Edward J. Hackett et al- Tokamaks and turbulence: research ensembles, policy and technoscientific...

8/3/2019 Edward J. Hackett et al- Tokamaks and turbulence: research ensembles, policy and technoscientific work

1/21

Research Policy 33 (2004) 747767

Tokamaks and turbulence: research ensembles, policyand technoscientific work

Edward J. Hackett, David Conz, John Parker, Jonathon Bashford, Susan DeLayDepartment of Sociology, Arizona State University, Tempe, AZ 85287-2101, USA

Received 1 May 2003; accepted 1 December 2003

Available online 20 February 2004

Abstract

A comparative analysis of twofusionenergy research facilities is used to examine how the ensemble of research technologies

(materials, methods, instruments, techniques, and the like) constructed and used by a group not only connects the group to other

researchers and policymakers but also influences the groups trajectory, performance, and the work of its members. We use a

combination of historical, interview, and questionnaire data to describe the two facilities, position them within the field, and

examine the working conditions and job satisfaction of their members. We develop the idea of research ensemble, characterize

it in comparison with related concepts, explain how it reflects policy priorities and provides a new way for research groups

to accumulate advantage and disadvantage. Using multiple regression models, we demonstrate how differences in research

ensembles lead to differences in working conditions and job satisfactions. Some implications are proposed for policy in

fast-changing, large-scale fields of science and technology.

2004 Elsevier B.V. All rights reserved.

Keywords: Research groups; Scientific collaboration; Work life; Research technologies; Science policy

The advantages of having fusion available as a

power source on Earth would in fact be so immense

that they would motivate all the necessary scien-

tific, technological and economic support required

for realization.

(Hans Wilhelmsson, 2000)

Fusion is the energy source of the future and alwayswill be.

(Anonymous)

Otto Neurath likened the work of scientists to that

of sailors who continually rebuild their ships amid

the turbulence and resource scarcity of the high seas

(Cartwright et al., 1996; Nowonty et al., 2001, p. 178).

It is an apt metaphor for the field of fusion energy

Corresponding author.

E-mail address: [email protected] (E.J. Hackett).

research, which encounters turbulence in the plasmas

it studies and in its policy and resource environments.

Fusion research begins with a vessel built to create,

contain, and diagnose plasmas, and substantial invest-

ments of government funds and researchers energies

are required to construct the vessel and diagnostic in-

struments. Just as the turbulent seas damage the ships

containing Neuraths sailors, the vessels containing

plasmas are damaged by their turbulent contents and

by the demands of their policy environments, and

thus require continual rebuilding. Fusion researchers,

much like sailors, risk damage to their vessels in the

ordinary course of their work by operating them at or

beyond their designed performance limits.1 But where

1 Just as the builders of medieval cathedrals learned from the

damage incurred by smaller churches, the building and breaking

0048-7333/$ see front matter 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.respol.2003.12.002


2/21

748 E.J. Hackett et al. / Research Policy 33 (2004) 747767

Neurath imagines lone scientists tossed in the waves,

fusion research is performed by a coordinated fleet

of vessels, each with a sizeable crew and each neces-

sarily distinct in some important ways yet similar inothers. Individual vessels in the fusion fleet are co-

ordinated by policy and programmatic decisions, and

depend upon one another intellectually, materially,

technologically, and socially.

We wish to understand what it is like to build and

rebuild the vessels of research, how their construction

takes account of the turbulent content and environ-

ment, how different sorts of vessels sail, and what it is

like to work within them. Our central idea is that the

ensemble of research technologies (materials, meth-

ods, instruments, established practices, and the like)

constructed and used by a group not only connects the

group to others in its field but also positions it on the

maps of politicians and policymakers (who participate

in the construction and direction of the ensemble) and

influences the groups performance and the work of its

members. We develop this idea using a comparative

case study of two fusion research groups.

1. Conceptual framework

Our conceptual framework combines ideas from

three distinct streams of literature. The first stream has

to do with the role of research technologies, which are

constructed by groups and in turn structure and con-

strain the groups activities. The second is concerned

with how science and technology policy influences re-

search through the amounts and purposes of support,

through specific direction of goals and objectives, and

through decisions about the research technologies. The

third literature treats research as a form of work, em-

phasizing the importance for knowledge workers of

autonomy, development opportunities, intrinsic quali-

ties (such as challenge and variety), and job satisfac-tion. Of particular concern to us is how the ensemble

of research technologies, co-produced by researchers

and policymakers, affords researchers the opportunity

to do certain sorts of research and not others, to de-

velop their ensemble in certain directions but not oth-

ers, and in those ways influences the quality of their

work lives.

of fusion machines teaches researchers what works and what does

not, and sometimes why (compare Knorr-Cetina, 1999, 36 pp.).

We blend these streams to view research groups as

socio-technical entities that integrate a social group

with an ensemble of research technologies (research

ensemble, for short) that are the groups means ofproduction, positioning it within the field, offering a

point of influence for research policy and a point of

contact for interaction with other groups. The research

ensemble shapes the life course of the group and the

work lives of its members through interactions with

other groups and with policymakers and through the

sorts of research performed.

1.1. Ensembles of research technologies

Research groups accomplish their work using an

arrangement of materials, techniques, instruments,

ideas and enabling theories that we call an ensem-

ble of research technologies. We prefer this term to

others (reviewed below) because it retains the in-

sight that doing research requires an integrated set

of tools, materials, and related ideas while removing

the restriction to experimental work implied by other

terms. Observational sciences and social sciences use

research technologies: remote sensing techniques,

econometric data and models, ethnographic protocols,

and conventions of survey sampling and measurement,

for example see Price (1984, p. 13). Incorporatingtechnology into the term also establishes a connec-

tion with studies of technology and work (Liker et al.,

1999) and invokes structuration theory to explain the

process through which the research ensemble is both

constructed by researchers and constrains their action

(e.g., Adler, 1992; Barley, 1986; Giddens, 1979).

Systematic attention to the dependence of science

on technology began with Derek Prices work on in-

strumentalities, which are techniques of science . . .

an understanding of the way to do things . . . that pro-

duce something new . . . . It will not do to call them in-struments [because the] term must allow us to include

parts of the experimental repertoire that are labeled

effects. . . [and] must also include chemical pro-

cesses, such as polymerization and Lowrys method

for protein determination, and biological processes,

such as recombinant DNA (1984, p. 13). Chandra

Mukerji (1989) emphasizes how a research groups

signature or identity shapes its research investments

and marks the path inquiry will take. For Franois

Jacob the choices a researcher makes have enduring


3/21

E.J. Hackett et al. / Research Policy 33 (2004) 747767 749

consequences: any study begins with the choice of

a system. On this choice depend the experimenters

freedom to manoeuvre, the nature of the questions he

is free to ask, and even often, the type of answer hecan obtain (Jacob, 1988, p. 234). Rheinberger ex-

tends the paradox of choices and ensuing constraints

in research, explaining that the more a scientist learns

to handle his/herexperimental system, the better it re-

alizes its own internal capacities. In a certain sense, it

becomes independent of the researchers wishes just

because he/she has shaped it with all possible skill

(Rheinberger, 1997, p. 305, emphasis original; see

also Creager, 2002). Over time incremental decisions

shape a genealogy of research ensembles that exert an

independent force on the groups research trajectory

(Ben-David and Collins, 1966; Knorr-Cetina, 1999).

Such ideas evolved mainly from studies of fields

where researchers act individually and almost inde-

pendently, where (at least initially) they have substan-

tial freedom of choice in the design and development

of their research ensembles, and where the goal is

to produce new knowledge. But for several reasons

researchers in fusion are unlikely to enjoy such lat-

itude in their choice of a research ensemble. First,

fusion research is literally technoscience: new tech-

nologies are constructed both to develop new energy

systems and to produce scientific knowledge (Latour,1987, pp. 174175). The two aims are linked in place,

personnel, and performance. Second, their research

ensembles are expensive, visible, and accountable to

other researchers and to the public, and so become

more tightly coupled to diverse communities. Third,

the technological dimension of technoscience entails

an array of objectives, milestones, and deliverables

that provide accountability and opportunities for guid-

ance by policymakers. Fourth, larger and more expen-

sive research ensemblesin the extreme, think of the

Superconducting Supercollider or the Hubble SpaceTelescopewill have longer gestation periods, greater

involvement of policymakers and the community, and

tighter coupling to predecessors and prototypes. Fifth,

when big technoscience is conducted at scattered sites,

the work will be centrally coordinated either actively

or implicitly by a division of tasks. Finally, big techno-

science is collaborative, so choices are made by and

influence groups.

