On the face of it, Brian Tonner looks like a bit of a slacker. Despite being
in charge of a host of instruments on the world鈥檚 brightest beam of X-rays at
the Lawrence Berkeley National Laboratory near San Francisco, he hardly ever
goes to work there. The X-ray beam passing through his equipment constantly
churns out fascinating insights into the miniature world of atoms and
molecules鈥攁nd yet weeks can pass without Tonner laying hands on his
experiments.
But appearances can be deceptive. In fact, Tonner works very hard nearly 3000
kilometres away at the University of Wisconsin in Milwaukee. Every day, he puts
his nose to the grindstone over the Internet. Tonner is part of a
鈥渃ollaboratory鈥�, a new way of carrying out scientific research, controlling
instruments and interacting with colleagues over vast distances.
For years, scientists have used the Net to collaborate鈥攊n fact, the
seeds of the Net were sown in the 1960s by researchers at four US universities
working for the US Department of Defense. They formed a computer network known
as the Defense Advanced Research Project Agency network, which allowed them to
communicate easily and exchange theories. This was the forerunner of today鈥檚
global network.
Advertisement
Now the speed and capabilities of the Net are far greater, allowing
experiments to be carried out in real time over the network by investigators
around the world. Collaboratories are bringing together groups separated by
thousands of kilometres at the same virtual lab bench, enabling them to pool
their knowledge and work with a global view.
Breaking down walls
Tonner and other physicists are members of one of several collaboratories
that have access to the Advanced Light Source, a synchrotron-radiation beamline
at the Lawrence Berkeley National Laboratory. Tonner is principal investigator
for the spectro-microscopy facility, and studies the atomic structure of
materials and the surfaces of solids. His work involves running experiments on
the collection of electron microscopes and other instruments attached to the
X-ray source, chatting with his postdoctoral assistants there, and consulting
with fellow scientists at other laboratories around the country.
The collaboratory system allows him to do all this from his desk in
Milwaukee. And when he does decide to visit Berkeley, he can keep up-to-date
with progress at Milwaukee. Wherever Tonner goes, he can 鈥渨alk鈥� into the virtual
laboratory as easily as he can wander among the benches at his home
university.
The teams using collaboratories claim they will change the way that science
is done. They say such long-distance systems will break down the walls that have
confined much cutting-edge research to scientists from elite institutions. They
will also allow students to participate in proper scientific research at a much
earlier stage in their careers, motivating them in a way not possible before.
快猫短视频s, it seems, are preparing to 鈥渢ake back鈥� the World Wide Web.
Dozens of collaboratories are already under way. For example, upper
atmospheric physicists from six American universities have been able to make
observations of the interactions between the solar wind and the atmosphere over
Greenland, courtesy of the Upper Atmospheric Research Collaboratory (UARC) based
at the University of Michigan in Ann Arbor. The system links the researchers
with a US-funded radar in Greenland that continuously monitors the upper
atmosphere.
The US government鈥檚 Pacific Northwest National Laboratory in Richland,
Washington, has developed an environmental and molecular sciences collaboratory,
which enables people to operate remotely two nuclear magnetic resonance
spectrometers and other instruments at the laboratory. The Department of Energy
has just launched two major collaboratory schemes. The Diesel Combustion
Collaboratory Project will link researchers at three US government laboratories
and the University of Wisconsin, Madison, with the aim of designing diesel
engines that produce far less pollution. The other scheme, known as the
Materials Microcharacterisation Collaboratory, will link researchers at four
government laboratories and the University of Illinois at Urbana-Champaign.
These researchers hope that studies of catalysis will help them to develop
corrosion-free materials.
鈥淭his is our way of making all our facilities a virtual collaboratory. This
is the first step,鈥� says Martha Krebs, head of the energy department鈥檚 Office of
Energy Research. To develop the necessary new technology, the agency plans to
spend $18 million over the next three years.
The aim of such a system, apart from allowing scientists at one location to
use an instrument at another, is to foster a 鈥渧irtual community鈥� with a sense of
team spirit as close as possible to the real thing. Researchers should be able
to bandy about ideas before the experiment starts, mull over data from an
experiment together, and even change the experiment in mid-stream if
necessary.
鈥淵ou really need the feeling that you鈥檙e there鈥攖hat you can reach out
and talk to the people, adjust the instrument and feel that you are a part of
that experiment,鈥� says Deborah Agarwal, who leads the development of
collaboratories at the Berkeley laboratory.
Different collaboratories have different methods of providing that sense of
鈥渂eing there鈥�. The ability to control instruments remotely plays only a small
part. Other ways include establishing videoconferencing and audioconferencing
links between remote users and the central laboratory, an electronic
鈥渨hiteboard鈥� that collaborators can use to draw diagrams to accompany their
comments, a chat system for exchanging text messages, an electronic notebook
that records data from experiments and software tools for writing documents as a
team.
