¿ìè¶ÌÊÓƵ

Almost everything you do today will be forgotten in just a few weeks.
The ability to retrieve a memory decays exponentially, and after only a
month more than 85 per cent of our experiences will have slipped beyond
reach, unless boosted by artificial aids such as diaries and photographs.

Given that our memories are our identities, this is a frightening rate
of loss. But what if a machine could record our lives, keeping nearly everything
that happened accessible to us? It is likely that computer technology and
video cameras will continue to get cheaper over the next decade, so why
not build electronic memory aids that simply record every waking moment?

At Rank Xerox’s EuroParc laboratory in Cambridge, a team of 30 computer
scientists, psychologists and sociologists have been working on an electronic
memory prosthesis. Even as a crude prototype, Xerox’s technology is impressive.
More startling still is the glimpse it gives of a world to come: a world
where computers are ubiquitous, pervasive – intimate, even. A world where
computers begin to enclose us in an electronic womb, becoming seamless extensions
of our own minds.

Despite its grand ambitions, EuroParc is an easy place to miss. A small,
four-floor building standing anonymously among furniture shops and sandwich
bars in a busy Cambridge street, the European offshoot of Xerox’s famous
Palo Alto Research Center (PARC) in California looks more like the offices
of a moderately successful solicitors’ practice than a playground for blue-sky
visionaries.

Cluttered store

Inside, EuroParc is a dimly lit warren of rooms, cluttered with mountain
bikes, straggly pot plants, inflatable animals, dog-eared posters, and all
the usual debris of academic life. The only clue to the laboratory’s greater
mission lies in the Heath Robinson assortment of computers, video recorders
and cameras, TV monitors stacked high in the corner of every office, and
the thick, metal cabling ducts snaking their way across every wall.

Mik Lamming, a senior researcher, rummages in what appears to be a
pile of telephone pagers, looking for one with enough battery charge to
demonstrate Forget-Me-Not, the laboratory’s latest electronic memory project.
While he searches, he talks about the potentially huge market for ‘wearable’
computers that augment human memory.

Personal experience tells us that our memories provide a far from perfect
record, he says. We forget most of what happens, and even what we do remember
is largely a sketchy reconstruction, based on only a few hard facts.

A favourite example of memory’s reconstructive capacity among psychologists
comes from the Watergate hearings, in which President Nixon’s legal aide,
John Dean, was asked to recall dozens of meetings he had with Nixon as the
bugging conspiracy unravelled. Dean’s recollection of these meetings seemed
so detailed that it earned him the nickname of the human tape recorder.

But when Nixon’s own secret tapes of the Oval Office meetings were discovered,
it turned out that Dean had got the gist of the conversations right, but
the actual words and dates were very different. Dean’s 245 pages of testimony
were a much condensed and tidied-up version of what had really happened.
Dean had also rewritten events to give himself a more central role. Crucially,
Dean believed his memories to be accurate, though the recording equipment
told a different story.

Reliable records

With human memory so fallible, Lamming says there is an obvious market
for artificial memory aids. Computer-based recording systems for straight
storage could give us a complete electronic record of our lives, but computers
could also be used to navigate around our own biological memory banks.

Over the past six years, EuroParc has explored these ideas by building
several memory prostheses. One project, NoTime, combined a camcorder and
an electronic notepad to record meetings. Every pen stroke made on the notepad
was time-stamped so that if people later wanted the full text behind a hasty
jotting, they had simply to click on the handwritten note to bring up the
relevant segment of video tape.

Another system, Marcel, was designed by William Newman, a principal
scientist at the laboratory, to keep track of paperwork passing across a
person’s desk. A video camera mounted over a desk was connected to a computer
recognition system that could automatically follow the movement of documents.
Lamming says one American study showed that so many bits of paper get lost,
misfiled or destroyed in big companies that office staff have only a 1 in
10 chance of retrieving any particular document within a reasonable time.
While this may be an exaggeration, Lamming says many people could still
do with a tracking aid which could tell them that a vital invoice was last
seen fluttering off the corner of their desk.

With video equipment and computers improving all the time, it is easy
to contemplate even more advanced systems. Lamming says many have joked
about the day when we will walk around with tiny camcorders glued to our
foreheads, recording every act for posterity. But many less obvious kinds
of memory prosthesis are to be found among EuroParc’s memory-jogging devices.
Rather than providing a raw record of life, they are tools to mine our own
memories for lost information.