Research ensembles are shaped by the epistemic

cultures of their field, a term Karin Knorr-Cetina

coined to represent the most prevalent amalgams of

arrangements and mechanisms . . . which, in a given

field, make up how we know what we know (1999,

p. 1; emphasis original). Within an epistemic culture,the ensemble of technologies a group (or laboratory,

in the quotation that follows) uses in its research is

created and developed through a process that takes

account of many elements of its environment:

Laboratories, to be sure, not only play upon the so-

cial and natural orders as they are experienced in

everyday life. They also play upon their own pre-

vious makeup and at times upon those of compet-

ing laboratories . . . . [O]ne can link laboratories

as relational units to at least three realities: to the

environment they reconfigure [through correspon-dence of lab conditions to those of the world outside,

through treatments and interventions that process

a partial version of the outside world, or through

representations that examine signatures of phenom-

ena], to the experimental work that goes on within

them and is fashioned in terms of those reconfig-

urations, and to the field of other units in which

laboratories and their features are situated.

Laboratories introduce and utilize specific differ-

ences between processes implemented in them and

processes in a scientific field . . . from which epis-

temic benefits can be derived . . . . We need to con-

ceive of laboratories as processes through which

reconfigurations are negotiated, implemented, su-

perseded, and replaced.

(Knorr-Cetina, 1999, pp. 4445; emphases original)

At the same time, however, the epistemic cul-

ture of fusion research is fused to a technogenic

(technology-building) culture: a culture that values

and guides the production of new technologies and

new physical phenomena. While the cultures are notopposed, they are often in tension. At times the epis-

temic is ascendant, and the production of new knowl-

edge takes precedence, at other times the technogenic

is dominant and the emphasis shifts to the devel-

opment of energy systems. Thus fusion researchers

experience an enduring form of sociological ambiva-

lence as they are caught between inconsistent values,

principles, and practices (Merton, 1973).

Researchers invest energy and expend resources on

their research ensembles under a variety of constraints,


4/21


including their own abilities, the possibilities afforded

by the ensemble of technologies currently in use, and

their relationships with other groups and their ensem-

bles. For fusion research and other such fields, addi-tional constraints are added by the demands of policy

and politics, perhaps conveyed through processes of

co-production of science and policy that take place

within boundary organizations (Guston, 2000, dis-

cussed below). Within those constraints the precise

ensemble of technologies a group uses is established

through the processes of alignment, negotiation, or

enrollmentprocesses at the heart of many studies of

laboratory work (Fujimura, 1996; Lynch, 1993; Law

and Hassard, 1999)with the aim of establishing both

specific similarities and specific differences with other

groups. In short, researchers build technologies, but

not in times and circumstances of their own choosing,

and those technologies not only make inquiry possible

but also shape and constrain their actions (cf., Adler,

1992; Barley, 1986).

Following Cook and Brown, we call the pos-

sibilities and constraints of research technologies

affordances, which are how a material, design, or

situation affords doing something and, by impli-

cation, does not afford doing something else (Cook

and Brown, 1999, p. 389). The interaction between

a research group and its ensemble of research tech-nologies dynamically affords both the acquisition

of knowledge and the use of knowledge once ac-

quired . . . doing epistemic work that the knowledge

alone cannot . . . . (Cook and Brown, 1999, p. 390;

emphases original).

For fusion research, producing knowledge now,

constructing the means to produce knowledge in the

future, and developing a machine that can produce us-

able energy all rely to some degree on the interactions

among researchers, policymakers, research ensembles,

and their affordances that take place within particu-lar epistemic and technogenic cultures. Viewed over

time, fusion research technologies appear in continual

flux, as diagnostics, plasma properties, and tokamak

designs are modified and sometimes transformed. Ma-

jor transformations produce genealogies of research

ensembles, some designated by letter suffixes to the

machines name, others by merely appending a U for

upgrade, while lesser innovations are taken in stride

(Ben-David and Collins, 1966; Knorr-Cetina, 1999).

Viewed synchronically across the research field, the

machines operating and in the design or construction

stages fit together in a particular way, each with its

distinctive contributions, affordances, and limitations

that reflect the guidance of the policy and politicalcommunities, the epistemic culture of the field, and

the preferences and judgments of researchers (Cook

and Brown, 1999, pp. 381400). Affordances, con-

nections between research ensembles within a field,

and arguments that position an ensemble with respect

to science and technology policy create epistemic

and technogenic commitments that limit the latitude

of a group to negotiate, align, enlist, and otherwise

maneuver to reposition itself advantageously.

Having established the centrality and character of

research ensembles we wish to take two further steps

with them: a step outward, to their connections with

policy and politics, and a step inward, to their conse-

quences for work within the groups. Moving outward,

studies of research ensembles have contributed much

to our understanding of the technical and material

dimensions of research, but say little about connec-

tions with the larger environment of research, policy

and politics. Mukerji (1989, pp. 132134), for exam-

ple, indicates that laboratories tout and tweak their

signature research approaches to jockey for position

with funding agencies, but she does not examine the

process or its consequences. Knorr-Cetinas (1999)account of the formation and functioning of epis-

temic cultures scarcely addresses policy at all, yet

the accelerators essential for high-energy physics are

fabulously expensive and totally dependent on gov-

ernment funding. Moving inward, research ensembles

are interposed between the daily work of a group and

the environment of research and policy; they embody

the essence of previous research and mark the path of

future inquiry. While studies of research ensembles

are all concerned with the laboratory practice and

knowledge production, none systematically examinesproperties of research work.

1.2. Research and policy

Where, how, and with what consequences (for re-

search and for researchers) do the worlds of policy

and research intersect? David Guston (2000) of-

fers a systematic framework for thinking about the

influence of policy on knowledge production in sci-

ence. He views science policy through the lens of


5/21


principal-agent theory, with the federal government

as the principal that engages its agentsscientists

and engineersto perform research that the govern-

ment cannot perform for itself. In this view, VannevarBushs social contract with science, which built

a boundary between politics and science through

which only money could pass in one direction and

science-based technology in the other, has been re-

placed by boundary organizations that engage sci-

entists with policymakers in the co-production . . .

of knowledge and social order (2000, pp. 58, 149).

Guston (2000, pp. 2627) offers a thoughtful distilla-

tion of constructivist thinking about science, but does

not connect the process of collaborative assurance,

which occurs within boundary organizations, to the

production of knowledge within groups.

The connection between policy and knowledge pro-

duction works in many ways, and for big science

the most powerful influence operates through the en-

semble of research technologies used by a group. Fu-

sion is certainly big science, and perhaps even

megascience, a term coined to characterize a re-

search program that addresses a set of scientific

problems of such significance, scope and complexity

as to require an unusually large-scale collaborative

effort, along with the facilities, instruments, human

resources, and logistic support needed to carry it out(OECD, 1993). Megascience projects are expensive,

with lifetime costs exceeding US$ 1 billion; they are

conducted in unique research facilities, too large and

expensive to duplicate; and they may be spatially

concentrated (as are traditional high-energy physics

projects) or distributed, with substantial central coor-

dination (OECD, 1993, p. 50; Sandstrom, et al., 1999,

pp. 1314). In megascience, as Robert Smith reports

in his case study of the Hubble Space Telescope, the

research ensemble reflects the power relationships

between the various institutions, groups, and individu-als engaged in policy-making . . . these activities [are

not] only political, involving negotiations and com-

promises among different groups; instead, they also

involved the hardware and design of the telescope as

well as its planned scientific objectives. (Smith in

Galison and Hevly, 1992: 191, 208).

