Then you're one of today's lucky 10,000: "Robust De-anonymization of Large Sparse Datasets" http://www.cs.utexas.edu/~shmat/shmat_oak08netflix.pdf

A lot of environmental medical research results in dense and small data instead of sparse and large data, which obviously means it's even easier to de-anonymize.

And the really important take away from the above paper is that you need to know about all the other public data in the universe in order to really ensure anonymity. Even when anonymity seems like it's "obviously" not going to be a problem.

You never know what dataset will be leaked/published tomorrow correlating columns D-F in the spreadsheet you gave EPA to someone's first and last name. So privacy is like computer security: you can never be totally sure. So when the stakes are high you should be extra cautious. And the stakes when sharing medical research are very often high.

> If that's true, how people validate any medical research at all?.. I have very hard time believing that's the only way to do modern medical research known to our civilization.

This isn't exactly a new problem, so scientists have had a little while to figure out how to make peer review work.

You share the data. You share with trusted, relevant third parties who promise to preserve the anonymity of study participants. Just becuase you do not publish possibly de-anonymizable and definitely sensitive medical information on a public website does not mean you don't share the data with interested and relevant third parties.

> But you seem to have arrived to pre-determined conclusion already, and there in fact can be no middle ground that would be acceptable to you, as it seems.

On the contrary, I do literally exactly that in my very first post in this thread: There are some common-sense workarounds. For example, requiring an EPA or third-party audit of the original dataset and analysis. Or simply carving out exceptions for research whenever it can be demonstrated that an IRB (and therefore federal law) would not have allowed the research without restricting public access to data/analysis.

The situation where IRB says "no" to public disclosure should explicitly trigger exemption, with the duty falling back to the agency to prove the IRB was wrong in the first place. The fact that this exemption is not automatic is what makes this a catch-22 that pins researchers in-between their IRB and the EPA administrator in charge of evaluating their exception.