Drop-in centres
Collaboratory developers are trying to include remote researchers in every
step of the life of an experiment, and not just its execution, says James Myers,
a senior research scientist at Pacific Northwest and leader of the collaboratory
project there. For example, scientists can collaborate on finding grant sources,
writing proposals, analysing data from an experiment and writing papers.
鈥淭he real collaboratory vision is to go beyond text exchange and
videoconferencing,鈥� says Myers. The goal is to allow researchers, who are
separated by hundreds or thousands of kilometres, to work together as easily as
if they were located down the hall from one another. Myers believes you should
be able to say: 鈥淗ey, I鈥檝e run into a problem while I鈥檓 doing this analysis of
my data. Can you drop in for a second and take a look?鈥�
The UARC, for example, focuses on the Sondrestrom Upper Atmospheric Research
Facility in Kangerlussuaq, Greenland. The station monitors 鈥渟pace weather鈥�, the
interaction of the solar wind with the Earth鈥檚 atmosphere. In the past, the only
way to use it was to fly up to Greenland. 鈥淵ou would sit in a trailer with
instrument displays from all the various instruments that you had running at
that time,鈥� says Gary Olson, associate dean of the school of information at the
University of Michigan and one of the architects of the UARC. 鈥淵ou would watch
what was going on in the upper atmosphere, and if a particular phenomenon
happened, you would point the radar in a particular direction.鈥� The researcher
would record data on computer tapes or discs and take these back to their home
laboratory, where they would be analysed over the next few months.
Instant feedback
When the UARC was set up in 1993, it gave distant researchers the chance to
obtain data over the Net directly from one of the instruments connected to the
radar. 鈥淚t allowed them a window into the world of upper atmospheric physics
from a particular location [in Greenland],鈥� says Olson.
Today, researchers can access data from five different instruments, which
between them cover many of the effects of the solar wind on the Earth鈥檚
magnetosphere and ionosphere. Data from three other upper atmosphere
radars鈥攊n Massachusetts, Norway and Puerto Rico鈥攁re also being fed
into the collaboratory. 鈥淭he whole focus of the project has changed to a global
view of what鈥檚 going on in the upper atmosphere,鈥� rather than a view just from
Greenland, says Olson.
The UARC software running on researchers鈥� home computers includes a
text-based chat window through which the researchers can talk about the data and
send instructions to technical staff in Kangerlussuaq. UARC is changing the
entire process of how space physics is done, says Olson. In the past, scientists
didn鈥檛 know what they had observed with the radar until months later, when they
finished analysing the data. But now other researchers can analyse the data
while monitoring continues, which allows the scientists in the field to change
their observations as they go along.
Moreover, from October the UARC will provide participants with links to
supercomputer models of the upper atmosphere. By feeding their observations into
these models, the researchers will obtain a real-time impression of what is
happening. Development of the UARC has spurred more collaboration, says Olson.
The number of space physicists participating in the research has trebled.
Despite the success of projects like the UARC, there are several hurdles to
overcome before collaboratories can be widely introduced. The main problem is
the fact that scientists can鈥檛 depend on the Net to provide enough bandwidth, or
data transmission capacity, to routinely send enormous amounts of data out to
the participants in a collaboratory. As the Net鈥檚 popularity has boomed, its
reliability has faltered.
During the 1995/96 observations of UARC, 鈥渨e had enormous difficulties with
system performance because of the incredible usage of the networks,鈥� says Olson.
Interruptions in an online discussion between scientists might be seen as a mere
annoyance. But if a researcher is remotely controlling an instrument,
transmission problems over the Net could stop a crucial instruction getting
through, and may even damage the instrument.
鈥淭he Internet is simply being used too much,鈥� says Bruce Schatz, a
collaboratory designer at the National Center for Supercomputing Applications,
based at the University of Illinois at Urbana-Champaign. This means that
collaboratories can鈥檛 guarantee communication without delays between distant
participants.
Video is essential for giving participants a sense of 鈥渢elepresence鈥� in the
distant laboratory, says Agarwal, who has a video link to the Berkeley
accelerator beamline on the computer in her office. It allows her to watch what
the physicists are doing as they prepare for experiments. 鈥淚 feel so much more
involved in what they are doing, just by being able to see this low-rate feed of
what鈥檚 going on at the beamline,鈥� she says. 鈥淚鈥檓 not just off in my own separate
building, building tools for them. I鈥檓 now a part of their environment.鈥� But
limited bandwidth forces collaboratories to restrict their use of
videoconferencing. For example, some videoconferencing systems try to conserve
bandwidth by reducing the rate at which pictures are transmitted to one every
few seconds. Such low-rate video is a result of the collaboratories鈥� reliance on
Mbone, a subsystem of the Net that is used for 鈥渕ulticasting鈥濃�攕ending
video among multiple Net users鈥攂ut only at a limited rate. Although having
any video at all is a plus, low transmission rates mean that if the camera
captures a participant at an awkward moment, in mid-yawn say, that picture will
be splashed across the collaboratory for a mortifying length of time. 鈥淵ou look
kind of silly,鈥� says Myers.