Unlike the Hubble Space Telescope, fusion re-

search is conducted by a mosaic of devices and their

associated groups, mutually coordinated and centrally

directed. Using OECDs language, fusion is an ex-

pensive, unique, distributed megascience. In fusion,

researchers and policymakers co-produce research

ensembles, knowledge, and social order within bound-

ary organizations and research facilities, and thoseensembles of technologies have strong and enduring

consequences for research and researchers.

1.3. Research as a vocation

Max Weber eloquently dissected the external cir-

cumstances of scientific work and the inner life of

science in his lecture Science as a Vocation, and

his ideas have endured remarkably well (Weber, 1918

[1948]; Hackett, 1990). Science is one of the most

rewarding forms of worka vocation, for Weber, to

which few are called and those called become deeply

dedicated. The personal rewards of scientific work

derive from its distinctive qualities: autonomy, chal-

lenge, personal development, and the intrinsic value

of producing knowledge for its own sake. In Webers

haunting words, whoever lacks the capacity . . . [to

believe that] the fate of his [or her] soul depends upon

whether or not he makes the correct conjecture at this

passage of this manuscript may as well stay away

from science (1918 [1946], p. 135).

A small but persistent literature follows Webers

lead, treating research as a form of work and ar-guing that working conditions are not only instru-

mentally important for knowledge production but

also intrinsically important for researchers (e.g., Pelz

and Andrews, 1976; Andrews, 1979; Miller, 1986;

Tuttle et al., 1987; Hackett, 1990; Trankina, 1991;

Jones, 1996; Keller, 1997). Much of this research is

concerned with the Quality of Professional Life

(Miller, 1986) experienced by researchers, arguing

that autonomy, challenge, development, meaningful

work, and variety are important intrinsic characteris-

tics of work that increase the job involvement and jobsatisfaction of researchers, which in turn enhance job

performance (Jones, 1996; Cotgrove and Box, 1970;

Miller, 1986; Pelz and Andrews, 1976; Raelin, 1985;

Jones, 1996; Keller, 1997).

But studies in this genre usually treat scientists

apart from their group contexts (e.g., Trankina, 1991),

and even those concerned with research groups con-

sider them only as social entities, with no attention

to their ensembles of research technologies. Andrews

(1979), for example, shows that the motivation and


6/21


diversity of research units enhance their performance,

but gives no attention to the technologies groups use

to conduct research.2 We view researchers as mem-

bers of a group that uses a research ensemble to doits work, asking how characteristics of the research

ensemble influence work life.

1.4. Toward a synthesis

The social and technical arrangements of a research

group mediate the influence of the policy and re-

source environments on the research trajectories and

careers of its members. Groups build research ensem-

bles that embody aspects of their abilities and history,

and they do so in collaboration with other researchers

and with policymakers. Each research ensemble hasa distinctive orientation toward the epistemic and

technogenic cultures of its fieldaiming to develop

affordances that confer specific differences and spe-

cific similaritiesand for each group those cultures

appear substantially established and non-negotiable.

Collectively such decisions, constructions, and af-

fordances constitute the epistemic and technogenic

cultures of the field, and the accumulation of changes

in research ensembles will change those cultures.

Once built, a particular device presents an upgrade

pathway with limited possibilities. For any particulargroup, properties of its research ensemble will shape

its life chances and performance and the work lives

of its members. We will illustrate and develop these

ideas with a comparative case study of two groups in

fusion energy research.

1.5. Fusion as a research site

Fusion occurs when hydrogen isotopes (deuterium

or a mixture of deuterium and tritium) are heated

2

Even in the highly directive, bureaucratic environments ofbig technoscience, researchers find intrinsic satisfaction, as Lil-

lian Hoddeson observes in a paper examining Manhattan Project

research on the implosion trigger: intriguingly, it appears from in-

terviews as well as written sources that the scientists felt generally

fulfilled in their scientific study of the implosion. While working

on this strongly mission-directed problem, they experienced the

joy of research and the sense that they were working on their own

problem! The issue for historians to unravel is how it is possible

for a large laboratory to create an environment in which many or

most of its scientists can experience such a sense of free inquiry

while in fact they are working directly in line with the mission

(in Galison and Hevly, 1992, 286 pp.).

to the point of ionization (between 10 million and

100 million degrees), which separates electrons from

nuclei and allows the nuclei to be forced so close

together that they form a single helium nucleus, re-leasing energy. The hydrogen plasma created in a

tokamak is turbulent and short-lived, writhing within

a magnetic bottle for about a second before it is

extinguished by contamination (if it contacts the ves-

sel walls) or exhaustion (when the input power is

expended).

Fusion requires high temperatures (initially driven

by outside energy, then sustained by the fusion reac-

tion itself), high pressures, high densities of electrons

and energy, and containment to keep the plasma from

contacting any material surface. The work is collab-

orative and dependent on research funding to pay for

machine construction and maintenance, staff salaries,

and electricity. A small operation costs about US$ 12

million per year; a moderate-sized one (such as CTX)

runs US$ 35 million; a substantial one (such as MAT)

costs US$ 15 million a year, and the large facilities

cost three to five times that amount (for fiscal year

1995, in 1994 dollars; see Department of Energy Bud-

get, [http://www.ofe.doe.gov];).

Fusion energy research has several strategic advan-

tages for this study. First, it is a big science, with

research groups larger in membership and annualbudget than benchtop sciences, but much smaller in

those ways than high-energy physics. Since much

of what we know of big science is really about

high-energy physics (e.g., Traweek, 1988; Galison

and Hevly, 1992; Knorr-Cetina, 1999; but see Shrum

et al., 2001; Chompalov et al., 2002), the case by

itself has intrinsic value. Second, the field is inher-

ently collaborative, integrating a range of scientific

and engineering specialties, so it is advantageous for

studying group processes. Third, fusion has a history

of changing research priorities, funding levels, andcommitment to international collaboration, offering

a window on the effects of policy on the research

process.

The two academic tokamaks we studied, here called

CTX and MAT, are similar in size, age, organiza-

tion, performance, and university context but differ-

ent in their research ensembles and, consequently, in

their working conditions and fates. Both are relatively

large operations, with about 7080 researchers housed

within research institutes on the campuses of major
http://www.ofe.doe.gov/


7/21


research universities and additional off-campus col-

laborators.

2. Methods, data and measures

The study began in 1993 with a questionnaire survey

and face-to-face, taped, transcribed interviews with 15

researchers at CTX and MAT. We interviewed at ev-

ery level, from facility directors through faculty and

career researchers who led research teams, to team

members, postdocs, and graduate students. Question-

naires were administered at both facilities to measure

researchers social background characteristics, percep-

tions of the field and the groups place within it, work-

ing conditions, and job satisfactions. Response rateswere 42/81 (52%) at CTX and 38/77 (50%) at MAT.

Specific measures were taken, with appropriate revi-

sions, from the job diagnostic survey (Hackman and

Oldham, 1976), the Quality of Employment Survey

(Quinn and Staines, 1979), and organizational assess-

ment surveys (e.g., Mirvis et al., 1991). To understand

the policy environment for fusion energy research we

attended meetings of the Fusion Energy Science Ad-

visory Committee (FESAC) and the 2002 meeting of

the fusion community at Snowmass, Colorado, and ex-

amined policy documents, agency budgets, annual re-ports, proposals, news articles, and briefings prepared

by the Department of Energy. Publication and cita-

tion data were obtained from Science Citation Index

0

100

200

300

400

500

600

700

800

900

1957

1959

1961

1963

1965

1967

1969

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

Year

Constant(2000)

Dollars(Millions)

Source: Rowberg, 2000, Appendix

Fig. 1. Congressional funding for magnetic fusion R&D fiscal year 19572000.

(available online as Web of Science), using as search

terms the facility name and the keyword fusion as

elements of the articles title, combined with the ap-

propriate universitys name in the address field. Theyears 19842000 were chosen to provide a substantial

publication history before and after our fieldwork and

survey.

3. Elements of fusion history and policy

The possibilities of fusion were first glimpsed in

the early decades of the twentieth century, but the

field did not attract much attention until 1958, when

researchers from the Soviet Union presented results

from a magnetic fusion device at the second Atoms forPeace Conference (Bromberg, 1982; Fowler, 1997).

