Sunday, October 19, 2014

Original Research

Paul Pietroski sent me this nice link on what it is to do original research. It eloquently expresses thoughts similar to those that I posted  (here) on the same topic a while ago. I thought you might find the discussion interesting. To me, it rings true. Let me add a few stray thoughts (but read the linked piece, it’s put better there).

There are at least three problems with doing original research.

First, the skills that got you to the point of doing it (usually in 2nd year grad school) differ from those that will help you to do it well. Earlier skills involve mastering a technology that has been road tested against understood problems, your job being to prove that you can answer the test questions without looking at the back of the book. But, and this is key, there is a back of the book and there are answers there.  This is not the case when working on a new problem.

Second, a good chunk of doing original work consists in finding a good question to ask. In other words, part of the transition to being a researcher is shifting from being a hot shot question answerer to being a fertile question poser (not to be confused with the often superficially similar poseur). Great researchers know how to ask the right questions. Indeed, the questions if good always outlive the answers, which, if the question is really novel, will be replaced by better answers pretty rapidly. What makes a question good? Well in part, it's a little like porn, you know it when you see it. But there are some surface properties of note: Good questions must be worth answering. Good questions must be answerable. Really good questions lead to others that meet the two criteria above. If all of this sounds vague, well it is. And that’s the problem with original research.

Third, and this is the point of the first paper linked to above; you never know enough to answer the good questions. This can make you feel dumb and inadequate and scared and feel like your wasting your time and consider a change in career or inspire you to clean your desk, office, apartment, building, accessible public areas, go to the gym, movies, overeat, diet, scream, go into a depressive shut down, yell at your loved ones, kick your dog… Or, it can, as Martin Schwartz expresses it, it can liberate you. In his words:

That realization, instead of being discouraging, was liberating. If our ignorance is infinite, the only possible course of action is to muddle through as best we can.

And, I would add, if the problem is really hard, then there is nothing embarrassing about not being able to crack it wide open and there are huge psychic rewards for being able to nudge it forward even a little bit. In addition, again as Schwartz notes, there is something exciting about having a problem of your very own. When it comes to your problem you are the world’s expert on that topic (or at least one of the very very few experts). You are going where no one has gone before and making up the rules as you go there.  It can be a real rush, sort of like bungee jumping with an untested cord.

There is at least one problem however. Nobody is really prepared to do original research. Not only don’t you know enough (ever) but there is no guarantee that you ever will. This I believe is the final indignity of real research work. It’s so damn unfair. Hard work need not be rewarded. Ingenuity may be nugatory. Perseverance may go unrewarded. It’s hard and can succeed, but it need not do so and you never really know until the problem is cracked how well you are doing.[1] Schwartz again:

What makes it difficult is that research is immersion in the unknown. We just don't know what we're doing. We can't be sure whether we're asking the right question or doing the right experiment until we get the answer or the result. Admittedly, science is made harder by competition for grants and space in top journals. But apart from all of that, doing significant research is intrinsically hard and changing departmental, institutional or national policies will not succeed in lessening its intrinsic difficulty.

So, new work is tough and it’s tough because it is new, which also makes it exciting and scary. It requires imagination rather than mere competence, it requires an ability to tolerate your own ignorance and it requires a capacity to live with the realization that you may not get anywhere despite your best efforts.  Schwartz says that we don’t really prepare our students for this in their training, and he makes several reasonable suggestions about how to help our students become “productively stupid.” His comments are both sane and humane. But I suspect that in the end they will only be marginally effective. As he also notes, the problem is that there is only so much one can do to “lessen its intrinsic difficulty.”

That said let me suggest one more useful crutch. Though research is tough, it need not be lonely. And even if it ends inconclusively, it can be lots of fun along the way. One of the things that I have found endlessly helpful is talking with fellow researchers. I love lunching with colleagues and students, kibitzing with them, arguing with them, joking with them, playing volleyball with them etc.  The pains of original work can be mitigated in part by the social pleasures of an active research group. When students ask me what they can do to push their work along I always suggest that they talk to their friends about it over lunch, beer, gym. Have fun. Make jokes. Be irreverent. Laugh a lot. Argue. Be silly. Talking about good ideas (and sometimes bad ones) can be very enjoyable. And an infectious delight can often spur the imagination. And as Schwartz rightly emphasizes, one’s imagination needs all the help it can get given that most of the time if we are doing our jobs right we will be really really clueless.

[1] I suspect that this, even more than material advancement, is what tempts people to cut corners. It’s not the desire to deceive, so much as the desire not to fail. Here is a nice discussion I found on various ways that this is done. The discussion focuses on model building and the ways stats can fudge matters, but the maneuvers and temptations described circulate very widely. I admit that they have winked at me more than once and I am not confident that I’ve always avoided yielding.


  1. Boy, this is embarrassing. I just re-read the earlier post I linked to and realize that it discussed the very same piece that Paul sent to me. There it is, early onset forgetfulness. At any rate, I am happy to report that I was more or less consistent in my reactions to the piece over about a period of 12 months. At any rate, the Schwartz piece is very good and contains useful comments about life as a professional researcher. Take a look. Sorry for the duplication.

  2. That's another good feature of blogs -- they're happy to publish replications.