Bias in physics may be not that important, but with social sciences it can be pretty strong. I know that the best papers have this system of review, that the reviewers don't know in which country the experiment took place, who conducted it etc, they also don't know other reviewers. The other big problem is publishing papers that PROVED something and not publishing the ones that didn't prove anything extraordinary, which also distorts the overview of the field. So I heard the best journals 'order' the research and is going to publish it regardless of the results. I don´t know how spread are those kind of practices, but this is a way to make (social) sciences better and more accurate. Blockchain could be a mediator in that for sure.