Low video rates also make it difficult for remote participants to signal that
they want to speak. The ways in which people signal to each other in
face-to-face communication, such as tilting their head, are lost in
videoconferencing, particularly if images are transmitted only every few
seconds, says Agarwal.
Videoconferencing presents other problems that may seem mundane at first
glance but which subtly prevent remote participants from seeming like full
collaborators. For example, videoconferencing requires a remote site that is
equipped with a camera and a television screen. A participant normally looks at
the television screen and talks back to it. But the camera is not located behind
the screen鈥攊t is placed at the participant鈥檚 side. This means that
collaborators only get a side view of their colleagues. 鈥淵ou look very
disinterested to someone on the other end,鈥� says Myers.
Eventually, participants in a collaboratory may end up wearing wireless
microphones so they can be seen and heard as they work and not have to worry
about remaining near a stationary microphone, says Agarwal. That would help
produce a sense of 鈥渢ransparency鈥� about the collaboratory. 鈥淚n a normal
environment, if you鈥檝e got to walk over to the soldering table and solder
something, the people that you鈥檙e having a discussion with just walk over with
you,鈥� she says. 鈥淲e want to create that same environment for the remote
耻蝉别谤蝉.鈥�
Tempting targets
Another major issue for collaboratory designers, says Agarwal, is security.
Many large instruments now rely on 鈥渟ecurity by obscurity鈥�. In other words,
potential hackers don鈥檛 know where instruments are located or how to find them.
鈥淭hat鈥檚 all going to change with these collaboratories,鈥� says Agarwal. 鈥淲hen you
go to the collaboratory environment, by its very nature you are publicising this
instrument.鈥� That publicity may make the instrument a tempting target for
someone who would like to break into it across the Web.
Such security concerns have made collaboratory designers wary about giving
remote users too much control over laboratory instruments. The Berkeley
collaboratory, for example, has been set up so that remote users cannot operate
an instrument in a way that could damage it, says Agarwal. She believes that the
solution to guaranteeing security may lie in public key encryption.
Videoconferencing, which is essentially open to anyone with the right
software, poses special problems. 鈥淧retty much anyone can eavesdrop,鈥� says
Tonner. He cautions people that they should think of it like speaking on
television or over a CB radio, but many don鈥檛 heed the warning. 鈥淸Even] I tend
to forget it,鈥� he says.
Yet another issue is how to make sure that remote participants in the
collaboratory have up-to-date software on their computers, to keep up with
changes in the software on the collaboratory鈥檚 central computers. To solve this
problem, says Agarwal, collaboratory designers are pinning their hopes on Java,
the new Internet computer language. Java 鈥渁pplets鈥�, or small programs that
execute specific tasks, can be instructed to automatically update themselves
from a central library without bothering the user.
Despite the technical and security challenges, collaboratory designers
believe that their systems will change science in all sorts of ways. For
example, graduate students in particular are benefiting from the UARC. Olson
says that students rarely have the opportunity to travel to Greenland where the
data are gathered. 鈥淚f you were a graduate student studying space physics, maybe
once during the time before you got your PhD, you might go to a site and
actually participate in real-time science,鈥� he says. 鈥淢ore often you sat at home
and your adviser went and came back with lots of data and you got to analyse
it.鈥� But with the UARC, students can participate in experiments and interact
with experts in their field across the globe, all without having to spend
anything on plane tickets.
Myers says that collaboratory technology could also enfranchise scientists at
small colleges that cannot afford expensive research facilities. Several
colleges could band together to form a 鈥渧irtual institute鈥�, linked by
collaboratory software. Together, those colleges could buy a sophisticated
instrument, and researchers at the participating institutions could share its
operation.
Agarwal believes that collaboratories will help to foster collaborations with
experts who might not have the time or the inclination to travel and meet other
scientists, but who might have expertise that the other scientists value.
鈥淐ollaboratories can provide new scientific opportunities,鈥� she says. 鈥淵ou can
collaborate with a scientist who cannot travel to your site or who cannot spare
even five hours [for a meeting] but who can spare 30 minutes to talk to you
about a result.鈥�
With this flexibility and such a focused approach to discussing research,
collaboratories might ultimately not just be the next best thing to being
there鈥攖hey may be better than being there at all.