Over the decades US support for fusion research has

ebbed and flowed with national energy policy, and

programmatic emphases have changed with chang-

ing perceptions of scientific and technical possibili-

ties. Significant federal support began in the 1970s

and rose sharply with the energy crisis in the lat-

ter years of that decade (see Fig. 1). Through fiscal

year 2000 the US has invested about US$ 16 billion

(constant US$ 2000) in the field (Rowberg, 2000).

Early research involved the construction and operationof a variety of magnetic fusion devices: tokamaks (a

torus or doughnut-shaped device), spheromaks (more

like a doughnut hole than a doughnut), stellarators,


8/21


reverse-field pinches, magnetic mirrors, and others.

The aim of building such a variety of fusion devices

was to explore alternative architectures, energy induc-

tion strategies, and confinement regimes in hopes offinding one that could be developed from a research

and demonstration facility to a commercially viable

energy device.

The exploratory years of fusion energy research

ended in the early 1980s with redirection of research

from exploration to energy production, leading in the

latter years of that decade to an international collab-

orative program to build the International Thermonu-

clear Experimental Reactor (ITER; see Kay, 1992;

Fowler, 1997, pp. 113126). ITER, a greatly enlarged

version of the standard tokamak, would produce a

self-sustaining plasma and provide an engineering

test bed for designing components of a working fu-

sion energy plant. At the time of the initial interviews

and surveys (19931995), the US fusion program was

heavily invested in ITER, spending some US$ 63 mil-

lion (20%) of a fiscal year 1994 budget of about US$

328 million on the program (Department of Energy

Budget Request, fiscal year 1995). The commitment

to ITER in the Budget Request was unequivocal:

A broad international consensus now exists on how

ignition and burn of a magnetically confined fusion

plasma (i.e., the fusion fuel) can be achieved . . . .To date, the most effective way to confine a plasma

magnetically is to use a toroidal, or doughnut-shaped,

device called a tokamak . . . . During the next decade,

the program will focus on demonstrating the sci-

entific and technological feasibility of fusion in the

International Thermonuclear Experimental Reactor

project . . . . (Department of Energy Budget Request,

Fiscal Year 1995, p. 425).3 Note the specific com-

mitment to a particular research ensemble. Unsur-

prisingly, respondents explained to us that a facilitys

value to the national fusion energy research effortdepended upon its relevance to the development of

ITER, and took pains to explain their facilitys ITER

relevance..

3 ITER has had its ups and downs among policymakers: the

US pulled out of the collaboration in July 1998, at the specific

direction of Congress, and in January 2003 was making plans to

rejoin.

On the technical front, the quest for ignitiona

self-sustaining or burning plasmaslowed as con-

temporary machines were pushed beyond their design

specifications. While setting a record for energy out-put of 10.7 MW of fusion power (which has since

been surpassed), Princetons TFTR exceeded the

design specifications for its magnetic field by 8%

and for its neutral beam injectors (which heat the

plasma) by 20% (Glanz, 1994). This exceptional per-

formance extended the machines working life by a

couple of years, as funds were rearranged to exploit

this success. But the machine made its final plasma

in April 1997, and the last segment of its vacuum

vessel was taken off its pedestal on 26 March 2002.

TFTR did all it was designed to do and more, yet

when its work ended a burning plasma remained on

a distant horizon. Fusion energy policy does not ap-

ply diffuse pressures that emphasize some aspects of

research and development while de-emphasizing oth-

ers. Under policy pressure, fusion devices routinely

operate above their design limits and frequently un-

dertake technological upgrades, leading to significant

spells of planned and unplanned downtime, with pre-

dictable difficulties for the group. Policy change is

often accomplished by abruptly funding or defund-

ing particular activities, including entire facilities and

those they employ: the MFTF-B device at LawrenceLivermore Laboratory was mothballed on the very

day its construction was completed (Fowler, 1997,

p. 179).

This sketch of fusion science, technology, and pol-

icy suggests several preliminary generalizations. First,

during the 1990s the policy environment for fusion re-

search was unsettled, with the US first committing to

ITER as a priority that pervaded the field, with conse-

quences for researchers that will be described below,

then withdrawing from the collaboration in July 1998.

What initially seemed a guidestar in the firmament be-came a fading landmark in contested terrain. Second,

Congress involved itself in fusion policy not only by

adjusting funding levels but also by compelling sub-

stantive and technological changes in the national fu-

sion energy program. On occasion decision makers

defer to technical experts, but this was not such an

occasion. Third, ensembles of research technologies

offered convenient handles for policy to influence the

conduct of research in the field, so those become the

entry point for our case studies.


9/21


4. Technology and performance of two tokamaks

The name of the game is to push parameters, to

push the frontier. You cant sit back in a nice cozylittle room and take data, because thats not where

the action is. The action is always in pushing the

plasma toward the very edges of disruption, push-

ing instruments toward the edges of their measur-

ing capability. Already were pushing people to the

limit. Theyre all being pushed right to the limit of

how much they can work . . . .

Between April and December of 1985, in a

nine-month period, we will gradually push every

parameter we can. We will get the magnetic field

up to full value, get the plasma current up to full

value, push on the density, and hopefully then get

the temperature we need.

(Dale Meade, Princeton Plasma Fusion Laboratory,

quoted in Heppenheimer, 1984, pp. 59, 66).

Fusion occurs under great pressure, at both the

atomic and social scales, so the science, engineering,

policy, and work life of fusion converge at the edge of

the possible. To understand work under such circum-

stances one must first understand the ensembles ofresearch technologies used to accomplish the research.

Three categories of components make up the en-

semble of technologies used in fusion research:

plasma properties (such as shape, electron density,

energy density, temperature, and duration), design

characteristics of the device itself (such as aspect

ratio of the torus, composition of the vessel walls,

means of introducing energy and fuel into the plasma,

and presence of a divertor to remove impurities from

the plasma edge), and the number, capabilities, and

quality of diagnostic instruments on the machine(which may include interferometers, reflectometers,

Langmuir probes, Thompson scattering detectors,

heavy ion beam probes, and others).

Differences in research ensembles embody and

enforce differences in research programs, positioning

groups within the epistemic and technogenic cultures

of fusion. Some research groups, for example, em-

phasize the engineering challenges of fusion research:

they push plasma parameters, achieving hotter, denser,

and longer-lasting plasmas, then codify their results

in scaling laws that allow extrapolation from research

results to larger, higher-performing machines. They

also design and study tokamak technologies: fueling

devices, plasma impurities and their removal, andother operational and engineering issues. In contrast,

other groups are committed to the scientific side of

fusion research, seeking fundamental understanding

of plasma physics, which they pursue not by pushing

the envelope of plasma parameters but by developing

theories of plasma behavior and improving the quality,

precision, and specificity of plasma measurements.

Fusion laboratories may claim originality in various

combinations of these dimensions, establishing their

salience to the field in terms of the engineering or

intellectual contributions made possible by their spe-

cific differences. CTX and MAT are positioned quite

differently on these dimensions.

4.1. CTX and fundamental physics

CTX was conceived in the late 1970s and made its

first plasma in 1980. Three properties of CTX estab-

lish its strategic differences from other fusion labora-

tories. First, and most importantly, CTX is dedicated to

understanding the fundamental physics of fusion, par-

ticularly the proposition that turbulence in the plasma

causes the transport of energy from the core of the

plasma to its periphery, cooling it and ending the fu-

sion reaction. As a senior CTX researcher said,

we would never give up our turbulence emphasis

here. Turbulence causes transport, which has to be

understood. The reason turbulence is good to us

now and will continue to be good to us in the fu-

ture as a research area is because, although its

clear that transport is correlated with turbulence,

nobody knows what causes the turbulence. [Other

fusion groups] would not broach the issue becausethey dont care. Particularly in the larger machines,

they try to change plasma parameters, like the den-

sity, and at each different density they measure the

transport and they say, Okay, we have a straight

line, so as density goes up, transport fluxes perhaps

go down, and we know that ITER is going to have

a particular density, and so we draw the straight

line. Thats called the development of scaling laws,

and there, they dont care what causes the transport

. . . . They work on scaling laws only, no physics.


