No one likes to change. But if we continue to solve the same problems (in slightly different contexts) we will be replaces by others who view the problem differently. Or at least that will happen in the area of ubiquitous computing.
I'm spending my day in the dark insides of a big silver bird, at an
altitude and speed which are not natural for someone who is a member of
Aristotle's feather-less biped class. I'm in cramped quarters, finding
that the airline now wants money for the marginally more spacious seats
they used to give out for free to those of us who spend too much time in this
position. I've dined on a soggy sandwich that I brought on
board to avoid paying even more for a soggy sandwich that the airline
brought on board. The person in front of me has fully reclined her seat,
and is now strengthening her six-pack abs by doing a continuous crunch that
puts her into an upright position.
All of which, for some reason, has gotten me thinking about ubiquitous
computing. I guess it is either that, or chew my leg off trying to
We in the computing industry think that we have invented ubiquitous
computing, and to some extent I guess we have. But given our background,
we tend to have a somewhat strange view of the field. We often think of
our mobile phone, or how we can use mobile computing devices to do, on
the run, what we now do with our desktops when we are stationary.
All of that is fine and good, but we tend to ignore some other uses of
computation that have been embedded in the real world for some
time. There all all sorts of simple computing devices used to control
the heating and ventilation in most commercial buildings. There are lots
of sensors on all kinds of moving things (like this plane that I'm
trapped in). There are systems for getting telemetry data from sensors
on space vehicles. In point of fact, there are few places where
computing is more ubiquitous than those places bearing the NASA logo.
But those of us who do ubiquitous computing tend to ignore these other
groups, labeling them industrial automation or
telemetry rather than ubiquitous computing. They tend
to have computers that only talk to other computers, with a fairly long
set of filters before a human gets involved, if one gets involved at
all. They program in real-time languages on real-time operating
systems. They worry about things we've decided we don't need to worry
about (like very high reliability, and failure characteristics, and
determinacy). The work they do is really different from the work we do,
and we don't understand it, so we do the kinds of things that we are
good at, and that we understand (which generally looks like n-tier
computing with a human looking at some kind of screen at one end and a
database of some form at the other end). And we call
that ubiquitous computing.
I think we may be missing out on a lot of expertise and a lot of
interesting problems and techniques. If the best user interface is no
interface at all, then a lot of the industrial automation systems have
pretty good user interfaces. Telemetry systems deal with large amounts
of information in ways that let that information be used to make things
that are more reliable, more predictable, and more useful without
requiring the intervention of the users of the things that are
generating the telemetry data.
Every time I read a ubiquitous computing paper about delivering ads for
local restaurants to my car as I drive by, or systems that will tell me
(and everyone else using the system) where the best parking place is (can you
imagine the carnage this would cause in Boston?), I
have to admit that I'm not at all sure that I'm seeing the future of
computing. I'd much prefer sending a bunch of data to my car
manufacturer so that I could be emailed about a problem before
the problem happened. I don't want to see all the information, but I'd
love it if some computer in Detroit was analyzing the information,
looking for anomalies.
Of course, this would require that we think about new and different
problems, rather than clothing our old problems in new devices. It is
harder, and riskier, and takes us into areas where it isn't clear that
there are journals for publishing (for the academics) or a business case
(for those in industry). Maybe those doing industrial automation or
telemetry will come in and make all of our work irrelevant, showing
everyone who spends money on this sort of thing what ubiquitous
computing could be.
Which is just the sort of thing that is happening to the airline
companies like the one that I'm currently using. They haven't really
changed the way they do business for a long time, so the really cheap
carriers are undercutting them, and the luxury carriers are siphoning
off the specialized runs that are less price sensitive. The ones who
can't change make some minor alterations that only irritate their
clients, find themselves bankrupt, and try to keep flying anyway. If you
don't change, you may find that you are no longer able to survive.
But, of course, that won't/couldn't/isn't going to happen to us....