Well, yes and no. Yes, to a certain extent papers don't have enough details for replication, but this is mainly because you just can't fit enough details into t

Well, yes and no. Yes, to a certain extent papers don't have enough details for replication, but this is mainly because you just can't fit enough details into the scope of a journal paper. In a PhD thesis, you should be able to. And no, it's not that replication becomes a guessing exercise, but it's the fact that the time spent setting up, debugging and then running a replication of even a perfectly described benchtop-style experiment costs much much more than the actual equipment involved. And the ROI is poor, since replication studies are hard to publish.