10/21


We claim that you cant do a scaling law, that you

have to understand the physics, and that if you un-

derstand the physics, that, in fact, is your scaling

law. The people who do scaling laws say, Well, wecant wait around while you guys try to figure out

what the physics is. Weve got to go ahead.

Second, CTX offers excellent diagnostic instru-

ments that make detailed physical measurements of its

relatively low-energy plasma. Using its diagnostics,

CTX

can study things in quite fine detail that cant

be done on other machines, because on the

smaller-scale machine the diagnostics are more

affordable, so you can put more of them on . . . . So

its a facility where you can do more physics . . . . It

is the best-diagnosed tokamak anywhere on Earth.

Finally, CTX is a dependable machine that recov-

ers rapidly between shots (brief episodes of plasma

production) and reliably generates many plasmas a

day, usually producing a lot of data quite efficiently.

This allows it to function at times as a user facility

for students and visiting researchers.

Fusion groups face a version of the essential ten-

sion between tradition and originality in science

(Kuhn, 1977): they must be traditional enough to es-

tablish strategic similarities that connect their work

to others in the field, and original enough to establish

strategic differences (Knorr-Cetina, 1999) that impart

novelty to their work. The strategic choices a facility

makes are built into its ensemble of research technolo-

gies and guide its research path, and by choosing cer-

tain pathways a group precludes others. For example,

plasma characteristics that permit precise measure-

ment and promote fundamental physics researcha

relatively small, cool plasma that is not very

denseinterfere with the ability to push parameters

and develop scaling laws. As one researcher observed,moderate-sized tokamaks (such as CTX) cannot play

the scaling game, because if we get a scaling [law]

on our machine nobody would care because its too

small. But if we understand a piece of the physics

then people will care because thats going to apply

to almost any machine.

There is a crucial tradeoff between the qualities

that make an inexpensive, reliable, well-diagnosed

tokamak and those that make a cutting-edge, parame-

ter pushing machine: sensitive diagnostics, such as a

heavy ion beam probe, are difficult to use with plasmas

that are very dense or very energetic (because the beamwould not pass through the plasma to the detector on

the other side). Further, for a facility to host many

researchers from other universities it must be reliable,

relatively inexpensive to operate, and quick to recharge

and recover between shots; but high-performance ma-

chines take longer to cool down and to recharge, and

run a greater risk of damage and downtime.4

CTXs epistemic commitment to understanding the

fundamental physics of turbulence and transport is a

strategic choice: they were not relegated to this role

but chose it to establish their place in the national

fusion program, and their commitment is built into

the ensemble of technologies used in their research.

Despite belief that the strategy that was good to them

in the past would remain good to them in the future,

CTX was undone by changes in the fusion policy

environment and ceased operations on December 31,

1995. Most of its researchers scattered: extensive

follow-up research located 52 of the 81 researchers

on the CTX roster in 1994 (64%), and from these data

we estimate that about a third of CTX researchers

remain active in fusion.

4.2. MATs dense, diverted plasma

MAT was conceived in 1969, soon after a campus

visit by Lev Artisemovich, one of the Soviet inven-

tors of the tokamak. Inspired by that visit, a professor

organized a group to devise a relatively inexpensive

way to study the physics of tokamaks in a university

environment (Fowler, 1997, p. 180). Capitalizing on

the universitys strength in high-end magnet engi-

neering, they

combined the physics idea [to build a tokamak] with

a technology that existed here and invented a new

kind of tokamak. We put in a proposal for that and

4 That is, between shots the capacitors that store the en-

ergy needed to create the plasma must recharge and the coils

that cool the vesselwhich on some machines contain liquid

nitrogenmust dissipate their heat. The bottom line is that CTX

can make about 100 shots per daya shot is the production of

a plasma, lasting about a secondwhereas MAT makes about 30

shots, and TFTR at Princeton makes about 5.


11/21


it was funded. So all of a sudden we had a million

dollars, which was a lot of money back in those days.

The result is a machine that uses strong magnets to

produce a dense, energetic plasma. Through succes-sive generations, MAT became a machine that

enjoys a rather unique corner of parameter space

that were able to explore. Most of the other main

line tokamaks are much larger, lower [magnetic]

field, comparable plasma current (which is one

measure of how high performance your tokamak

is in terms of its confinement parameters and how

hot it gets). And the reason we can do that with a

smaller device is because of our higher magnetic

field . . . [which] . . . allows you to run a lot of cur-

rent density in the plasma which in turn allows you

to get these high temperatures and very high densi-

ties. So the area of parameter space that we actually

explore is the high density and high power density,

and its very important for a number of reasons.

Unlike CTX, MAT produces plasma conditions suf-

ficiently similar to those of a working fusion reactor

that it can contribute to the development of scaling

laws. But MATs plasma is difficult to study and not

as well diagnosed as CTXs.

In addition to its powerful magnetic field and high

densities, MAT has a second specific difference that

establishes its relevance:

a large part of our research is focused on the di-

vertor [a magnetic device that scrapes off the

cooler outer layer of a plasma, to help it maintain

its high temperature] . . . . One of the key issues for

ITER and probably for any tokamak reactor is how

do you handle the power in the particles that come

out of the edge of the plasma? And we are in a po-

sition, because of our very high power densities, to

explore a lot of those issues at the same parameterrange in terms of power density and electron den-

sity and edge temperature that ITER will face. So

we can actually look at the physics of those ques-

tions on our machine. . . .

Finally, MATs vessel has metallic walls, which

help keep the plasma free of impurities, while many

others machines have graphite walls. This technical

innovation was produced through interaction with pol-

icymakers in a boundary organization:

People from here are on various ITER committ-

ees. . . called expert committees, and we have peo-

ple that go to these things and not only are [they]

involved in trying to do things that people buildingITER think are important, but [they] also have in-

put into what we think are important. . . . the fact

that our walls are made out of molybdenum, rather

than carbon, was a major issue that we had to go

through which is maybe changing the way people

think about how the wall ought to be of the machine.

And so I think we have autonomy in so far as that

we can sort of suggest what we think is important

. . . [and] what were interested in studying .. . .

Both the technology and the sense of autonomy that

accompanies it are constructed through interactionwith collaborators, competitors, and policymakers.

The components of a research ensemble are not

independent dimensions but interdependent tech-

nologies and phenomena that afford certain research

opportunities and technological developments while

precluding others. The MAT divertor not only allows

study of a technological innovation and improves the

quality of the plasma, but it also poses compelling new

questions of fundamental physics. This interaction is

recursive, with properties of the plasma influencing

characteristics of the technology: as tokamaks get

more and more dense and hotter and hotter it turns

out that you cant just pour gas into the side because

it doesnt penetrate well. So in place of gaseous

deuterium fuel, at MAT

we have the pellet injectors that actually freeze deu-

terium . . . and turn it into a sort of a slushy frozen

snow and then we put that in a tube and more or

less like a blow gun we inject high pressure helium

gas and it pushes it through a tube and then accel-

erates it at very high rates into the plasma and then

that fuels the plasma . . . . One of the neat things isthat, because the pellets are going at such a high

rate that they ablate on the way in and carry fuel

all the way into the center, you can get very peaked

density profiles that have different characteristics

than just a gas fuel system.5

5 A peaked density profile is a plasma property that may be

very important for achieving a burning plasma, so this is a very

significant characteristic for MAT to achieve (National Research

Council, 2001).


12/21


So the dense plasma requires the addition of pellet

injectors to the tokamak, and the pellet injectors in

turn improve the plasma, enhancing MATs research

program.This may be a novel form of the process of accu-

mulative advantage, which has been well established

in science studies (Merton, 1973; Latour and Woolgar,

1979; applied to groups by de Haan et al., 1994).

