After hearing this I first thought about similar experiences I have had when I see computer scientists talk about ideas in physics, biology and economics that they seem to have really learned about the other field and tied the areas together well.
My next thoughts went in a different direction. When I have seen economists talk about computational models and complexity, my bread and butter, I just want to tear my hair out (or more accurately tear their hair out). How they get so excited about some stuff we consider completely obvious or known several decades ago, or worse using the completely wrong computational model for the problem they look at, for example examining the precise number of states of a finite automata when they shouldn't be using automata at all.
On the other hand I have heard many economists and physicists and biologists complain that (with some exceptions) they don't understand the relevance of much of the interdisciplinary research that comes from our community.
Part of this attitude comes from the general arrogance that you find among all academics. We computer scientists feel we know the right way to think about problems in all academic fields and researchers in other fields feel the same.
We give strong weight to interdisciplinary research, a good thing. unforunately we typically do this research in the context of impressing those in our own field with a system that encourages this behavior. It is our peers, other computer scientists, that will hire us, write us recommendation and tenure letters and review our grant proposals. So we find it more valuable to have STOC and FOCS paper than papers in good economics or biology journals that most CS people have barely heard about. And we give these talks mostly to our peers designed to impress them.
True interdiscplinary research is difficult enough because you have to deal with not just another field's language but also their culture about what kinds of problems one finds important. But we need a system that truly rewards those who can truly bring computer science ideas into other fields above those who just try to impress our own.
As someone in a research-oriented CS dept., originally trained as an electrical engineer and having spent several years in industry, I would suggest that this has something to do with a lack of breadth in the experience base of most academics' careers, and the constituent biases.
ReplyDeleteWhen I look back to my days in industry, I was solving problems that needed to be simultaneously looked at from two perspectives - the technical challenge of solving problems that were more or less open in the corresponding academic area, and the engineering challenge of making sure my algorithms did what they were supposed to in the final working product that would make money for the company.
I took away two lessons from this:
(a) If you want to do inter-disciplinary work, learn to wear many hats and figure out when each hat is appropriate
(b) When wearing the hat corresponding to a specific area, make sure that you are able to hold your own in that area (independent of the rest of your CV) and pull your weight in terms of the significance of those results.
I suspect that too many people engage in inter-disciplinary projects just because it is popular, a good source of funds, or whatever and do not really have the skills (or the inclination, which is worse) to engage in this way.
Most often, the models we use are the ones the people before us used. Not pretty, but there you go.
ReplyDeleteSo. I'm in CS and unfortunately my knowledge about complexity is restricted to what you learn in your first-year algorithms course. However, I still would like to answer things such as whether an FSM is the right model for my computations. Where do I start?
It's also important to recognize that what other people think constitutes the right question, and what constitutes a good answer, can be very different from what CS theory people think. For instance, to people in other fields the idea of an algorithm with a worst-case guarantee of an approximation ratio seems rather strange: what they are usually interested in is an algorithm, or we would say "heuristic", which works extremely well on the instances they care about --- and they don't care at all if it works extremely poorly on a few adversarially chosen instances. This kind of cultural gap can be confusing and frustrating for people trained as CS theorists, but it's important to be flexible about this sort of thing when you engage with people from other fields.
ReplyDeleteThere's a thesis in there somewhere, still:
ReplyDeletehttp://arxiv.org/abs/adap-org/9910002
LOL! ... and thank you Bill for the pointer to your fine article, which I greatly enjoyed!
ReplyDeleteLike much high-quality humor, your essay makes a serious point (is P. G. Wodehouse the main exception?).
You wrote your essay in 1999 ... now it's nine years later ... so maybe it's time to check whether your humorous ideas are evolving into serious enterprises?
One such enterprise is SynBERC, an NSF-sponsored enterprise whose explicit goal is to make biology into an engineering discipline.
SynBERC is governed by the fundamental organizing principle that (as SynBERC's Paul Rabinow says) "philosophy is the ability to make friends through the medium of a written text." Isn't that a great quote!?
AFAICT, SynBERC is conscientiously extending this principle from the domain of philosophy to the domain of synthetic biology---which obviously is quite a radical and innovative (even transgressive) step forward, and yet (equally obviously) is a necessary step, not only for SynBERC, but for any global-scale enterprise.
To make my point explicit, isn't laughter one of the very best ways to being making friends?
And under the mapping text→algorithm don't we obtain a good working definition of 21st century computer science?
Bill, it seems to me that your late 20th century article (and others like it) helped initiate a cascade of thought that has led to a broad class of exciting (and radically new) 21st century global enterprises like SynBERC.
This cuts both ways. I know at least a couple of situations where it has been an issue of language not content. Namely, the significant contribution of an outsider is overlooked because the language in which it is expressed is too unfamiliar. Only after someone redid the work independently or rephrased the arguments so that they could be understood did the community take notice.
ReplyDelete