What is the matter with e-Science? – thinking aloud
about informatisation in knowledge creation
Paul Wouters
Networked Research and Digital Information (Nerdi)
NIWI-KNAW
The Royal Netherlands Academy of Arts and Sciences
PO Box 95110
1090 HC Amsterdam
The Netherlands
T 3120 4628654
F 3120 6658013
http://www.nerdi.knaw.nl
paul.wouters@niwi.knaw.nl
This long quotation is a future scenario taken from a rather old article (De Jong &
Rip, 1997). It describes a possible future in which “discovery environments” have
fully developed. Scenarios like this, but usually less well written, are now quite
popular to make the case for e-science. By staying rather close to what is presently
technically possible these scenarios avoid the stigma of science fiction and are able to
operationalise the technical challenges into real-world computer science problems. I
think they capture the sometimes rather grandiose future visions that are dominant in
the escience movement.
In this presentation I wish to address the question: why should we study e-science and
if we wish to do it how can we conceptualise it in a somewhat productive way?
E-science is generally defined as the combination of three different developments: the largescale sharing of computational resources, the provision of access to massive, distributed and
heterogeneous datasets (in the order of tera to petabytes), and the use of digital platforms for
collaboration and communication. E stands not in the first place for “electronic” but for
“enhancement”. The core idea of the e-science movement is that knowledge production will
be enhanced by the combination of pooled human expertise, data and sources, and
computational and visualisation tools.
I see it as significant that most of this is still promise rather than practice. But a promise with
a very material financial and infrastructural embodiment. Not something to lightheartedly
dismiss as the next fad in science policy. E-science is a discursive construction at the
interface of technoscientific practices, computer technology design and science policy
that is being embodied in material infrastructures, a demand for new sociotechnical
skills in research, and a pressure on existing scientific and scholarly practices. The
justification is the promise. The promise is a new version of the old idea of the World
Brain of H G Wells which also played an important role in the construction of the
internet. We should not think that the promise limits itself to computational research.
The UK develops a large e-social science programme. In the Netherlands there is a
modest move to create something called e-humanities. The e-Science community is
interested in the sociology of science in order to change habits and structures that may
hamper the further development of e-science. STS might be instrumental in the spread
of e-science across the board.
This means that, apart from funding opportunities for new case studies, there may be
an interesting intellectual challenge for STS. I see four different types of approach and
conceptualisation of e-science. I guess somehow they are all represented in this panel.
First, of course, it seems obvious that the analysis of e-science as a political
movement might be very interesting. Second, since very broadband computer
networks and distributed storage and computing facilities seem at the heart of escience, there is scope for technology assessment studies. Michael Nentwich’s book
Cyberscience (Nentwich, 2003) is a beautiful exemple of this approach. It tends to
lead to very inclusive encyclopediac descriptions of all sorts of developments that
seem relevant to cyberscience or e-science. More importantly, this approach by the
nature of its description tends to reify the phenomenon of e-science. In the end escience is everywhere, difficult to escape and difficult to skepticise. Here the third
approach might come to our rescue: the trusted case study zooming in not on the
technology with its promises, but on the hands-on scientific practice. And suddenly,
e-science seems to be nowhere anymore. We see the usual tinkering of the scientist,
easily making use of both online and offline resources, mixing heterogeneous stuff as
they have always done. The local is everything. Whether an e-science resource is
being used or not is determined in the local context in which all the supposedly global
stuff is recontextualised. The case study might of course focus on the creation of a
particular e-science project, for example the Virtual Observatory. A clear case of the
application of the social shaping of technology.
If it is the case that the local is connected to other localities through networks, a fourth
approach seems promising: to focus on those networks and connections. It might
overcome the inherent myopia of local case studies while avoiding the totalising
perspective of a unifying drive to e-science. Several theoretical and methological
frameworks have been developed that seem relevant to this fourth take on e-science:
virtual ethnography developed by Christine, ANT, and perhaps the combination of
network analysis with the social shaping of technology.
I think that a very interesting candidate here is the notion of the “epistemic object” as
it has been developed by Rheinberger (Rheinberger, 1997). In two ways: it seems
very productive to analyse some specific features of e-science, as I try to explain in
this paper. But also, it may give as a good theoretical drive to study e-science. In
using the epistemic object to deconstruct and analyse e-science, e-science might help
us to find out in more detail and across a wider spectrum of epistemic cultures how
epistemic objects actually work, how they can and more importantly cannot be
recombined with each other, and transmuted in technical objects and vice-versa. This
relates to digitisation, signification and spaces of representation. If the epistemic
object is a digital representation and if many if not all technical objects in the research
practice are digital, and if epistemic objects of other fields are also both digital and
available, there seems no techno-material barrier anymore for the endless
recombination of epistemic objects. This is the promise of e-science.