Ordinarily, inequalities of performance increase over

time as success begets success: resources become re-

sults, results become publications, publications en-

hance reputation, and reputation brings resources in a

cycle of credibility that increases the variability of

reputation, rewards, and resources in science. In this

novel form, advantage accumulates through a process

that centers on the research technologies groups use

and the contingencies that affect their changing place

within the research and policy realms. Properties of

research ensembles interact with one another, with the

groups interests and abilities, and with the accom-

plishments of other groups and the priorities of poli-

cymakers to produce advantages or disadvantages. At

MAT properties of the plasma interact with technical

features of the machine and research priorities of the

field to bestow future advantages. At CTX, in contrast,

plasma properties and epistemic commitmentsthe

cool but well-diagnosed plasma and commitment tounderstanding the fundamental physics of how turbu-

lence causes transportconferred initial advantages

that later became disadvantages.

Tokamaks are always in flux, upgrading tech-

nologies and performance in a turbulent policy en-

vironment, so they routinely operate near the edge

of breakdown. Whether this danger is a liability

or an asset depends on social characteristics of the

group. Throughout their histories, both CTX and

MAT experienced significant downtime, accompanied

by reductions in research and publication. Making avirtue of an occupational hazard, for example, some

MAT researchers study

halo currents, which are when we actually lose con-

trol of the plasma and it goes unstable, it slams into

the walls, and when it does that it can drive very

high currents in the vessel. It does interact with the

magnetic fields and can cause large stresses on the

vessel . . . we had always thought that the plasma

would go up or down sort of uniformly all the way

around the torus. But it turned out, in fact, that it

rotates around and crashes sort of locally and can

in fact generate much more force than we thought it

could, because rather than going down uniformly,it sort of goes down, you know, at one location and

then sort of spirals down.

To summarize, through machine upgrades, failures,

and downtime MAT adjusted to the shifts in fusion

policy, parlaying advantages of their research ensem-

ble into other sorts of advantages (including relevance

to shifting policy priorities), earning a growing share

of a shrinking pie.

4.3. Performance: publications and citations

Publications and citations are widely accepted in-

dicators of the scientific and technical performance of

research groups (Andrews, 1979). Admittedly, these

are quite simple measures, and for some purposes ex-

pert evaluations of intellectual and technical contribu-

tions, patent counts, and other indicators may be pre-

ferred. But these are adequate for sketching the mag-

nitude and trajectory of each facilitys performance.

Fig. 2 shows publications for each facility from

1985 to 1999, presented as 3-year moving averages

centered on the middle year. Thus the publicationcount of 1985 is one third of the sum of publications

in 1984, 1985, 1986. The moving average smoothes

out sharp annual differences that can mask trends and

misrepresent publication patterns (because there is a

certain amount of arbitrariness and error in a publica-

tion date). In this period CTX published 104 articles

that received a total of 1217 citations (11.7 each)

while MAT published 142 articles that received 1759

citations (12.4 citations each). MAT published at a

much higher rate than CTX during the mid-1980s,

then CTX became much more productive, outstrippingMAT during the early 1990s, peaking in 1992, then

declining as MAT rose during the critical 19941995

biennium. The pattern reflects turbulence in the policy

environment and continual adjustments of the re-

search ensembles to new contingencies. Importantly,

there is no evidence of large initial differences or

accumulative advantage (a widening publication gap

over time). The argument that one facility is vastly

better than the other, or acquired and cultivated an

enduring advantage, would be difficult to sustain.


13/21


1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

0

2

4

6

8

10

12

14

16

18

20

ArticlesPerYear

Year

CTX

MAT

Fig. 2. CTX and MAT articles per year 3 years moving average.

We treated citations similarly, tracking them for

the same years, calculating a moving average, divid-

ing by the number of publications in the period, and

graphing the time series (see Fig. 3). Citations are

somewhat more difficult to interpret, because we did

not limit the range of referencing years but included

all citations received up to the search date.6 The cita-tion patterns begin with CTX much higher than MAT,

the result of peculiar circumstances: each facility pub-

lished its most highly cited article in 1984, and CTXs

has received 148 citations to date, MATs 167. But

MAT published more articles than CTX in 19841986

(17 versus 4), so its blockbuster citations are spread

more thinly. The pattern soon smoothes out, with

MAT receiving more citations than CTX for work

published during the mid-1980s. Then it is overtaken

by CTX, just as it was in the publication chart. But in

1993 MAT overtakes CTX and the lines run roughlyparallel for the duration (that is, through 2000).

In sum, publication and citation data present a

consistent and supportive picture: there is no birth

advantage for MAT; the temporal pattern is turbu-

6 Citations are given to articles with whatever improved per-

spective the times allow, so the rising published output of MAT

may directly cite earlier MAT publications and may also call at-

tention to that work, thus attracting citations from others. CTX

publications would experience complementary disadvantages.

lent, reflecting downtime for each machine and rising

and falling fortunes with the field; and there is a

transition point in both graphs that marks a sharp

change in fortunes that occurred at the time of our

fieldwork.

4.4. Summary and comparison

CTX and MAT are academic facilities of similar

age, size, social composition, and performance that

conduct fusion energy research using tokamaks, but

they do so with strikingly different ensembles of re-

search technologies. CTX is a relatively small and

reliable tokamak, with excellent diagnostics and ded-

ication to a fundamental question in plasma physics.

For its connection to ITER (and through that to the

main concerns of US fusion research policy in the

mid-1990s), CTX argues that understanding funda-mental physics is essential to the construction of a

working fusion reactor. It makes this argument at a

time when the largest and most expensive machines

in the world are pushing parameters and develop-

ing scaling laws. MAT is a small machine that uses

strong magnetic fields to produce hot, dense, energetic

plasmas. Its divertor cleans and improves the qual-

ity of the plasma, and its vessel walls contribute to

plasma quality and technological importance. MATs

plasmas are similar to those of the largest machines


14/21


1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998

MAT

0

5

10

15

20

25

30

35

40

45

CitationsPerArticle

Year

CTX

MAT

Fig. 3. CTX and MAT citations per year 3 years moving average.

and approach those needed in a working fusion

reactor.

5. Technology and work in fusion

The vocation of science, Weber reminded us, entails

personal rewards and significant perils. As one of the

most skilled and engaging lines of work, research is

generally known to provide considerable intrinsic sat-

isfactions and rewards: autonomy, challenge, develop-

ment, meaning, variety, collegiality, and the like. But

such rewards are not available to all researchers: char-

acteristics of the vessel and the seas it sails influence

life on board. An episode from CTX and a compara-

tive analysis of survey data will illustrate the point.

5.1. Upgrading technology and degrading work:

a cautionary tale

Once its achieved then we can study it . . . . Its

actuallyI dont know if its been explained to

youits a little embarrassment that we dont have

it, and it could have a major funding consequence,

which is why were also worked up about it. And so

its putting a lot of stress on a lot of people. Were

pretty good about not pointing fingers: Nobodys

saying its your fault or anything like that. But

after youve been trying to do something for days

and days and days you get very frustrated, and that

makes the interactions harder. Its been happening

now for about [pfffft: an unintelligible sound that

implies a long time] and makes the environment

harder than it normally is.

(A CTX physicist)

It is H-mode, a sharp rise in energy confinement

time caused by electrostatic fields at the plasma edge

(Fowler, 1997, pp. 3738).7 H-mode is a new scal-

ing, which means that as energy input rises, confine-

ment time increases at a rate about double that of low

mode. Simply put, the plasma becomes surprisingly

well behaved for its energy level. Machines that can

attain H-mode produce plasmas of greater scientific

and engineering interest, so it is very important to do

so. The Department of Energy established H-mode asan objective for CTX, and the facility spent years in

the effort. The quest for H-mode at CTX reveals the

hazards of work in big technoscience:

We should have gotten it [H-mode] three years ago.

Then the divertor coils broke and we had to take

7 H-mode was first observed in the ASDEX machine in Ger-

many in 1982, then confirmed and explained through a series of

experiments at UCLA and DIII-D in La Jolla.


15/21


them away and make new ones. And then we finally

had divertor coils but the gyrotrons werent work-

ing and it took us a long time to get them working.

Actually, the gyrotrons were a very peculiar prob-lem in the sense that, many other problems you can

throw people at them and with more people it will

become solved. But in that particular one there were

the people who were the specialists and they could

only do what they could do. So that was very hard

on them in particular. . . . You can ask for somebody

to work 10 h a day for a week, maybe two, but not

for a year. [Laughs].

