I am a college professor. We are supposed to do research from time to time and get it published. Collaboration is a pretty significant part of the way all this research gets done. But it is also a deeply vexing issue for many researchers who have attempted collaborations. Egos get bruised. People feel that they have been had, cheated, duped, etc. What follows is my take on my most recent adventure in collaboration.

This other prof, let's call him Pi Beta, and I had met at academic conferences and I had earlier this year given a seminar, at his invitation, at his university.

A week or so ago he emailed me the manuscript of a paper he had written and requested my comments. This sort of thing happens all the time in academia; we are usually desperate for others to read our stuff and tell us what they think about it so that we can fine tune the work before submitting it to a refereed journal.

I read Pi's paper pretty much as soon as I got it. This, by the way, says something about me, I am sure. Typically, one has to keep bugging the other guy to get moving. Anyway, the paper didn't look to me like anything I would have been proud to have written. The main idea did not knock me off my chair, to put it mildly. But then, probabilistically speaking, I wasn't expecting to be upended; most research in my field -- economics, but I suppose this is true in other fields too -- is ordinary and is meant only to impress uninformed readers of curricula vitae. Pi's paper looked unfinished, like somebody's jottings rather than something the author himself would seriously consider a finished product. It seemed that the author did not even know how to set up what economists call a model. There were no results worth mentioning, no fresh insights about how the world works. It was all spectacularly haphazard and incomprehensible. I emailed some polite comments to Pi and thought that was that.

But that wasn't that. Pi surprised me by writing back and offering to collaborate with me to finish the paper. I was pleasantly surprised and wrote back to accept. You have to understand that this guy teaches PhD students at a New York City university, whereas my college in suburban New York offers only a bachelors degree. I was flattered. I quickly got to work on the project, putting aside my own research. I was excited about this unexpected prospect of another publication in my unprepossessing c.v. I consider myself a spectacularly luckless person and this sudden development had come like an auspicious bolt from the blue.

Today my lucklessness reasserted itself. The collaboration went down in flames.

Before I can explain what happened, I will need to get into a few boring details. A paper had been published in the May 2002 issue of the American Economic Review by three profs named Gollin, Parente and Rogerson (GPR). Among other things, it referred to the well-known fact that as countries get richer, agriculture becomes less and less important as a source of jobs and income. GPR also gave a hypothesis about what it was about typical economies that made agriculture fade away the way it does. Prof. Pi Beta had read the GPR paper and thought that he could come up with another hypothesis that could also explain the same phenomenon. This had led to the paper, or sketch thereof, that I had received. I had pointed out to Pi that while GPR's argument had assumed autarky -- that is, that people do not trade with each other -- Pi's admitedly unfinished argument assumed that people could and did trade with each other. Pi had not noticed this and was pleasantly surprised at the greater generality of his argument, assuming of course that his sketch would in course of time become a fully developed proof. This bit of good news from me may have had something to do with his invitation to me for a collaboration.

Anyway, I need to point out that it is notoriously difficult to think through the effects of trade on the way a country gets richer over time. This field is called growth theory and growth theory generally makes things easy for itself by imagining that the world has just one person called the 'representative agent'. If you begin your argument with 'Let's assume there is only one person in the world' who's she going to trade with? If you want trade, you need to say 'Let's assume there are at least two people in the world.' Current growth theory doesn't do that because, as I said at the beginning of this paragraph, it is far too complicated to work out what would happen if you let trade be a part of your argument. Not surprisingly, I soon found out that it was impossible to say anything crisp and comprehensive by working out the consequences of Pi's thinking with pen, paper and an average ability to do math. I had set up 'the model' and worked out the mathematical properties it ought to have; in the jargon, these are called the first order conditions and market clearing conditions. But that was as far as I could get. For any further progress we would have had to feed those mathematical conditions I had worked out to a computer and do a 'simulation'. The computer would then tell us whether agriculture really would fade away in an artificial world that runs along the lines imagined by Prof Pi Beta.

Now, I do not know how to use the computer to do simulations. So Pi would have had to do them.

Early this morning, I got a call from Pi. We discussed how to proceed. He suggested that I read up on a paper by someone named Campbell so that I could help Pi with the simulations as well. I said I would. At one point in the conversation Pi revealed that he was going to write a paper on his own based on his initial idea but without assuming that people traded with each other. He thanked me for the work I had put in and asked whether I would mind if he used in his solo paper some of the work I had done for the paper we were going to write together. I said fine and that was that.

Later I realised what had happened. By writing a solo paper that developed his idea, Pi was making sure that he would get complete credit for his initial idea; had the idea appeared only in the joint paper, he would have had to share credit with me. Moreover, by embarking on the collaboration, he had gotten me to do work that he could use in his own paper; without it he might not have been able to get his idea published, going by the initial version I had received. As far as I was concerned, on the other hand, Pi's move destroyed any reason I might have had to continue the collaboration.

Had Pi's initial idea been expressed only in a joint paper with me, I would have gotten some of the credit for Pi's initial idea just as Pi would have shared the credit for the work I had done. Now, with this solo paper by Pi set to appear before the much more complex joint paper with me, I would not get any credit for Pi's initial idea whereas Pi would have a share of the credit for my work for the joint paper. For Pi this was fabulous, for me it was unbearable. Had I known beforehand that Pi would write his own paper, I would not have touched this collaboration with a ten foot pole. Pi had started out with a paper that allowed trade. After I had worked several days on the project, he decided to embark on the elementary project of a solo paper without trade. This meant that the project I had been working on would be seen as derivative, merely an extension of Pi's solo paper. So, I withdrew, wishing Pi the best. He thanked me again for the work of mine that he would use in his own paper.

And that, as they say, was that.

Let me end this cautionary tale with another reason why I would not have entered into a collaboration had I known that Pi would write his own paper. It has to do with something called conflict of interest. With Pi working simultaneously on his solo paper and his joint paper with me, where would you say his attention would be, on his own paper or on the joint paper? With priority of publication being crucial to who gets how much credit, would Pi be more likely to press hard for early completion of the joint paper or the early completion of his own paper?

You may take these questions as rhetorical.

Moral of the story: If you are considering getting into a collaborative research project, ask your collaborators whether they would promise not to publish their own papers on the idea before the completion of the joint project. If they refuse, say 'No thanks' and walk away. And, please, no spitting and cursing while you walk away. It's always important to be polite!