On the Nature of Computing

I would like to propose that the natural agenda for Computing
is the Virtual, in contrast to the Natural and the Artificial.
Let me explain.

Over the years, one encounters periodic bouts of navel gazing in the
Computing community about the nature of the business we are engaged
in. Typically, the debate polarises people along a spectrum (to mix
metaphors) between Engineering, and Science.

The ends of the spectrum are usually occupied by Practice (Systems,
Networking, OS, software engineers and so forth) and Theory
(Computability, Complexity, Information Theory, etc). An extreme
view of each end places practitioners within EE departments, and
theoreticians within Mathematics departments respectively. 

I studied Natural Sciences in Cambridge as an undergraduate - we were 
taught the value of study of the natural world, and the use 
(and advance of mathematics) to describe and understand (and predict) its behaviour. 
I have also spend over a decade teaching courses in an Electrical Engineering
department, where artificial systems are built according to models
(often mathematical) that have reliable behaviour. Computing has never
lain on a simple line between these two, and nowhere is this clearer
than in the confusion often facing PhD students in their first years,
trying to select a problem to attack in their thesis work - this is
the key to understanding where Computing is a third place in the world
of discourse, distinct from the natural and the artificial of sciences
and engineering: computing is about creating systems that may never
exist, either in nature or by creation of humans. These systems are
virtual. PhD students find it hard to find a topic, because the
possibilities are endless, and the topic may have no intersection with
the real world, either in understanding a phenomenon, or in creating
an artifact. Here I completely disagree with Richard Feynman (Footnote 2).

Of course, at some point, much computing does result in a model of
some thing. An object or process which does interact with,
or describe  the real world may clearly be the outcome, but it does
not have to be. (Footnote 1)

This has a very important consequence when explaining what it is that
we do, to the public (whether to school children, to non-computing users in
general, or to funding agencies and decision makers). Unlike the artificial,
engineering, some of what we do may not be obviously useful, and
therefore attractive to commerce and governments in optimising social
welfare or profit. Unlike the natural, science, some of what we do may not necessarily 
be simply "for the advancement of pure knowledge", and therefore 
a priori worthwhile! In some sense, though, what we do underpins both
of these other more worldly activities. I really like the claim that
Computing is less worldly than, say, Cosmology. On the other hand, due to the
possible use of Computing as an underpinning, we can also claim to be more
useful than Engineering.

Some Examples

I would like to take several examples of the virtual from the history of
computer science to illustrate my argument. 

Within the discipline of computing itself, we use the concept of
virtualisation as a first class tool. When confronted with intransigent
engineering limitations of memory, processors, I/O, and networks, it has been
commonplace for decades to abstract: We create virtual memory systems to
replace one piece of hardware with another as needed to overcome 
capacity/performance problems and to choose when to do so appropriately;
we replace inconvenient low level processor interfaces (the instruction set)
with virtual machines (VM, vmware, Xen, Denali), to provide a more convenient
(and stable) interface for the systems programmers; we provide a single API to
all I/O devices so that programs need no longer worry whether (say) an MP3 is
being loaded from a tape, a magnetic disk, flash RAM, or even networked media;
we replace a network with a virtual private network, so that the users can
behave as if they were in an Internet of their very own.

In the recent world of GRID computing (and especially within the UK e-Science
program), we are creating virtual communities of scientists with virtual
laboratories, with computing resources dedicated to supporting "in silico"
experiments to replace the expensive, error prone and risky "in vivo" or "in
vitro" experiments of the past: here we have virtualised natural systems
whether fluids such as atmosphere, ocean, or plasma etc,, or complex
biological systems such as genomes, proteins, or even ecologies.

However it is in the world of entertainment that the convergence of computer games and the movie
industry has led to the clearest evidence to support my view that Computing is
a wholly new discipline. Here, there are no natural or artificial
constraints on what a system may do. The limit is the imagination of the creator, combined with the 
knowledge and skills from the discipline that is computing. Of course, there
are constraints imposed from the discipline itself (e.g. computability,
complexity, or just plain affability!), but may often be orthogonal to the
goals of the computation (if there are any goals!).

Historically, there have been simple examples of virtual worlds (both for
games and online environments, or for playing with alternate realities, such
as artificial life games and so on), and so this is not supposed to be
something that suddenly became true - it has always been one of the exciting,
but difficult aspects of working in Computing that the bounds are not set from
outside, but by our own choices of what to work on.

Conclusion

To conclude, Computing occupies a third place which involves creativity and
imagination, as it is not bound by the needs to describe what does exist (as
is natural science) or what can be built in the real world (as does
engineering), and that place is the virtual. We do not need to be complete or
consistent or correct, but we possess the tools to choose to be at least any one
of these, if we wish.

Additional.

Footnote 1
I am tempted (as Penrose does in his latest book, lamentably) to lay
claim to the term "magic". A lot of what we do is now certainly 
seen by the lay public as magical - programmers (especially systems
people) are often referred to as Gurus, and Sorcerers or Wizards.
Given the high goals of white magic, understanding the power of names,
the value of pure thought, and so on, this is indeed very attractive.
However, there are in fact many historically compelling reasons to
avoid this connotation, including the sad history of Newtons later
days in alchemical pursuit of the philosophers stone and eternal life,
and, of course, the events at Salem and the like!


Footnote 2.
"Computer science also differs from physics in that it is not actually 
a science. It does not study natural objects. Neither is it, as you 
might think,  mathematics; although it does use mathematical reasoning 
pretty extensively. Rather, computer science is like engineering - it 
is all about getting something to do something, rather than just 
dealing with abstractions as in the pre-Smith geology.":-
Richard Feynman from the Feynman Lectures on Computation.

Reference 1
The Road to Reality: A Complete Guide to the Laws of the Universe
by Roger Penrose, pub. Jonathan Cape, July 29, 2004, ISBN: 0224044478