Another senior scientist said H-mode is our cur-

rent problem right now . . . . We were spending as

much as half of our time just on development of thatone discharge condition. Pursuit of this mandated

upgrade produced a cascade of mechanical failures

and interrupted research, which greatly disturbed

work life at CTX:

Theres a lot more stress, a lot more fear. People

have been fired in the meantime. Competition is

healthy. I dont see people trying to do people in,

which could be possible, considering that one of us

could be fired next year. . . . But that doesnt seem

to make people spoil things for people. Were still

working together in this. But there is a lot of stress,there isa lot offear. . . . We all feel were being asked

to do too many things too quickly. Thats difficult,

and it seems to be normal nowadays. [Laughs].

Among researchers autonomy is highly valued in

itself and as a pathway to personal and professional

development, and development is valued more highly

than conventional advancement or promotion. Asked

about the chances to get ahead at CTX, a physicist

replied I do not think any of us wants to go up: in our

field the way up is just to stay where you are but have

less requirements imposed on your time. You dontnecessarily want to be the boss of anybody, you just

want to be the boss of yourself. Yet autonomy of pre-

cisely that sort, and all that it entails, is diminished in

pursuit of the mandated upgrade of CTX. Comparing

current conditions to those at CTX a few years earlier,

a scientist recalled that

There was a time, a few years ago, when I had the

energy to tell others, Look, Im not going to do what

you want. Im doing this thing, its more important.

And I got away with it. But because were now in

crisis mode I dont feel like it will be responsible to

do that. . . . Ill eventually get around to it, as soon

as we get over this H-mode thing. Hopefully Ill getback to doing what I want to do. But it has more to

do with the situation of the whole group. And the

problem is that the budget constraints dont make it

look good. People are being fired: it means the ones

that stay have that much more to do. The things that

need to be done are the same. So it is not very likely

in the near future that Ill be the boss of my time.

The picture for getting more autonomy doesnt look

very good. The picture for fusion doesnt look good,

either. As our budget shrinks and our needs expand,

well have less and less autonomy. Its actually veryhard for any of us to leave the field, because there

is very little to go to . . . . Most of us cant really

imagine ourselves doing anything but research, be-

cause thats what we do.

Autonomy is at the core of a researchers need to

sustain and develop competences and to remain em-

ployed, and pressure to upgrade the machine lowered

autonomy and reduced opportunities for professional

development.

I let the group take precedence [in establishing myresearch agenda and day-to-day responsibilities]

because were in such a crisis situation . . . what the

group needs becomes a lot more important because

we actually need it . . . . Its not particularly good

for my career. . . . As a theorist I am losing ground

because Im so busy with experiments I dont have

time to keep up with the theory . . . . I would never

be hired as an experimentalist, because thats not

what I am. I may be hired as a theorist [but] it could

damage me [to remain so involved with experimen-

tal work that my theoretical skills remain underde-veloped] . . . . The choices I make are affected by

the environment, they certainly are. It is a strange

balance and Im not always sure which is the right

answer . . . . As long as Im doing what the experi-

ment needs, my job working with the experiment is

safe if the experiment is safe. On the other hand, if

the experiment is gone, Im less safe than I would be

if I chose to do other things . . . . The path I am tak-

ing does not take me to [another, more prestigious

fusion facility]. If I had made the other choice and


16/21


gone more with theory I might have had a chance

to go [there].

(A CTX physicist)

To remain operational CTX needed first to upgrade

its machine by achieving H-mode, which would gain

some operating time, then to compete successfully

for a new machine. The upgrade was a years-long

struggle and the ground rules for the competition did

not favor research competences and properties af-

forded by CTXs research ensemble: the new machine

would require more power and more radiation shield-

ing than CTX now has, and its research would focus

on magnetohydrodynamics, not a CTX strength.8

Policy influences drove CTX to pursue objectives

that strained the groups competences and the affor-

dances of its research ensemble, diminished auton-

omy, reduced research productivity and jeopardized

careers.

In summing up the prospects for CTX, a scientist

opined that

the consequence of not getting the [new] machine

is that its possible CTX will have to shut down

. . . . DOE is the funding agency and they decide

where the money goes . . . . Officially there are ad-

visory committees made up of scientists, and the

advisory committees tell DOE what they think is

important, and then DOE goes ahead and acts on

it. But in reality DOE has much more of a say in

8 The DOE influence on research ensembles is sometimes as

dramatic as supporting the construction of a facility or mothballing

one that has just been constructed (as happened at the national labs

in Livermore and Oak Ridge). But DOE influence can be more

measured, as when the agency directs the academic researchers

it supports to locate (or relocate) their diagnostics on particular

tokamaks. Asked about the history of his work with CTX, a

researcher saidThe history was. . .we were asked to do so by DOE . . . . It was

determined by DOE and I think also by CTX that they needed

to have a more focused program in transport and turbulence, and

toward that they should hirebasically subcontractto Hilltop

Tech and ourselves, and I think there was one other group . . . .

So there were basically three people that were put into the

system to come and jump start a program in turbulence and

transport at CTX.

Another added that theres never any implication of force in what

I am saying. All such things are discussed and agreed on well

beforehand, prior to DOE says you need to go here.

what happens with us. It depends on the particular

person in DOE. . . .

DOE is an active presence in the direction andfate of the group, and it acts not entirely imperiously

and not always abruptly, but in concert with others

in the field (constituted as advisory committees) and

with a degree of delicacy. While not ordered, as a

principal might order an agent, such directives were

discussed but never negotiated. Thus DOE guides

fusion research groups by using boundary organiza-

tions to exert pressure on budgets, personnel, and per-

haps most powerfully, on the technical requirements

and capacities imposed on the research ensemble.

Autonomy, development, and teamwork may become

casualties of the machine-mediated pressure applied

by DOE. And the influence is neither vague nor dis-

tant: during our fieldwork we noticed daily phone

conversations between CTX administrators and DOE

officials, some occurring in our presence, and re-

searchers telling us of weekly and monthly progress

reports.

Twin threatsthat a single researcher may be lost

or that the entire vessel may capsize or founderare

vividly present in the researchers mind. As matters

unfolded CTX was shut down in December 1995, af-

ter 15 years of operation. At this writing the scientistinterviewed above is employed elsewhere in the fusion

research community.

5.2. Measures of work life at CTX and MAT

What is his personal dedication to the job? Has he

worked in a group? Can he sacrifice himself com-

pletely to the team? Thats very tough; often these

qualities dont go together. Some people who work

hard will work that way for themselves and no-

body else. That is, theyll really put a lot of effort

if theyre interested in a problem, but if theyre not

interested in it, even if their boss is or their col-

leagues arethey wont . . . . I look for people who

have enough of this group spirit to maintain their

motivation. Because theyll be working with a team,

a very dedicated, professional team . . . . I may not

be able to offer these guys a high salary, but I try

very hard to offer them professional development,

professional recognition.


17/21


(Dale Meade, head of the Experimental Division of

the Princeton Plasma Fusion Laboratory, explaining

his hiring standards. Quoted in Heppenheimer, 1984,

pp. 5657, 61)

We have argued that characteristics of the ensemble

of research technologies used by a group will influ-

ence the work and work lives of members. Differences

in the research ensembles of CTX and MAT have been

established through descriptions of the facilities, mea-

sures of performance, and in the words and experi-

ences of researchers themselves. In this section we use

questionnaire data from small samples of researchers

to confirm and refine these results by comparing work

life and job satisfaction at CTX and MAT. We find

that researchers at the two facilities are quite similar

in social background, work-related values, and task

characteristics, but differ significantly and predictably

in perceptions of the field and the groups place in it,

working conditions, and in the satisfactions received

from their work.

In educational attainment, sex and ethnic composi-

tion, citizenship, age and seniority (measured within

the profession, the organization and the group) the

two facilities are nearly identical (data not shown).

Each group is predominantly composed of white,

male, highly educated US citizens. On average eachgroup is about 40 years old, with a bit more than

12 years experience since attainment of the highest

degree, employed in the current organization, and

with the current group, for almost all of that time.

