This interesting exchange took place in the comments to a posting in John Cook's Google+ account, date 20111106 (not sure how to link to it for the long term):
John Cook: "The softer the science, the more political it is. That matches my experience."
Lao Tzu: "In maths departments, people feel they can use each other's research so it's more collegial. In philosophy departments, moving forward means cutting others down."
John Cook: "Grants also make people less collegial. Someone at my institution said that when the faculty were first required to cover part of their salary in grants, people immediately began to view colleagues as competitors and quit sharing data."
Idea for a metric of "softness" in scholarship: proportion of multi-author to single-author articles (books are more likely to be joint projects).
Though my interests are rather technical, all my articles (not books) were single-author until I left academia and stopped having to think about making an impression on deans and promotion committees — neither of which I actually cared about, but which had the power to affect my circumstances.
I still want credit for my ideas and labor, but now I'm glad to share credit for help I've received — in a byline, I mean, rather than simply thanking people in a footnote. I feel this to be a decisively healthier state of affairs than the way I used to work. Sharing a byline motivates me to ask for sharing of ideas rather than just "help", and motivates others to give more fully of themselves.
But the humanities does seem to discourage collaboration. I continue to be contacted by other scholars who want me to advise them on something that they will publish alone, even though we could do a much better job if we worked together. Graduate students (I won't name the institutions) seem reluctant to do projects I've assigned unless they can turn them into a conference paper of some sort, which in the humanities means a single-author byline. The first really harsh review I've ever written came out recently — among other things, I criticized the author because he should have collaborated fully with a statistician and a historical phonologist; instead, he merely asked for their technical help and then drew his own conclusions and published the whole mess under his own name.