored by 5 million scientists. They refer to around 5,000 scientists worldwide in all fields, selected as those 250 that receive the largest number of total citations in any subject
area (0.1% of the total). In all 21 fields US scientists largely dominate, with a proportion of highly cited scientists ranging from 40% in pharmacology and agricultural sciences to over 90% in economics/business and social sciences and an average around 60-70% of the total. Among the 21 areas, only in other three areas non-US countries represent more than 40% of the total: physics, chemistry and plant and animal science (see Basu, 2004).
Countrywise distribution of Highly Cited Scientists
0% 20% 40% 60% 80% 100%
Mathematics
Physics
Geosciences
Space Sciences
Materials Science
Engineering
Computer Science
Chemistry
Pharmacology
Biology & BioChemistry
Plant & Animal Science
Molecular Biology & Genetics
Microbiology
Immunology
Clinical medicine
Psychology/ Psychiatry
Neuroscience
Ecology/Environment
Agricultural Sciences
Social Sciences
Economics/Business
US
UK
Germany
Japan
Canada
France
Australia
Switzerland
Netherlands
Italy
Sweden
Israel
Belgium
Denmark
New Zealand
Spain
Austria
PR China
India
Finland
Norway
S. Africa
Russia
Taiwan
US scientists dominate in each of the 21 subject areas of science
USA
(Source:Basu, 2004)
Piece of evidence # 2
We examined (with non-ISI sources) the publications of top 1,000 scientists by citations received along all their scientific career in
- Computer science- High energy physics
and all publications in nanotechnology for the period 1990-2001 (ISI source).We identified the most productive institutions in terms of total number of publications in the period and ranked the first 100.
Share in the list of top 100 affiliations
0
10
20
30
40
50
60
70
80
Computer science High energyphysics
Nanotechnology
Scientific field
%
NAFTA
Europe
East Asia
MIT University of California University of CaliforniaUniversity of California MIT Stanford University
Indian Institute of Technology Stanford University MIT
National Taiwan University Harvard University Harvard UniversityHarvard University University of Massachusetts University of Illinois
Cambridge University Cornell University Carnegie-Mellon University
Yale University Carnegie-Mellon University Cornell UniversityUniversity of Michigan University of Illinois University of Michigan
Seoul National University Purdue University University of WisconsinCalifornia Institute of University of Michigan University of TexasTechnology
Bachelor Master PhD
Piece of evidence # 3
Institutions awarding degrees of the top 1,000 scientists in Computer science. Top 10 list
The performance of European science• Europe (possibly with the exception of UK and
Scandinavian countries) has problems in matching a rapid quantitative growth with adequate quality and the ability to dominate the upper tail of scientific reputation
• These problems largely come from the mismatch between new leading sciences and the prevailing institutional setting in most European countries
• The European institutional setting:• exhibits weaker selection properties• has less flexibility• has few mechanisms for rapid massive
growth• encourages lower mobility
• Under conditions of rapid and divergent growth, opportunity costs for scientists strongly increase.
TPERS/INSTAG
543210
T_PER
S
120
100
80
60
40
20
0
Plot of rate of growth (average number of personnel per each year of life,
T_PERS/INSTAG) against size (number of personnel, T_PERS). CNR 1957-1997
Source: Bonaccorsi and Daraio (2003)
TRES/INSTAG
3.02.52.01.51.0.50.0
T_R
ES
50
40
30
20
10
0
Plot of rate of growth (average number of researchers per each year of life,
T_RES/INSTAG) against size (number of researchers, T_RES). CNR 1957-1997
Source: Bonaccorsi and Daraio (2003)
European Research Council
1. Based on a grant mechanism (portability): effect on scientific competition and researchers mobility across Europe
2. Purely merit-based: effect (at the margin) on national, collusive scientific systems
3. Encourages talented junior researchers to develop their research more rapidly
4. May attract foreign researchers