[nudq]

Controlled compromise of privacy for the sake of scientific insight seems like a good idea, until you realize that we either get profoundly non-replicable junk "science", or continued and unlimited re-breach of privacy for the sake of replication. Neither is any good.

I'm reminded of Raj Chetty who publishes papers based on exclusive access to IRS tax return data. (https://www.sciencemag.org/news/2014/05/how-two-economists-g...) Not real science unless you can have access to that data, too. You can't.

[tvanantwerp]

While I'm normally all for access to raw data used in research, that's pretty hard to do with IRS data without massively violating Americans' privacy and exposing them to identity theft. It's fair to be skeptical of anything using locked-up data, but I don't see a good way around the problem.