From Kate Clancy: "I Can Out-Interdiscipline You: Anthropology and the Biocultural Approach".
Being interdisciplinary isn’t the same as being a little good at everything, consistent with the saying “jack of all trades, master of none.” Heneghan analyzed the footnotes of one of the most popular interdisciplinary works, The Ecological Thought by Timothy Morton, and found it represents typical biodiversity quite well: the strongest influences by a few fields with, as he says, “a trailing edge of rarer species.” At least one model of being interdisciplinary, then, is to be very good in one field, pretty good in a few more, and then conversant across others. Some of the work I’ve been reading never masters that first field. And so there is something less than ideal in how we are training our students.
This notion of a "long tail" approach to interdisciplinary research is interesting. As one considers more and more fields, the amount of knowledge an investigator knows about those fields declines exponentially. This is of course how we train students in the first place. They get the basics of knowledge in many fields, and specialize in few where they get a deeper level of knowledge.
The long tail of internet discourse is an expression of Zipf's law. The traffic (or readership) of sites declines exponentially with their rank. A disproportionately large amount of traffic goes to the top few sites.
The "long tail" of interdisciplinary knowledge seems to work like this: As we rank fields by an individual's knowledge of them, the amount he knows about the (n+ 1)th field approaches the amount he knows about the nth field, as n becomes large, with an asymptote of zero knowledge.
The state of knowledge within a field changes fairly quickly. As a result, someone who thinks he is very broadly educated may have to work very hard just to keep up with the state of knowledge presented to undergraduate students in introductory courses in a range of fields. I cannot tell you how many times I have dealt with "interdisciplinary" scholars who are pushing research based on a 1975-era knowledge of the human fossil record. Most scientists over 50 know the 1975-era version of human evolution very well -- that is, the "long tail" of knowledge by nonspecialists about paleoanthropology basically contains the same information. When these people work together, the whole is not the sum of its parts, it is the sum of one part plus epsilon.
Going deep into an interdisciplinary problem may mean working with other people who have thought long about the other fields you need. It also means working long enough with those people to be able to perceive the times when you're not speaking the same language.