Recent research by Marge Eldridge, an independent consultant for EuroParc,
shows just how quickly our lives slip beyond the reach of conscious recall.
Subjects could remember about 15 significant events from the previous day
but only about two events from a similar kind of day a month earlier. So
dense is the cloud of amnesia obscuring the past that much of life might
just as well not have been lived, for all we remember of it. On the other
hand, it is common for an old school photograph or love letter to bring
back a flood of long-forgotten memories. Though poor at deliberate recall,
our memory banks respond powerfully to the right kind of trigger.

Lamming believes technology could exploit this difference between the
weakness of unaided recall and the power of prompted recollection. Memory
systems could provide a kind of scaffolding to prop up our powers of recall,
recording enough detail about our lives to be used as an external index
file on our biological memories.

The first of EuroParc’s context-based information retrieval systems,
as Xerox has dubbed them, was the Pepys project, completed in 1991. Pepys
was a way of generating individual electronic diaries from a record of a
person’s movements around specially wired rooms. EuroParc staff wore ‘smart
card’ identification badges which acted as beacons to the system, broadcasting
their location to infrared sensors in every room. This raw information was
collated by the Pepys software to give a minute-by-minute log of everyone’s
whereabouts in the building. Pepys then uses the patterns of movement to
try to guess what might have been happening. For example, if several people
were in the same room for a similar length of time, Pepys would interpret
this as a meeting.

Each day’s activity was processed overnight and a personalised diary
mailed to subscribers on the EuroParc computer network. One person’s entry
for a particular afternoon might read: ‘16.05: Attended part of event in
common room with Andrews, Morton, Hatton (7 minutes). 16.13: Mostly in
office (44 minutes). 16.57: Attended event in Wright’s office with Wright
(7 minutes).’

Lamming admits this sparse record hardly constitutes a fully fledged
memory prosthesis, but he says that even this humble automatic diary could
find wide application. ‘There are a huge number of people who have to spend
time clocking what they’re doing – there are service engineers who must
log how long they spend on site fixing an appliance. This sort of system
could do the job automatically.’ More to the point, says Lamming, a Pepys-type
activity log provides a foundation for a full-blown memory prosthesis system.

Lamming’s next memory project was Vepys, an extension of Pepys which
took advantage of the video cameras installed in nearly every room at EuroParc
as part of its videoconferencing network. The camera system was used to
take single-frame snapshots of people at 10-minute intervals throughout
the day, homing in on their smart-card badges to catch them wherever they
were in the building. The combined video still and activity log could be
used by EuroParc workers to jog their memories about what they were doing
that day.

In a test of the system, Eldridge used staff to compare unaided memory
with memory prompted by Pepys and Vepys. With the electronic aids, the number
of significant events recalled from the previous day rose from 15 to 35
and from 2 to 12 for events a month earlier. Not only did people remember
six times as much after a gap of a month, but as with the Watergate tapes,
the electronic aids made them realise that often their own memories had
jumbled up dates and details. It was a shock to find that the electronic
record was contradicting their own recollections.

Pepys and Vepys proved the potential of electronically extended memory.
But they also made clear that while capturing data was simple – certainly
in the controlled environment of an office – using it effectively was another
matter entirely. Sitting at a terminal flicking through hundreds of electronic
diary pages or thousands of video frames is hardly efficient, Lamming points
out. A true memory aid must be both portable and almost as easy to review
as our own memories.

With these problems in mind, Mike Flynn, a consultant for EuroParc,
is working on Forget-Me-Not. This latest project, which began trials in
January, makes use of Xerox’s ‘tab’ technology, the telephone pager-sized,
pen-driven computers that are Xerox’s version of a personal digital organiser.
Tabs can be clipped onto clothing and should allow EuroParc staff to look
up Pepys-style diary information at any time. Information can be called
up over the building’s infrared communications network.

Memory joggers

More importantly, Forget-Me-Not will have a rich cross-referencing system
to help search out memories. Where Pepys offered just a straight activity
diary of where a person was and who they were with, Forget-Me-Not will
also store information about what the weather was like that day, what was
in the news, who phoned and which computer files the person opened. This
information can be acquired automatically from local weather centres, news
feeds and computer audit trails.

‘In a database search, the depth of description is a lot less important
than the breadth. That’s the way human memory works. Even if we can’t go
straight to a specific memory, we can usually remember lots of bits and
pieces surrounding the event – like I had a hamburger for lunch that day,
or it was raining,’ says Lamming.

