While I agree on the data point, it would kill so much research. It is bad that a lot of research validation basically comes down to "trust me guys", but with d

While I agree on the data point, it would kill so much research. It is bad that a lot of research validation basically comes down to "trust me guys", but with data being both very valuable and often times highly sensitive, it can be really difficult to just publish the data along with the research. A decent compromise would be to at least require meta-data to sufficiently exclude some flaws. A different approach could be to have researchers document and publish the process of th research, similar to a git-repo with the main branch being completely off limits to history-rewriting.