Nevertheless, there will be many obstacles left. Many recombinations will not be
possible, much promise of e-science will turn out to be not even desirable. These
failures of e-science that may come to the surface in a very naked sense (no longer
hidden by the noise of material impossibilities) seem to me extremely promising to
deeper understand the nature of epistemology in the STS sense of the word, to
understand the culture of knowledge. This is I think a very good reason, and possibly
the best, to study e-science and take it seriously. We would turn the failures (which is
not the same as the controversies) into our main epistemic object with respect to escience.
Perhaps some more detailed explanation is possible. Rheinberger localises the
epistemic object in an experimental system, others have put it in other contexts. I
think the concept can be applied across the board of knowledge creation if we focus
on the circulation of reference that is at the heart of both scientific and scholarly
research. For Rheinberger the objectivity of science is generated by chains of
representation in which every referent turns into a representation as soon as we focus
on it. Representation looses its referential meaning. Epistemic objects are bundles of
traces. “The activity of scientific representation is to be conceived as a process
without ‘referent’ and without assignable ‘origins’.” Rheinberger distinguishes two
elements of experimental systems. The first is the research object, which is called the
“epistemic object” that embody that which is not yet known. The second element are
the set of experimental conditions in which the research objects are embedded, which
he calls the “technical objects”. Within a particular experimental system both types of
elements “are engaged in a non-trivial interplay, intercalation, and interconversion,
both in time and in space. The technical conditions determine the realm of possible
representations of an epistemic thing; and sufficiently stabilized epistemic things turn
into the technical repertoire of the experimental arrangement” (Rheinberger, 1997, p.
29). So what is the difference between a technical object and an epistemic one? It is
functional rather than structural, there is no “essence” here. “We cannot once and for
all draw such a distinction between different components of a system. Whether an
object functions as an epistemic or a technical entity depends on the place or “node” it
occupies in the experimental context.” (Rheinberger, 1997, p. 30). The main
functional criterion is that epistemic objects are generators of questions. A technical
product is stabilised and is first and foremost “an answering machine”. “In contrast an
epistemic object is first and foremost a question-generating machine” (Rheinberger,
1997, p. 32)
“What is significant about representation qua inscription is that things can be represented outside their original and local context and inserted into other contexts. It is
this kind of representation that matters.” (Rheinberger, 1997, p. 106) Thinking about
e-science this is especially interesting because the very core of what many e-science
projects aim for is the decontextualisation of objects and subsequent
recontextualisation on the fly and in any context. How is this being made possible?
By metadata, a rather dull word for information that should describe the “meaning” of
the object/data in such a way that other machines and humans can make use of those
objects/data in contexts that might have been unthinkable at the moment of the
production of the object/data. Metadata are representations of the original context of
epistemic objects in such terms that new contexts can be created for these objects to
generate new questions. The main trick that should do this work is not simply
querying the epistemic object in its new context, but basically the reconfiguring of
new epistemic objects by the recombination of already existing ones in new contexts
(or in Rheinberger’s terms new technical objects). This means of course that also
technical objects can be turned into epistemic objects and the other way around. Seen
in this perspective, the whole e-science business is nothing new at all, except for the
scale and ease with which objects can be interchanged since they are already digital
representations. By representing the whole universe of relevant stuff (both technical
and epistemic objects) in digital objects, the interconversion is indeed seamless (apart
from the hard work behind the scenes and the hard work of producing the material
conditons for the whole business of digital representation). But then again: isn’t scale
and ease all that matters?
I think this perspective does give both a way of speaking about e-science which is
sufficiently different from ‘actor’s speak” to be interesting (for them and for us), and
a way of formulating a research agenda that has the potential of critically
interrogating the very notion of e-science (Woolgar, 1988) while at the same time
studying it “in vivo/silico”.
References
De Jong, H., & Rip, A. (1997). The computer revolution in science: steps towards the
realization of computer-supported discovery environments. Artificial
Intelligence, 91(2), 225-256.
Nentwich, M. (2003). Cyberscience. Research in the Age of the Internet. Vienna:
Austrian Academy of Sciences Press.
Rheinberger, H.-J. (1997). Toward a History of Epistemic Things: Synthesizing
Proteins in the Test Tube: Stanford University Press.
Woolgar, S. (1988). Science: the very idea. London: Tavistock.