None of the small differences approaches statistical

significance at the 0.10 level.

Researchers at CTX and MAT value the same work

characteristics to about the same degree: autonomy,

collegiality, and development opportunities top the

Table 1

Perceptions of the places of CTX and MAT in the field

Perception CTX MAT Diff. (sig.)

The work of our group is central 4.2 5.5 1.3 (0.00)

We have trouble keeping pace with developments 4.2 2.2 2.0 (0.00)

Our group has changed research problems to avoid competing 2.9 2.1 0.8 (0.02)

People in this field agree about the important research questions 4.4 5.3 0.9 (0.02)

Groups in this field cooperate freely with one another 3.5 5.1 1.6 (0.00)

There is a lot of competition between groups in this field 5.3 4.7 0.7 (0.07)

Scale scores range from 1 (strongly disagree) to 7 (strongly agree). All significance levels are based on t-tests and are two-tailed.

list, followed in order by intrinsic rewards (such as

challenge and variety), control over work of others,

and extrinsic rewards (such as pay and job security).

Such results are unsurprising and confirm expecta-tions based on many other studies of scientists and

engineers (e.g. Pelz and Andrews, 1976; Miller, 1986;

Watson and Meiksins, 1991; Jones, 1996; Keller,

1997). The groups differ only slightly on particu-

lar items, and none of those differences approaches

statistical significance at the 0.10 level.

Doing research involves a variety of activities, and

the mix may vary from place to place. We subdivided

research work into 10 tasks, asking researchers how

involved they were in each one. The tasks follow

the sequence of a research project: choosing a topic,

reviewing literature, conceptualizing the problem,

gathering and analyzing data, and writing up the re-

sults. The tasks performed by researchers at the two

facilities differ in only one respect: MAT researchers

are more involved in building equipment and appara-

tus. Otherwise researchers at the two facilities display

similar profiles on the spectrum of research tasks.

Against this backdrop of similarity, several striking

differences emerge.

Table 1 reports researchers perceptions of the field

and their groups place within it, and there are con-

sistent differences in perceptions from the vantagepoints of CTX and MAT. Researchers at MAT see their

work as central to the field, have little difficulty keep-

ing up with new developments, and have not changed

research problems to avoid competition. From their

vantage point, the field is characterized by consen-

sus, cooperation, and relatively low competition be-

tween groups. For CTX the circumstances are much

different: their work is less central to the field, they

have trouble keeping up with technical developments,


18/21


Table 2

Working conditions at CTX and MAT

Working condition CTX MAT Diff. (sig.)

Career security 2.5 2.7 0.2 (ns)Autonomy 3.6 3.8 0.2 (ns)

Specialization 2.7 2.1 0.6 (0.00)

Resources 3.0 3.5 0.5 (0.00)

Collegiality 3.2 3.7 0.3 (0.01)

Development 3.3 3.6 0.3 (0.01)

Intrinsic job characteristics 3.3 3.6 0.3 (0.02)

Scales are scored from 1 (low) to 5 (high). All significance levels

are based on t-tests and are two-tailed.

and have changed problems to avoid competition. For

them consensus and cooperation are lower, competi-

tion higher. All but one of these differences (compe-

tition) are significant at p = 0.02 or below.

We propose that differences in research ensembles

interact with the policy environment to influence the

work life and job satisfaction of group members, and

Table 2 shows significant differences in several as-

pects of work. Researchers at MAT report significantly

better resources and collegiality, better intrinsic job

characteristics (e.g., variety and challenge), less spe-

cialization, and greater opportunities to develop as a

scientist or engineer. There are no appreciable differ-ences in job security and autonomy, which is reason-

able because fusion researchers are employed in very

similar jobs and their security depends upon much the

same funding environment.

Table 3 reveals statistically significant and substan-

tively large differences in research ensembles influ-

ence the job satisfaction of group members. Dimen-

sions of satisfaction most closely related to properties

of the research ensembleequipment, development,

recognition and collegialityare all significantly

higher at MAT than at CTX. Intrinsic satisfactions,such as challenge, autonomy, variety, and benefit of

the research to society, are also about a half point

higher at MAT. Finally, researchers at MAT are more

satisfied with their pay and with the contribution of

their research to their careers, but are no higher in job

security or safety and comfort of the work. In sum,

there are large and significant differences between

CTX and MAT in satisfaction with resources and, to

lesser extents, with the intrinsic and extrinsic rewards

of work.

Table 3

Work satisfaction at CTX and MAT

Satisfactions CTX MAT Diff. (sig.)

Resources (scale) 5.1 5.9 0.8 (0.00)Equipment 5.4 6.4 1.0 (0.00)

Development 4.6 5.6 1.0 (0.02)

Recognition 4.8 5.5 0.7 (0.03)

Collegiality 5.2 5.9 0.7 (0.05)

Intrinsic rewards (scale) 5.4 5.9 0.5 (0.02)

Challenge 5.8 6.1 0.3 (ns)

Autonomy 5.3 5.7 0.4 (ns)

Variety 5.6 6.3 0.7 (0.01)

Societal benefit 4.7 5.2 0.5 (ns)

Extrinsic rewards (scale) 4.3 4.9 0.6 (0.04)

Pay 4.0 4.8 0.8 (0.06)

Job security 3.8 4.3 0.5 (ns)

Safety and comfort 5.2 5.4 0.2 (ns)

Career contribution 4.2 5.3 1.1 (0.01)

Overall job satisfaction 5.1 5.8 0.7 (0.06)

Scales are scored from 1 (low) to 7 (high). All significance levels

are based on t-tests and are two-tailed.

Overall job satisfaction is an integrative psychic

response that spans the specific dimensions of the job

measured above (Hackman and Oldham, 1976). To es-

timate the relative influence of the research ensemble

on overall satisfaction, we regressed satisfaction onto

three composite scales: extrinsic satisfaction (pay,

security, comfort, and career contribution), intrinsicsatisfaction (autonomy, challenge, development, and

variety), and resource satisfaction (equipment, col-

leagues, and recognition). We expect the facilities to

differ in one principal way: the effect of resources

on satisfaction should be positive and significantly

stronger for MAT than for CTX. The regressions

were run in two steps, first including only intrinsic

and extrinsic satisfactions as predictors, then adding

resource satisfaction to the equation.

The analysis presented in Table 4 supports this

prediction. The upper panel shows that intrinsic sat-isfaction has roughly equal positive effects at the two

facilities and that extrinsic satisfaction more strongly

predicts overall job satisfaction at CTX than at MAT.

Resource satisfaction has a dramatically positive ef-

fect on satisfaction at MAT: the coefficient is the

largest in the table and the proportion of variance

explained leaps from 0.38 to 0.71. In contrast, re-

source satisfaction has a slight (but not significant)

negative effect on satisfaction for CTX and the overall

R-squared is essentially unchanged. Taken together,


19/21


Table 4

Regressions of overall job satisfaction onto facet satisfactions

CTX MAT

Model 1 b t (sig.) b t (sig.)Extrinsics 0.56 0.45 2.9 (0.01) 0.20 0.22 1.3 (0.20)

Intrinsics 0.44 0.32 2.0 (0.05) 0.67 0.49 2.9 (0.01)

Constant 0.31 0.84

R2 adj. 0.46 0.38

Model 2

Extrinsics 0.63 0.51 3.2 (0.00) 0.02 0.02 0.20 (0.84)

Intrinsics 0.57 0.41 2.4 (0.02) 0.38 0.28 2.3 (0.03)

Resources 0.29 0.20 1.3 (0.20) 0.89 0.67 6.4 (0.00)

Constant 0.86 1.90

R2 adj. 0.47 0.71

N 34 37

these analyses confirm that characteristics of the en-

semble of research technologies used by a group,

in interaction with the wider context, influences the

work life and job satisfaction of group members.

6. Summary, conclusions, and implications

We have developed the idea of resear

Edward J. Hackett et al- Tokamaks and turbulence: research ensembles, policy and technoscientific...

Documents

Transcript of Edward J. Hackett et al- Tokamaks and turbulence: research ensembles, policy and technoscientific...