When planning a publication strategy for a dissertation, invariably the question comes up where to submit your papers. Ph.D. students naturally are biased towards conferences, because if a paper gets accepted to a conference they get to travel to a (usually) nice place. I nip this bias in the bud right away: For a journal paper, every Ph.D. student gets a conference to attend for free. This lets us focus then on the economic value of a journal vs. a conference paper and how to best reap the benefits of hard research work. Here, I’m a contrarian (to most colleagues): I’m in favor of journals. It is also the economically smart choice for a Ph.D. student.
It is really simple: In computer science, journals are underrated, and conferences are overrated. An established computer science professor typically has a long history of conference publications and hence will want to keep the value of these high, recommending a Ph.D. student submit to conferences. To most anyone from outside computer science, however, only journals count; this is due to historic reasons and also because the various indices used in evaluations until recently only provided data on journals. Within the scope of their university, computer science professors were always with their back against the wall having to justify their conference publications.
Evaluations (“bean counting”) will only increase. Therefore, the value of journals will be forced upon us: Resistance is futile. As long as researchers submit to conferences (for historic reasons), and those are viewed as second class citizens by science at large, as a matter of degree, within computer science, conferences will remain overvalued and journals will remain undervalued. A Ph.D. student, well, anyone, should try to get the biggest bang for their bug and that would mean submitting to journals. End of story, so far.
I understand that many more arguments can be made: Conferences afford things that journals can’t, publishers are including conferences in their evaluation indices, the world is chunky and not a matter of degree, etc. This doesn’t change a thing that the life-time reputation value a researcher acquires for their C.V. through publications is presently higher with a journal paper than a conference paper. This benefit is temporary: Once computer science professors give in and accept that journals have long won, the inefficiency will go away.