Science is about a lot of things. But amongst the most fun and eye-opening things are certainly collaborations. Now, collaborations can - depending on your field - come in all sorts of flavors, but they all can broadly be divided into either domestic or international ones. You might argue that the latter are better, because they give you an excuse to travel more and further. Conversely, you could argue they are useless as they slow you down due to language issues, different time zones or whatever other problems may arise.
So which one is the way to go? Thankfully, somebody has really gone to town for you to find out. Jonathan Adams (formerly director of research evaluation at Thomson Reuters) has combed through a whopping 25 million papers (!) published between 1981 and 2012 to look at the outcome of collaborations, in particular comparing the outcome of domestic and international team efforts. While the measure of success in this particular case was the average citation impact of the work (a measure under increased scrutiny these days), the outcome provides a clear message:
Go out there, find international collaborators that are willing (and able) to contribute to your work!
Yes, you may end up with an entire cricket team worth of names on your paper, but chances are the outcome (and impact) will be much further reaching than just teaming up with the folks next door.
But: before you ignore the expert from down the hall, keep in mind: setting up a collaboration (domestic or not) is better than not getting one up and running at all. You can only learn from other experts (who in worst case scenario only teach you new drinking games).
For some concrete numbers on the outcome of domestic and international team efforts, check out this article in Nature Magazine (subscription / library access required).