Of course, by itself, Forget-Me-Not is little more than a navigation
tool: a database search engine designed around the natural characteristics
of our own memories. Lamming says the real benefit of a Forget-Me-Not style
system would come from using it as an access mechanism to call up electronically
stored memories such as those gathered by a NoTime or Marcel device which
record meetings and the flow of paperwork. It is this combination of technologies
that would allow us to use electronic memories as an efficient extension
of our own.

There is a long way to go before we enter a Buck Rogers world where
people walk about with memory prompts strapped to their wrists while video
recordings of past experiences are projected onto their retinas by contact
lens-mounted laser-diode scanning guns. Even so, memory prosthesis technology
may find a use sooner than many might suppose.

Lamming says an obvious application would be a Pepys-like recording
system to help elderly people cope with degenerative memory disorders. And
even rudimentary memory systems would be valuable in the office, especially
where document retrieval and action recording are important. And there
will always be unexpected applications – recently Lamming was button-holed
by an officer in the US Army asking for a diary system to track the service
life of costly items of military hardware, such as tanks and missile launchers.

The storage requirements of the office systems being developed at EuroParc
are surprisingly trivial. Lamming says that 40 megabytes of computer memory
is enough to store a year of Forget-Me-Not data on one person. ‘You’ll be
able to hold that much on a flash card (memory card) in a year or two,’
he notes.

A much greater barrier to the commercialisation of memory technology
is the equipment used for recording environment. Lamming says people would
need to carry electronic beacons to record their movements linked to a centralised
network to cross-reference their activity with that of others. It is easy
to see a company paying to wire up a workplace to allow this, but far harder
to see where the infrastructure for a general-use memory prosthesis network
might come from.

Outer reach

‘You’ve got a problem as soon as you move out of doors, into public
spaces or into homes. But I think the infrastructure for doing this could
come gradually, piggy-backing onto other information or entertainment services.
People might say that they would happily pay 10 bucks to have the time of
the next bus displayed on this device on their wrist. Then, once the system
was in place, you could start adding other applications to it. An infrastructure
would emerge,’ says Lamming.

Ultimately, the real questions are social, not technical. At their
most dazzling, memory prosthesis systems and some of the other intimate
computing technologies under research threaten to redefine what it is to
be human. Beyond the perfect electronic memory may lie the networked memory
– what would it be like to have access to the recorded life events of other
people? Intimate computing could blur not just the boundaries between humans
and machines, but perhaps between individuals as well. Even disregarding
such futuristic fantasies, the implications of filling every room of a building
with cameras and tracking devices make Orwell’s Big Brother pale into insignificance.

Another EuroParc project, ‘awareness servers’, highlights both the promise
and the dangers of the coming wave of intimate technology. The idea behind
awareness servers (an idea being pursued by others apart from Xerox, such
as the University of Toronto and the US telecommunications research company
Bellcore) is to create the virtual open-plan office, using a videoconferencing
network to give people the feeling of sharing the same room even when they
happen to be working on different continents.

A camera in nearly every office at Xerox’s labs in Cambridge and Palo
Alto allows researchers to make the video equivalent of phone calls to each
other. Having established a basic network, Xerox has added other services,
giving the kind of peripheral awareness that usually comes only from sharing
the same office and which is known to be vital to the smooth running of
any team.

One example of an awareness service is the OfficeShare link, in which
the connection between two people’s monitors is simply left open all day
so they can chat to each other as freely as if they were sitting at adjacent
desks. Another service is Glance. At the press of a menu button, researchers
can take a quick three-second peek into another room – analogous to poking
your head round the door to check if someone is in.

A refinement of Glance, a service called Portholes, allows a researcher
to keep a continuous, low-level eye on the whereabouts of a dozen or so
colleagues. A set of video pictures of selected offices are displayed in
a window on the researcher’s computer terminal. These images are called
portholes because they are just low-grade video snapshots, refreshed once
every three minutes, rather than continuous video pictures. Even so, these
small, fuzzy snapshots are enough to give a general idea of when people
are in their offices, what they are doing, and whether they are likely to
mind being disturbed.

The final video service is Background, in which a monitor is left hooked
up to a public space camera, such as the one in the top floor meeting room
where EuroParc staff gather for their coffee and seminars. This helps staff
keep in touch with the general flow of events in the lab, letting them know
when a meeting is about to start – or rather, allowing them to put off leaving
for a meeting until they have seen everyone else arrive.

