Well, what does the scientific community need in terms of publications?

* Somebody who organises some sort of peer review. This does not necesarily have to happen in the currently common form, but some mechanism that avoids total nonsense and ensures quality is needed. This cannot be done (in my opinion) on basis of a voting and commenting system, as people would either not do it (and thus no peer review would happen) or abuse the system (which is not to say that no abuse of peer review is happening now, but this would be worse).

* Somebody who publishes papers, be it online or in printed form.

* Somebody who typesets papers or otherwise renders them in a digestable form. While many scientists may be able to do this themselves if provided with a proper LaTeX template or similar, they are still the minority, and at least I do not want to have to read papers set in Word or similar.

Now at least in my opinion, the above points are also the defining properties of a *journal,*¹ and some journals, e.g., PLoS One do nothing more. So unless you want to propagate a different definition of *journal* (in which case this question becomes one of definition), we cannot do without journals.

Note that I do not claim that there is nothing wrong with the current publication system – on the contrary: there are a lot of things that need radical improvement. But the mere existence of journals in general is not the problem and cannot really be avoided.

----
<sup>¹ and yes, this means that even conference proceedings, which are the dominant form of publication in computer science, are some variant of a journal</sup>