Letter to fellow UK computing researchers about the proposed new Academy of the British Computer Society (BCS)

From Aaron Sloman Fri Oct 16 12:48:12 BST 2009
To: UKCRC Subject: UKCRC and BCS Academy of Computing

Comment on some draft documents.

You can probably guess that I would be very disappointed about the lack of any explicit mention of the enormously rich and deep interdisciplinary connections of computing, and I don't mean that people in all other disciplines use computers and software as *tools.*

There is an opportunity for the academy to play a major role in a process that is happening anyway, but could be happening with more clarity, consistency and depth, namely the use of ideas about information, information processing, and computation in forming questions, theories and explanatory models in a wide range of disciplines, including not just the empirical sciences but also philosophy and mathematics.

Here's one example. Our school of biosciences recently had a visiting speaker who talked about the problems Darwin had defending (with help from Huxley, etc.) his claims that humans had descended from the great apes. It seems that the only kind of evidence they could present was evidence about morphological similarities (e.g. pictures of skeletons) and also reports of more or less human-like behaviour in a variety of species. There is a very useful online version of Darwin's Descent of Man

    See chapters III and IV for example:
    III - Comparison of the Mental Powers of Man and the Lower Animals
    IV - Comparison of the Mental Powers of Man and the Lower Animals (continued)
For anyone deeply engaged in recent developments in cognitive science and cognitive robotics it is clear that Darwin's conceptual framework had severe limitations that made it impossible for him even to think about the core questions: what changes occurred in information-processing mechanisms, architectures, forms of representation, and develomental processes creating them, between different species?

Of course, the problems are huge because information processing systems don't leave fossil records and are very difficult to observe -- brain imaging systems merely provide pale shadows of what's going on.

Thinking about those things is totally different from thinking about physical changes, and thinking about similarities in behaviours. It requires an appreciation of ways in which the right sorts of information processing mechanisms (including layers of virtual machinery) are capable of explaining complex observed behaviours and altering themselves dramatically as they develop and learn -- and also evolve.

Thinking about those things leads to entirely different questions about similarties and differences between species from the questions that Darwin was struggling with. It also explains why his opponents (rightly) felt that the kinds of evidence Darwin, Huxley, and others produced were not nearly adequate to show that human *minds* evolved from ape-like minds.

As it happens, I am collaborating with bioscientists who are now trying to wrap their heads round those ideas and I see how difficult it is for them, and how much better equipped they would be to do psychology and biology if they had had a deep education involving designing and and testing *working* information-processing systems. (E.g. as required for perception, motive generation, motive selection, planning, problem solving, reasoning, ontology development, language understanding, social interaction....)

It might be reasonable for computer scientists not to want to bother: let all those non computing people do their own learning and teaching, etc. while we get on providing the material from which they can learn, if that's what they want to do.

But insofar as some of the engineers are trying to build machines that can interact deeply with humans, and in some cases perform as well as or better than humans, then that policy is obviously deep folly, since computer scientists will need a lot of help from people in other disciplines, e.g. in formulating the problems and constraints.


I have given one type of example. I think there are many more cross-disciplinary overlaps that the academy, if it is to be worthy of the name, can promote, involving various other disciplines.

One that particularly interests me is getting philosophers thinking about how to characterise virtual machines and how the networks of causation that they support are deeply different from the kinds of causation in the underlying implementation machines. This seems to me to have deep implications for theories of causation and theories about mind-body relationships that are at the core of much philosophy.



If the academy decides to take this commitment to fostering *deep* interdisciplinarity on board then it should be made very explicitly, perhaps wherever phrases like 'advance computing' are used. e.g. something like 'to advance computing and its interactions with related disciplines'.

The Academy could take on a commitment to including as board members at least one third (?) of people from other disciplines who also have a deep commitment to the development of the interfaces.

It should be committed to separating membership of the Academy from membership of BCS, with fees low enough to attract people from other disciplines, and free/open access to interdisciplinary publications.

It could take on a commitment to attempt to alter computing education in schools so as to support this kind of role for computing science, in addition to the other more obvious roles.

(Simon Peyton Jones and Andrew Herbert are already involved in interactions with school teachers, most of whom seem to me to find this point very hard to understand -- because they are products of the current one-sided educational system.)

Unfortunately, my guess is that I am in a minority of one in thinking this should be a key feature of the academy.

And I am not even a member of BCS.

So I'll shut up now.


Last Modified: 16 Oct 2009
Created: 16 Oct 2009
Maintained by Aaron Sloman