334. How much should you publish?
Uncategorized — By Dmitry Podolsky on April 3, 2009 at 10:56 pm
Save This Post as PDF
As far as I remember, it was believed that one has to publish about 2-3 papers/year during glorious and relaxing PhD years (i.e., about 1 paper per 4-6 months, not a too heavy burden indeed)… This would give necessary 6 publications in the end of the term (in Russia, getting your PhD takes 3 years) for your thesis to be officially considered worth defending. If you were publishing less than 2 papers/year, you were assigned a mark “loser”, which is, I guess, quite reasonable 
Being loser meant that you had to spend another year preparing your thesis, then – another one, until the overall number of your publications would hit the magical number 6. Believe it or not, I happen to know (many) very smart people belonging to this category.
Now, the first question is – how many published papers are you supposed to have in the end of your first postdoctoral position term in order for your career to be considered successful? Does the vacuum expectation value of the number of published papers differ for different institutions and countries? What are your thoughts, what is your personal standard of scientist’s productivity?
20 Comments
I’ve seen people go different routes. Some postdocs focus on one big usually solitary idea and then publish a hard hitting heavy citation count paper, and the remaining 2-3 papers are either letters or support/generalizations of the big paper.
Other postdocs prefer fewer solitary papers but with a steady stream of papers that are done in collabaration with a big name or two to get their street cred up.
The latter tend to do better on average i’ve found when it comes to hiring time, b/c they develop friendships with the collaboraters. The former route also has luck factors involved(it really sucks to find that the thing you’ve been working on by yourself for 6 months night and day ends up not working as planned)
Ok, no diversification strategy – too risky from my point of view, since your whole life as a physicist is at stake during first 3 years.
How many places are you typically applying to?
Cheers,
Dmitry.
“How many places are you typically applying to?”
I can’t speak for everyone, but after my postdoc I already knew where I was going to be hired more or less. I still applied to about 7 other positions just in case a random famous person decided to appear and take my spot, but I had an in (again b/c of collaborations). Also, i’m a phenomenologist, so things are an order of magnitude less tight than what it must be for a theorist.
So much of those first 2 years is unfair to the extreme. If you are lucky to be at a prestigous university, the road is so much simpler. I’ve heard stories from theorist friends who worked at Princeton (which is essentially a tenure track factory), that they more or less hold your hand right through the whole process. Eg ‘Get idea from Witten’, ‘Get theory group to explain how to solve the problem’, ‘Cosign with famous physicist’, ‘Get recommendation letter from Gross’.
Uhoh, I see. The total number of applications I send during the last round was around 70
I get your point, but “7″ still seems to be a bit of no-diversification strategy: what if someone with a better profile files out an application to your collaborator, do you really have a backup when you apply for <10 positions?
That’s probably the statement hard arguing with but I’ll try a bit: I actually happen to know a couple of counter examples. One counter-example is the guy who had to return from Princeton to Russia due the lack of support (although he was considered one of the best students there). Another example is a friend of mine, currently a PhD student at Princeton… I wish I would be ever able to work so hard as he does. In a few words – I have an impression that nowadays our little world is so competitive that simply being a Princeton alumni does not completely guarantee your success… There are also MIT, Harvard and Stanford physics departments anyway, and the number of professor positions is essentially lower than the overall number of PhD students those departments release per year
Let us now have a bit of weekend fun solving the following problem
Conditions are: suppose you know that your competitor is a PhD student from Princeton having 4 publications with overall number of citations around 50. He has a short weakly personalized recommendation (< 10 sentences) by Witten or Gross (“very talented, must hire”). You apply to a theoretical physics (not just HEP) department where no professor has ever worked with aforementioned people. Neither you, nor your competitor collaborated with people from the department. How many papers/citations should you have in order to really compete with the guy? 
Cheers,
Dmitry.
70??? Wow, I thought I was exagerating when I said it was an order of magnitude harder to be a theorist =) I didn’t even know there was 70 theory positions even available at any time during hiring season.
But yea I knew roughly where I stood amongst the people I was competing with (by citation count) and I roughly knew where the people ahead of me were going/applying, so I wasn’t that worried (most people I know apply about 15-20 places, I was on the very low side). Theres also a weird collusion going on, people are often locked into one of a few places from the getgo. A close friend/collaborater knew he was going to one of three places for instance.
As for your game. Its going to be heavily biased based on your publication record and if you happen (by luck) to have a paper that caught the eye of a senior member on the hiring commitee. Recommendation letters go a certain length, but its way too common to have gushing letters full of praise so I think their utility is overemphasized. Institution prestige unfortunately often ends up being a tiebreaker, so perhaps lets say you have to do 20% better than someone from Harvard or Princeton to end up with an advantage.
A lot of this game is word of mouth ultimately. Faculty chitchat about their postdocs all the time, and it helps if you know who your advisors have talked with or are friends with.
I have actually built up the list for several months and yes, in the end, it turns out that market is not that bad after all, especially in the US… And of course sometimes you submit to places that did not advertise this year
especially if you are a string theorist 
– some condensed matter theorists and cosmologists, friends of mine, complained that they have to deal with many somewhat irrelevant applications.
Also, I did not send my application to places that didn’t advertise this year which however seems to be a common practice nowadays.
