Personally, I think the main driver of hostility was the skyrocketing popularity of Wikipedia between 2003 and 2005. I think that prior to 2003, essentially the Wikipedia editing community was puttering away doing a good job writing good articles, mostly pretty much ignored by the outside world. Then in 2003-ish, readership skyrocketed. That resulted in:
- A huge influx of vandals;
- A huge influx of POV-pushers and self-promoters;
- A huge influx of new editors who wanted to help but didn't know how;
- A ton of fairly stupid and ill-informed media coverage, much of it making fun of Wikipedia, or hostile to it;
- A huge wave of interest in commercializing/monetizing Wikipedia (business proposals, etc., many of them clueless)
I think it's hard to overstate the importance of that enormous surge in popularity, and how influential it was in shaping Wikimedia community attitudes and behaviours. I wasn't editing Wikipedia at the time, but I've talked to, and observed, lots of people who were, and I think I can imagine what it was like for them. I think they felt beleagured, like an enormous tidal wave of input was crashing over them, much of it threatening Wikipedia's integrity and quality, and there weren't enough of them to successfully manage it. And I think they felt a heightened sense of personal accountability: Wikipedia was starting to be more influential, to matter more. Which I think made editors even more worried about maintaining its quality.
I believe that sudden, unprecedented surge in popularity created an atmosphere of insularity and suspicion of outsiders that continues to this day. And we are still beleagured: the problem hasn't gone away -- Bodnotbod and Randomran attribute their own brusqueness/curtness/abruptness to being too busy, which I think is very true for many editors. It's a chicken and egg problem: if we are too busy to be welcoming, we will never get the help we need to become less busy. So, we need to upwards-prioritize bringing in new people -- we need to make an investment in friendliness :-)
I'll say too that there are obviously other factors at play here, in addition to the popularity surge. Super-fast: the Wikimedia projects are edited mainly by geeks, and geeks aren't known for being particularly warm and welcoming. Online culture in general tends to be sparky, anonymity drives sparkiness, and text-based communications tend generally to lack warmth. Communicating across languages sometimes, I think, leads people to overstate their points for clarity. We have no really good mechanisms for sanctioning / kicking out jerks. And the people who edit Wikimedia projects are super-smart; I believe a certain amount of hostility isn't intended as such, but is experienced by the participants as pleasurable intellectual debate. Plus, editors are mostly male, and men are not as socialized as women to be sensitive to social cues, and to care about social harmony.
So, there are lots of contributing factors. Most of which I think can be alleviated by broadening the contributor base.