During grad school I spent a year or so being supported by a very broadly defined interdisciplinary grant. At fist I hated it. It meant attending a weekly meeting with everyone at university associated with the program. A group of people I never would have voluntarily chosen to associate with professionally. They were nice people but. This is science?! This is research? What are these people doing?! ..these were my common complains. It took me a while to get to know these disparate fields. Once I did I saw how much their constraints (and questions) were totally different than mine and it started to make more sense. I just had no idea initially. What they were doing had seemed ridiculous to me because of my underlying assumptions t about research priorities, questions worth investigating, and constraints in answering those questions.
This takes me to the PLOS open data announcement. Let's get this out of the way: not a bad idea. But a tough roll out. I appreciate that PLOS is in a difficult position here. PLos ONE covers all of science. That's a big big tent.
I mean really big. Easily underestimated, when was the last time you seriously talked to someone who works in a research area vastly different than what you do, easy to underestimate.
Here's the list from PLOS One: Biology & Life Science, Computer & Information Sciences, Earth Sciences, Ecology & Environmental Sciences, Engineering & Technology, Medicine & Health Sciences, Physical Sciences, Science Policy, Social Sciences.
That is a massive diversity of approaches to Science. I don't even know what I could confidently generalize about all of these areas. Social Science itself includes many disciplines. This is not simply 5 different ways to do biology. The data that PLOS is requesting varies greatly in, size, type, difficulty of collection, uniqueness, etc. etc.
This difficulty with figuring out the in and outs and whathaveyous of the big tent is relevant for any big how to improve science proposal. These proposals frequently suffer from a bit of discipline (or institution type) myopia. Some the differences across disciplines are merely historical accidents. Others may reflect real differences in constraints, goals, current professional realities, etc. That these differences exist, are important to be cognizant of, and cannot be understated.
(I am a little amused at both large video data and human data being mentioned as unusual corner cases. There are three or four labs on my hallway that use both of those.)