I just left the following comment on the OSTP blog: (We submitted an NSF proposal yesterday, so this is particularly well-timed.)
As a graduate student and researcher in Computer Science I would welcome the application of NIH’s open publication policy to NSF funded research as well. I just helped submit an NSF proposal and would welcome such guidelines on our research. We already follow them for the most part, and it would help us ensure that conferences and publishers won’t object to the dissemination of the results of our research if it’s not just our own desire, but an obligation from our funding agency.
I would also appreciate if you look into the feasibility of establishing open data and open source requirements for the CISE division of NSF where funded projects would be obligated provide data generated as part of the research (this isn’t always reasonable, so some care will be necessary in formulating this policy) as well as source code used. Right now, especially in computer science, it’s extremely difficult to replicate the results of experiments and often we find that researchers have to invest time in reverse engineering someone’s experiment simply to be able to compare the results one research got with the results of another or their own work. In addition, university technology transfer departments sometimes get in the way of researchers who try and open up their work. Making this a firm requirement from the funding agency provides the researchers with the credibility and leverage they need to ensure technology transfer departments stop interfering with dissemination efforts.
It’s rare someone ever asks for a requirement, but in this case I think extending these types of requirements to NSF programs will give us the power and leverage we need to overcome the obstacles in the way of disseminating our work in the fashion we’d like to be doing anyways.
As a graduate student and researcher in Computer Science I would welcome the application of NIH’s open publication policy to NSF funded research as well. I just helped submit an NSF proposal and would welcome such guidelines on our research. We already follow them for the most part, and it would help us ensure that conferences and publishers won’t object to the dissemination of the results of our research if it’s not just our own desire, but an obligation from our funding agency.
I would also appreciate if you look into the feasibility of establishing open data and open source requirements for the CISE division of NSF where funded projects would be obligated provide data generated as part of the research (this isn’t always reasonable, so some care will be necessary in formulating this policy) as well as source code used. Right now, especially in computer science, it’s extremely difficult to replicate the results of experiments and often we find that researchers have to invest time in reverse engineering someone’s experiment simply to be able to compare the results one research got with the results of another or their own work. In addition, university technology transfer departments sometimes get in the way of researchers who try and open up their work. Making this a firm requirement from the funding agency provides the researchers with the credibility and leverage they need to ensure technology transfer departments stop interfering with dissemination efforts.
It’s rare someone ever asks for a requirement, but in this case I think extending these types of requirements to NSF programs will give us the power and leverage we need to overcome the obstacles in the way of disseminating our work in the fashion we’d like to be doing anyways.