0

Suppose there is some (public) data set that censors (in the common parlance) any data satisfying a certain criterion. By a rather obvious workaround, these censored data can be exactly recreated, giving one access to the full (uncensored) data set. I have thoroughly scrutinized the data use policy and this is not in violation of data use restrictions, so long as the censored data is not published.

I would like to use the full data set for some of my research, and I fully intend on obeying all data use restrictions in the process of publishing, but am a bit stuck on how I respond to someone asking exactly how I accessed the data in the first place. Refusing to provide details makes the work seem dubious, while I hope to be as transparent as possible for the sake of reproducibility.

Any help on this matter would be greatly appreciated.

9
  • 1
    What sort of "censorship" are you asking about?
    – Buffy
    Oct 4, 2022 at 17:41
  • 1
    Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking.
    – Community Bot
    Oct 4, 2022 at 18:04
  • 1
    Related: Is it ethical to reconstruct a dataset without permission? Explaining how your question is different from this one (if it is) might be a good way to clarify what you're asking.
    – cag51
    Oct 4, 2022 at 18:16
  • 1
    It seems unethical to me not to let those making the data available know how easy this is to bypass if they had a good reason.
    – Bryan Krause
    Oct 4, 2022 at 20:22
  • 1
    Do you mean censorship in the common parlance, censorship in the statistical sense, or redaction? Please edit your question to clarify.
    – shoover
    Oct 4, 2022 at 21:05

1 Answer 1

3

In your publication(s), You just state the truth: You took data from a public source and used the data therein to reconstruct the data. You also state that you do not believe you are in violation of the data use agreement and that your use of data, and in particular your publication of the results, does not violate privacy.

Because knowing that data can be reconstructed helps people looking for full data sets, publication of the existence of a work-around has at least the potential for a breach of privacy. To be more precise, the breach of privacy has already occurred, but publishing the existence of a work-around makes it easier to find a work-around. This means that before you publish anything, you need to give the agency a chance to remove the data (and hopefully replace it with better protected data). This also means that your data will not be reproducible from public sources, so that you will have to post publicly a privacy protecting version of the data set on which you base your conclusions.

Finally, if you use a computer program to regenerate the data, you might use an anonymizer within the program (e.g. replace names with a hash value or slightly encrypt identifying information such as addresses by XOR-ing with a fixed string) so that no human eyes including yours sees the actual data. I do not know how bad the privacy breach is, so this might be a stupid suggestion.

Privacy protection is an active research area in fields like Computer Science. Mistakes in privacy protection are important data points. Their discovery should be treated like the discovery of exploits in software, where you give the owner of the data / software a decent chance to fix the problem before you announce the discovery to the world.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .