Most of my collaborations, long-term and short-term, have been very positive experiences. I haven't had any trouble finding interesting and collegial people with complementary research interests. It can be frustrating working with other people who have a different style and pace of working (and writing/editing/communicating), but in general things work out for the best..
.. except when they don't, and then the situation and attempts to extricate yourself from it can be painful.
Last spring I wrote about Working with Jerks, as I believe it is nearly impossible for one to go through an entire career without ever working with someone of the jerkish persuasion. There are many different species of jerks, and some are more odious than others.
If you find yourself working with a difficult and/or unpleasant person and you want to get out of the situation, is there a good way to do it? The short answer is no, if 'good' means 'easy', 'uncomplicated', and/or 'without negative consequences'. These situations are seldom as simple as making a call (or sending an e-mail .. or text message .. or posting something on Facebook) to say "You are a jerk and I am not working with you any more." (even if you leave out the first part)
In some cases, there are ancillary, complicating issues, including personal connections and funding commitments. If there are such issues, you have to think carefully about whether and how to end a working relationship.
What if the jerkish colleague has some mitigating circumstances? For example, what if this person has had some difficulties in his career/personal life, has a 'mood disorder', has some difficulty relating to human beings but is not an evil person, and so on? Do all those what ifs add up to making excuses for someone who should be ostracized or are they sufficient reason to end a dysfunctional and/or unpleasant working relationship?
I know some people who would respond So what? to each of the what ifs above, and others who would take these into account and try to continue working with a difficult colleague. So, let's consider some more what ifs.
What if one of your jerkish collaborators runs a lab that you need access to and there are no other such labs on the planet? Unfortunately, I cannot recommend trying to end such a collaboration.
What if one of your jerkish current or former collaborators becomes an administrator at a major funding agency? If you've collaborated with such an administrator, they have a conflict of interest with you and cannot be involved in the review or decision process of your proposals, but it can make for a tense situation if you don't have a good working relationship with a grants officer. Unfortunately, I cannot recommend trying to end a collaboration with such a person either, though I've tried it.
What if you don't have a tenured or a tenure-track job (yet) and you can't go telling influential jerks to take a hike (yet)? Ditto.
What if the problem person is a dysfunctional student or postdoc and you have a lot invested in them (time, funding) and/or it would be a major problem for you and your research group if you fired them?
Do all these what ifs sum to the conclusion that, more often than not, we have to continue to work with unpleasant people even if we don't want to? Does this help perpetuate the existence and survival of the jerk species in academia?
I hope not, but only once in my career thus far have I successfully done a surgical removal of an extraordinarily jerkish collaborator, and it took many years of putting up with unpleasant behavior by that person before I reached the point of no return. The point of no return for me happened to occur in a charming alley of a major international city and involved my saying to my soon-to-be-ex-colleague, in a rather emphatic way that startled passersby: "You are a psychopath. I am done working with you." [FSP walks away]
It sounds kind of funny now, but at the time I was quite shaken by the experience because I had a lot invested in that collaboration (financially and emotionally), and I was worried about the collateral damage -- i.e. would there be negative consequences for my research group of severing this collaborative relationship? It was clearly the best thing to do, though, and the few negative consequences were worth the aggravation.
A more typical way to end an unwanted collaboration, however, is a more gradual, more diplomatic retreat.
As I get older, I have less patience with unpleasant people and in theory have the luxury of working with whomever I want, but it is never that simple. There is almost always some reason why I could put up with some amount of unpleasantness for the sake of the larger project, or the students, or other colleagues, or someone or something. I could, but do I want to? What level of difficulty is acceptable and what level justifies ending a collaboration despite the consequences?
There's no one answer, of course, in part because it's a moving target, but in general, if the bad outweighs the good, it's time to end the collaboration. Perhaps you can be open with the ex- or soon-to-be-ex-colleague about why you are no longer interested in collaborating with them, or perhaps you can "reevaluate your priorities". And perhaps you can gradually extricate yourself or find some other polite exit strategy that doesn't involve yelling about psychopaths in the middle of a street.