[8note]

I find it intriguing that reproducing the experiment isn't a part of the review, or visiting the lab to look for potential flaws in their set up.

These reviews seem far from what you hear about for disproving N rays and the like

[swiley]

Do you checkout your co-worker's branches and run the unit tests for every PR they make? In an ideal world that might be nice but in the real world with normal sized projects that's to time consuming so everyone just reasons through patches.

[andi999]

If reproducing were a requirement for reviewing, publication cost might be exorbitant (like when CERN tries to publish a paper you have to rebuild the particle accelerator?)