# Is it sometimes better to not publish at all?

I hope this is not a politically incorrect question to ask, but for a PhD student who usually publishes at CCC/ITCS/ICALP (and occasionally at FOCS/STOC), could it be harmful (career-wise) to publish less significant works in less prestigious conferences (e.g. MFCS, FCT, STACS, IPL)? Could it be better to just leave such papers laying at ECCC/arXiv?

-
it depends on the quality of results. focus on getting the best results 1st & getting them into the best journals that will take them. =) ... the electronic sites are useful for not-as-fully-cooked stuff & establishing priority etc. –  vzn Feb 23 '13 at 19:35
What kind of career are we talking about? –  Thanatos Feb 23 '13 at 23:38
I assume she means an academic career at a research university. –  JɛﬀE Feb 24 '13 at 1:29
surprised at high upvotes on this. so far answers below are assuming that "career-wise" is being measured by selection/promotion/acceptance committees & leading to much "inside baseball" talk, but "career-wise" is actually a very broad term verging on vague. for example, a researcher might be happy in their current position, not wanting to move, and it could depend on that particular schools committees attitudes. other note: there is some connection of the question to citation analysis, a growing area of study/application albeit controversial. –  vzn May 30 at 16:28

I was just referred to this question by graduate students that, in my opinion, were far too influenced by the answers. So let me start with two generic advises.

1. To the aspiring scientist: Don't assign too much weight to any answer on such matters, and don't assume that a small and highly non-random sample represents the common views among senior (or non-senior) people in the community. In general, think for yourself! See http://www.wisdom.weizmann.ac.il/~oded/advice.html for more details...

2. To the senior scientist: Be careful about what you say, since it may be misinterpreted in harmful ways and/or have more impact than what you intend and/or perceive.

Re the discussion itself, I think that the idea that credits are non-monotone is not only utterly non-intuitive but also utterly wrong, and I am talking as a person who sat on numerous committees that took various career decisions. A person who has $X$ fundamental contributions, $Y$ important contributions, and $Z+1$ nice/legitimate contribution is ranked higher than one who has $(X,Y,Z)$, regardless of the numerical values of $X,Y,Z$ and assuming that quality captured by the $(X,Y,Z)$ triples is exactly the same. Trade-offs between different types is a different question, ditto re how much credit does each increase give...

In other words, for any set of works $S$ and any additional work $a$, the credit of $S \cup \{a\}$ is (strictly) bigger than to $S$ [i.e., strict monotonicity].

In my opinion, people who claim the opposite just assume that a larger number $Z$ implies a decrease in what the value of $X$ (or $Y$) could have been. But this assumption may be wrong and more importantly is irrelevant to the comparison at hand. That is, if you compare a case of $(X,Y,Z)$ to one of $(X,Y,Z+1)$, you must rule that the second person (called B) was able to meet the performance of the first (called A) although B also did another work of 3rd type; so B is clearly better. Indeed, you may think that B could have done better investing more energy in Type 1 (which is not always true - see below), but that's a comparison against an imaginary B, not against A. (And when you have a case of $(X,Y,Z)$ against $(X,Y,Z+10)$, the same holds is stronger terms.)

In addition, I think there is also a confusion between the works and the publications. If a work already exists in writing, and assuming that it has its merits, then it can only be advantageous to publish it in a adequate venue, where by adequate I mean one that is intended for works of this profile (wrt quality and scope - publication in a too prestigious conference may actually hurt, since it may generate some annoyance and even bad opinions re the author). But if one still has to develop a work from an initial idea (or "only" write it - which always involves some more research...), then one may consider the trade-off between the amount of time required versus the importance of the work.

Finally, as I hinted above, it is not clear that one is better off aiming all the time at Type 1 (i.e., fundamental work). Firstly, this is infeasible and thus problematic/harmful. Secondly, and more importantly, one is always better off following the inherent logic of his/her own interests and ideas/feelings, and aiming to do as well as possible. See more in the aforementioned webpage.

Oded Goldreich

-
Welcome to cstheory ! Hope you stick around :) –  Suresh Venkat May 30 at 9:02
Sorry, but I don't intend to stick around since I find doing so very time-consuming and somewhat annoying at times (this feeling is based on some but not much experience with other blogs). But do feel free to call my attention (via email) to anything that you think I would be interested to react to, or to anything that you want my answer to. (This invitation extends to all readers.) –  Oded Goldreich May 30 at 9:45
+1 for "one is always better off following the inherent logic of his/her own interests and ideas/feelings, and aiming to do as well as possible"!! –  JɛﬀE May 30 at 15:33
ditto on the welcome! the attempt to measure something inherently subjective (candidate accomplishments/contributions) in formal/mathematical language is rather questionable & is reminiscent of statistical/quantitative-based/citation analysis debates going on in scientific circles (& there are various refs/surveys on this out now & TCS plays a role here). re "senior scientists", opinions on how to make the site better and "less annoying" have been solicited (albeit controversially) on meta. –  vzn May 30 at 15:37
Re the attempt to use math -- it was merely illustrative. You may prefer the alternative that talks of $S\cup\{a\}$. This is not math, only using math symbols, and it merely captures the claim of monotonicity while being careful to stress that one should compare the same set of works and not merely numbers. I am all in agreement against the objection to mere statistics. –  Oded Goldreich May 30 at 15:50