They do matter, I think our experience does match. This is part of the problem I was alluding to. All I am saying is, that I do no longer take for granted the

They do matter, I think our experience does match. This is part of the problem I was alluding to. All I am saying is, that I do no longer take for granted the paper/IF (& other Thomson ISI Corp.) science metrics as the golden standard, because it is being actively abused (clique self-referencing, grant business etc.). This is not a new controversy, and I think it is much like the RIAA/MP3 shift, which is not only taking place in publishing (web daily news, kindle etc), but also in scientific publication (NIH publication policy etc). arxiv.org seems to be one of the best technical compromises, but unfortunately, it doesn't fit very well in the establishment's established authoritative scheme (IEEE, ACM etc), which is a racket by anyone's standards. And it can be misleading too (there was an ARS report on this, about the level of trust in the result being published). Long story short: I, for one (and I think you too) are very unlikely to get tenure based on papers on arxiv.org about volcanoes, transportation, urbanization and JIT. No matter how good we are in Linux sysadmin an distributed systems. Though back on topic, publication per se is what early stage training is about (and not PhD), in my experience. Of course, one can always ''downgrade'' an EST to a PhD, but then one has to go through all the post-docs, to get to be staff again. Short of luck, of course. For those with tenure, it is very rare to actually write papers to publish (though it does exist) -- that is what students, assistants and fellows are kept around for.