I can give you data that say anything. The reproduction that counts is where you obtain similar data independently.
If you want to validate that my claims are justified BY MY DATA then you need my data.
If you want to validate that my claims are justified by the universe, you need to get your own data and should avoid being polluted by mine.
Having said that, I competely support publishing data by default. But not for the reason you state.
I can give you data that say anything. The reproduction that counts is where you obtain similar data independently.
If you want to validate that my claims are justified BY MY DATA then you need my data.
If you want to validate that my claims are justified by the universe, you need to get your own data and should avoid being polluted by mine.
Having said that, I competely support publishing data by default. But not for the reason you state.