Can Computational Modelers Be Nice To Each Other?

K. Schulten, Dinner Talk, Sept 21, 2009

Dear Friends!

Suited to this serious moment, our symposium dinner, I will approach the topic systematically. Surely you did not expect anything less. Open your notebooks and get your pens ready.

Computing touches on many disciplines as best explained through this question: How much is one plus one? The mathematics graduate student (RA) says: the solution exists; the computer science RA: depends on the processor; the engineering RA: 1.99999; the physics RA: pi / 2; the biophysics RA: what number do you need?; the PostDoc: 2, but I don't know if I can convince the referee; the biology RA: don't be obscene; and the psychology RA: I don't know, but it's good that we talked about it. This diversity of answers illustrates the disciplinary richness of computational biology.

To those of you who think psychology, in particular, is not involved, I offer the following true story. Once I received a referee report for a submitted paper that began: "This paper is the best psychoanalysis of a protein I ever read" … In other words I am a practicing psychologist. Let me then psychoanalyze our field.

You might say that only people can be psychoanalyzed, maybe proteins, but not a field. Sorry, but you would be wrong since any field is formed of people, say N of them. If N = 1, obviously, a field can be psychoanalyzed; same for N = 2, 3, … by induction. For N = 100, you may think that one converges to a boring John Doe personality. Wrong again, as personalities do not add linearly; there are the leaders of a field, like the very gracious ones present tonight, and the ruthless ones that were not invited. The leaders have a strong influence on the "personality" of their fields. In any case, fields can be divided, personality-wise, into four simple types: intelligent and nice; intelligent and mean; dumb and nice; and dumb and mean. There are quite a few examples of every kind; thank heaven in particular for the intelligent-and-nice kind. Obviously, it is very much up to everyone, but especially to the leaders, to keep computational biology intelligent and nice! Let's never forget this!

To remain systematic, I dabble with semantics and note that there are two ways to read the title. First, it can be read "Are computational modelers capable of being nice to each other?" Here I remind us that we have to take all our computational biology peers for what they are, as we have no others. We can not fire us all and replace us with real sweeties; we have to make the best of who we are. But is that so bad? Actually, I really love you guys, ladies and gents! There are definitely some interesting computational biology characters sitting here around me and also roaming elsewhere. With them I shared many insights and laughs, we had great debates and, even the nasty referee reports I got (you know who you are!) mostly turned out a savior for my eventual publications. Personally, I think I could do much worse than having YOU as colleagues. But there is a downside. Once I was asked "How can you bear the competition with mighty David Shaw?". My answer was: "You don't know what's the worst of it, David is a nice guy." If we were all mean to each other, we could hate each other and compete with gusto. You take your pick: Enjoy life and comradeship or hone your competitive spirit. I know what I prefer.

The second way to read the title is "Can computational biologists afford to be nice to each other?" as opposed to "should we better kick each other wherever we can in a survival of the fittest in the face of limited opportunity?". Here I note that our field is young and still has enormous development potential; computing is absolutely needed in biology, as realized more every day, in handling and interpreting experimental data, in complementing experiment that often needs modeling for its interpretation as a microscope that offers views simply not available otherwise, and as a data base generating, curating, and analyzing tool. Also, every morning we wake up to discover that technology has advanced overnight and offers us even more powerful tools. Meanness out of necessity in the face of limited opportunity? I don't see it!

Leaving semantics I turn to applying the antipodes of the utilitarian and moral imperatives to our title. Should (in the utilitarian sense) computational biologists be nice to each other? Computational biology as a field, on the one hand, is expensive because of costly computer hardware as well as high labor costs-computational biology requires highly trained and skilled people as well as machines. On the other hand, we are between a rock and a hard place, namely experimentalists claiming to be so much more realistic than we, and theorists claiming to be so much smarter. On top of this, our field is not yet well established in the institutions that we depend on. In this situation, we are faced with the choice of fighting for our own individual benefit or fighting for what is best for our discipline. If we choose the latter, we must stick together, support each other,and speak with a coherent voice. Here the choice is a tough one, though it may not necessarily be a case of one choice or the other; you must each decide for yourselves.

"Ought computational biologists be nice to each other?", the moralist asks. Obviously, being nice is the right thing to do. Being nice does not mean to tolerate shoddiness, but it does mean to be fair and to recognize other's achievements.

Lastly, I have some really good news! Computational biology is actually a uniquely nice field when viewed from a trans-discipline perspective. We computational biologists have been extremely nice to each other, pretty much from the beginning, in fact, so much so that we take it for granted and don't realize that our daily competitiveness, and yes, occasional sniping and pettiness, is nothing compared to the remarkable accomplishments we've achieved through scientific cooperation. Computational biology, unlike many other fields, does not compete about tools, but only about developing new tools and about applying the tools in novel ways. Computational biologists share their tools with each other freely; we even train each other in the use of tools so that all can do better science. Every leader sitting in front of me contributed in this regard in many ways, and so did the junior colleagues sitting here. Once I sat in a multi-discipline NSF panel that eagerly asked me about how computational biologists do this. After I had spoken for a while I was interrupted by an unbelieving voice: "You give your programs to your competitors?"

We computational biologists can be proud of ourselves as a field; our unique generosity in sharing software with each other has advanced our science in countless ways. Let us continue to build on our tradition of sharing, and let us train our students to value and to maintain that legacy. To return to my original question, "Can computational modelers be nice to each other?" - they have been nice all along. May we all stay that way!

To the health of everybody and our field!