We need role models for diversity of deep uses
of computational thinking more than role models
for diversity of sex of thinkers
Message posted to the Computing at School Forum: 1 Dec 2013
(Unfortunately forum contents are available only to subscribers.
This hastily-written version is publicly accessible and will be corrected and updated
in response to comments.
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/girls-in-computing.html
)
This message was written as a comment on a lengthy discussion of whether and
how to provide more female role models in order to attract more girls into
computing.
I'll try to explain why I regard the discussion as seriously misguided, partly
because it ignores history, and partly because it ignores the future.
Some of these ideas are shared with the President of the Royal Society, Paul Nurse,
who emphasises the importance of information (in contrast with matter and energy) in
controlling processes in complex systems. That's the fifth great idea for the future
of scientific understanding in biology, in his presentation here (especially from
about 50 minutes into the lecture onwards):
http://www2.warwick.ac.uk/newsandevents/events/distinguishedlecture/paul-nurse/
History and future
1. There used to be a higher proportion of girls applying for CS degrees before the
teaching of CS as IT grew in schools. There were also national differences. By the 1990s
the proportion of girls doing CS in UK universities was much lower than in some other
countries (e.g. Poland, where I was told they were closer to 50%). I am fairly sure
this difference was not a result of role models but of the content of what was
taught. The reduction in numbers of females choosing CS here (in the 1980s-90s?)
could not have been a consequence of reduction of number of female role models.
2. I expect that 100 years from now computation scientists will refer scornfully to
the century when educators, politicians, and industrialists thought computing was
mainly about making useful or entertaining things, because they failed to understand,
and to teach, the deep relevance of computational thinking to a wide range of
disciplines
and hard scientific problems: e.g. what is life?, how does evolution work?
how can a fertilised egg build a brain? what is a language? What mechanisms are
required for languages to be learnt, used, taught, revised (not Java, Python, Lisp,
etc., but French, Swahili, Urdu, English, Russian, etc.)?, What are the relationships
between minds and brains, and what sorts of computational powers do brains require in
order to make various kinds of minds possible (insect minds, squirrel minds, elephant
minds, human baby minds, human toddler minds, minds of philosophy professors)? What
biological changes made it possible for our ancestors to start making mathematical
discoveries, eg. about lines, circles, triangles, cubes, spheres, cones, ellipsoids,
and not just empirical discoveries but results that could be proved to have no
exceptions? What information processing mechanisms produce and modify motivations,
and allow motives to be achieved, and conflicts between motives to be resolved
(sometimes)? What sorts of computational mechanisms make it possible for humans at
various ages, other animals, and future robots to have emotions, moral values, deep
concerns, deep interests, and passionate social ideals? What sorts of computational
processes make it possible for a working mind to become puzzled about something
discovered, and then to investigate the problem and come up with a new explanatory
theory, and go on testing and modifying the theory to make it broader and more
accurate? What computational mechanisms make it possible for minds to enjoy hearing,
or composing, or reciting, stories, poetry, songs, piano sonatas, orchestral music,
and even pop music?
3. Given that computers that we now use differ in so many ways from the very earliest
computers (size, speed, memory capacity, amount of parallelism, variety of
interfaces, network functionality, self-understanding and self-modification, variety
of programming formalisms, ...) in what ways might future computers differ from the
ones we now make and use, and how might they have to change in order to be able to
replicate some of the kinds of computation done in animal brains, minds, sense
organs, motor control subsystems, etc?
4. How many diseases, genetic abnormalities, effects of brain damage, forms of
treatment, types of recovery, depend on forms of biological computation that we don't
yet understand, and what attempts are being made to understand them, with what
results? E.g. Cancer studies are now increasingly computational.
I've probably gone on long enough. My main point is that if we want to get more girls
interested in computing we had better start emphasising far more than in the past the
growing and increasingly central role of computational thinking in a wide range of
fields of academic/scientific study including subjects as diverse as literature,
philosophy, psychiatry, evolution, developmental psychology, education, and cancer
research.
If we merely tell them they can have fun, get jobs in industry, contribute to
national productivity, make useful machinery, meet interesting fellow workers, get
degrees in computer science, then they will rightly respond: So what!
Start from things they really care about (e.g. how minds work) and show the relevance
of computing to that, and then perhaps things will begin to change.
But because this has not been done enough in the past, the people who are now
involved in teaching computing are mostly not the ones who are interested in or
knowledgable about these topics.
The problem is not to find role models but to find the right sorts of inspiring
teachers.
We also need to introduce far more variety, including varieties of programming
languages, styles and architectures, and varieties of forms of computation, so that
future science and invention are not held back by a straight-jacket based on current
fashions.
We must stop emphasising coding (as if the study of literature could be based on the
study of how to assemble character strings). Instead we must start including the
importance of construction of explanatory implementable theories that can be
tested partly by running them, and partly by varying them.
Some relevant links
(To be expanded)
_________________________________________________________________________________
Maintained by
Aaron Sloman
School of Computer Science
The University of Birmingham