The potential of awareness systems is huge. Not only would they allow
people working in different locations, and even from home or on the move,
to feel part of a close-knit team, they also create a special flexibility.
Team structures can be set up for a year or just for an afternoon, and people
might move through many different structures during the course of a day.
To use Xerox jargon, people would no longer be confined by physical location
but would meet in the ‘virtual world of media spaces’. Once again, the risks
of abuse of such ‘eye in the office’ technology, both by overbearing managers
and voyeuristic fellow workers, are only too obvious.

Far from shirking Big Brother questions, EuroParc has made them central
to its work. Its director, Bob Anderson, says that the impact of such technology
is likely to make preserving privacy fundamental to the design process.
Safeguards must be built into the systems, not tacked on as afterthoughts.

Psychologist Victoria Bellotti is one of several EuroParc staff paying
special attention to the privacy problems of intimate computing systems.
She points out that the threat of workplace surveillance is not a new issue.
Even in the 1930s, it was possible to build keystroke counters into typewriters
and, of course, time clocks were common. More recently, itemised phone billing,
computer audit trails and security cameras have all been seen as insidious
steps towards tighter management control over employees. And Bellotti concedes
that a memory prosthesis system creating a permanent record of a person’s
every action does raise the stakes somewhat.

So all the EuroParc systems are designed with two safeguards in mind.
The first is feedback: people should be continually aware of what information
about them is being captured and to whom it is available. The second is
control: people should be allowed to say what information can be captured
about them and by whom – to the point of refusing to participate.

With EuroParc’s awareness servers, for example, the kind of links that
other people can make to each office are controlled by the user. Bellotti
can set her terminal so that any staff member can call her up on a videophone
link, but only certain named colleagues can use the more informal Glance,
OfficeShare, and Portholes services.

The various links also offer feedback. A colleague may take a quick
peek into Bellotti’s room at any time, using a Glance connection, but each
glance is preceded by the synthesised sound of a creaking door and the system’s
announcement of the person’s name. This gives a brief warning of the coming
intrusion – long enough to take your finger out your ear, says Bellotti
– just as if a person was poking a head around a real door. The system also
makes a synthesised camera shutter click shortly before taking a video
camera snapshot for the Portholes service, or for Lamming’s video diary
project (Vepys).

Sneak preview

These privacy safeguards have been discovered mostly by trial and error,
with EuroParc staff acting as their own guinea pigs. Bellotti found one
particular design flaw on a day she was linked to a male colleague on the
floor above by an OfficeShare connection: ‘We were both changing for jujitsu
and he covered his lens with his jacket so it blanked out my view of his
office. I wasn’t thinking and assumed that because I couldn’t see him, he
couldn’t see me. I forgot my camera was still on,’ laughs Bellotti. EuroParc
video systems now come equipped with a second screen, a small ‘confidence’
monitor, which displays the outgoing signal to remind staff of the image
they are projecting.

Bellotti says that when she first arrived at EuroParc, the idea of cameras
and tracking devices in every office struck her as rather creepy, but in
practice it has been possible to build sufficient controls into the systems
to make them trustworthy. Also, says Bellotti, it is surprisingly easy to
get used to the idea of working in the midst of an electronic web. Like
newcomers to a nudist camp or an encounter group, most people are horrified
by the thought of such self-disclosure. But most soon forget their inhibitions
– especially if they feel the technology is there to serve them rather than
spy on them.

Bellotti admits that the relaxed atmosphere of a research laboratory
may not be the most realistic test of the Big Brother question. At EuroParc,
no managers lurk in the background, poring over daily charts of how long
workers spend seated at their desks and how long chatting to friends. But
the feeling at EuroParc is that there are two good reasons for thinking
that intimate technologies are likely to be used responsibly.

The first is that every management textbook now stresses that the secret
of greater productivity is to work smarter rather than to work harder. Successful
companies are moving away from a tight, hierarchical control of workers,
and towards small self-managing, self-motivating teams. When information
technology is installed in such companies, the aim is to extend the reach
of individual workers, not to impose yet another deadening layer of bureaucratic
control upon them.

The staff at EuroParc are optimistic that sense will prevail, but even
if enlightened self-interest does not prove enough of a safeguard, the ultimate
line of defence is likely to be what Lamming calls the ‘neo-Luddite’ response:
‘Humans are endlessly inventive about finding ways of defeating a system
that they don’t like – especially a complicated, hi-tech one. If people
objected to having their movements monitored, you would probably find that
their badges kept mysteriously breaking or getting lost. There would be
resistance and things just wouldn’t work.’

John McCrone is a science writer and author specialising in technology
and psychology.