I’ve described in the past how the U.S. government has adequate policies for data archiving on paper, but that NSF had been totally co-opted by non-compliant scientists and their administration of data archiving could be described only as ineffective or non-existent. Doug Keenan wrote in on some calls to NSF:
NSF grant conditions are specified here: http://www.nsf.gov/awards/managing/general_conditions.jsp
There are different grant conditions depending on the year in which the award was granted. In fact, though, all grants made since 1995 (possibly earlier—conditions prior to 1995 are not listed) include a condition entitled “Sharing of Findings, Data, and Other Research Products”. My interpretation of the condition is that researchers must share their data.
So, I telephoned the NSF and spoke with someone in the Policy Office about this. (The person was Beth Strausser.) I was informed that my interpretation was correct, although there could be some dispute about how raw the shared data had to be. Apparently, partially-processed data might suffice; it seemed unclear as to how that would be judged. I then asked what should be done if a researcher refused to share data. Strausser said that in principle the program manager might get involved, but that in reality the sharing policy was “self policed” by scientists in the field. She briefly elaborated: the NSF relied on other scientists in the field to put (social) pressure on a researcher who refused to share data.
I can imagine rare cases where that approach might not be fully effective. Perhaps the FOI Act would be of use in such cases? I telephoned the Justice Dept. and spoke with an attorney in the Office of Information and Privacy (which deals with FOIA matters). (The attorney was Anne Work.) I was informed that because the NSF does not have the data, there is very little chance that a request made under the FOIA would be achieve anything.
Obviously “social pressure” has not accomplished anything with Lonnie Thompson or other non-compliant scientists. What a ridiculous way to produce and verify results being relied on for climate policy?