[Eelongate]

It won't change until a track record of producing studies that fail to replicate hurts the career of an academic more than not publishing anything at all. As long as publish-or-perish provides a stronger incentive than the negative repercussions of publishing nonsense, this will continue.

[hermitdev]

I don't know if one necessarily wants to punish for failure to replicate, but I do think that falsified data/records should lead to a death sentence of a scientific career or result in criminal charges if necessary (e.g. for defrauding the associated institutions/grantor). I guess sort of like what Elizabeth Holmes is facing currently.