I’ve actually heard another strongly different opinion – rec. letters are of immense importance right now, and you better send 6 letters to the place which asks for 3
I think this is a kind of reasonable, after all how else could young relatively low profile postgraduates get a nice fellowship competing with people having longer research records? And of course word of mouth is important, as you say…
That’s what I wanted to hear – an estimation
But I think that 20% is too low, mine is more like 50%, since the competition is crazy nowadays.
Cheers,
Dmitry.
I should publish as much as I like. And I encourage everybody else to do the same.
That is – it is not correct to force anyone, students, scientist, teachers, amateurs…. to publish something.
That’s fine, but how do you suggest your boss to decide about the size of your salary?
well, scientists should not be limited by money. We should be limited by other people. So I suggest that each boss decides about the salary by knowing his employees.
Particularly, my boss knows me. He knows my work too. And this knowledge is not based exclusively on the number of articles that I have published. This knowledge is part of my boss job. That is his responsibility to know his employees and decide about their salary.
I am talking about basic human quality. If we only count papers, how do we know the top scientists are not bad or even evil people?
Counting articles is usefull only for picking people whom we do not know. But in this situation, maybe a recommendation letter could be used instead. What do you think?
I think that recommendation letters, word of mouth etc. should work relatively well within not too competitive, small niches, say, for a single country job market like Czechia or Russia. If you go international or if the job market is highly competitive, I am afraid, some metrics useful for evaluating someone’s performance should be introduced. The total number of published papers is not too good of a criterion, but it is a criterion with higher level of objectiveness than, say, the word of mouth.
Then again – if you base your evaluations only on recommendations letters and the word of mouth, how is the overall system stable w.r.t. corruption?
Cheers,
Dmitry.
P.S. Thanks for the comment by the way!
My understanding is that whatever metrics you introduce will spoil the measurement sooner or later, basicaly because it will be easier to perform mechanicaly according to this metrics. So I prefer not to introduce any such thing. Perhaps science should not be measured. Same way as art is not measured. If people like the art, fine. If not, that is fine too. This also imply that science itself is not objective. I would say that scientific results can be made objective very much. In fact, extremely objective. But not absolutely objective.
It is utopic to think that corruption will disapear. The most one can do is to promote less corruption. While counting citations certainly promotes less corruption, it also puts preasure on people not to publish high quality papers (since quality is less important than quantity in this case). I believe that corruption targets only relatively few people. But preasure to publish anything is put on everyone. That is what I consider very bad. Especially bad for science, but also bad for people and society.
Word of mouth and recommendation letters perhaps allow for corruption but also promote good human relationships and cooperation. One could imagine that it also motivates people to present high quality work to their bosses or other people whom they want to write a recommendation letter for them.
Surely, what I am describing here is how I think the world should be. Not how it is today. In this sence I understand the topic of this blog about article counts etc…
Finaly – you are too kind to me. It is me who should say thank you instead. Thank you for this whole blog, which I like to read.
P.S. one technical note: I cannot see any possibility to preview my comment, co I add is as it is, hoping that I have used the citation markup in a correct way.
hmm, no.
Hi Michal,
I fixed your comment above
That’s true for any system involving a social factor – for example, for markets. In quantitative analysis of markets, you invent a trick, you make some money using the trick before it becomes known to everybody else on the market and looses its value. The truth is that you don’t stop trying to invent new tricks because of that – you just try to invent more sophisticated ones.
Ohoho, here I will most seriously disagree with you
Moreover, I think that this kind of attitude – if it spreads – will bring a mortal danger to science. There is a serious difference between art and science: science is objective while art is subjective. Arguments that you use in your papers are objective, not subjective, because they are either correct or wrong. You accept them because they are correct, not because you like them or your advisor recommends you to like them.
Cheers,
Dmitry.
I guess we are addressing different questions. When I say that science is subjective, I am talking about “who profits from this discovery?” or “who decides about which way will this particular research go next?”. And that is definitely subjective. Obviously not everybody will have the same profit and not everybody will choose the same way to continue some research. And this is how science is measured these days. This is also why I am saying that it is better not to measure science, because without any metrics, everybody can do whatever he finds interesting, without having too much pressure from other people.
When you say science is objective, you perhaps address the question “is this scientific result correct or not”. This I call objectivity of scientific result and not objectivity of science.
Objectivity of scientific result is, as I wrote earlier, extreme. We struggle hard to get it as objective as we can. For most applications this is OK. But one should know very well the accuracy of each law / experiment and be rather careful. Since it is always possible to improve those laws / experiments.
One should not forget, that especially experimental results are most of the time based on the personal quality of people doing that particular experiment. When counting papers, more “fantastic” or “shocking” or “boulvar” results will get more profit. So there will be a tendency to get such results. In fact, there is already such tendency.
I guess one should promote good relationship between theoretical and experimental physicists. I imagine this would lead to higher quality of results.
hmm, after some more thoughts I came to the conclusion that the original question “how much should you publish” is completely wrong.
So I propose a better question instead: “What should I publish?”.
In the context of your original entry: What kind of papers should a postdoc scientist publish? How high should the standard be?
That’s a very good question, let’s talk about it on Saturday
Dmitry,
Very nice reading. Keep up!
Cheers,
Marco
Hi Marco
Thanks for dropping by! Enjoyed your recent conversation with Lubos and Terence Tao.
Cheers,
Dmitry.
Trackback responses to this post