However, if poor reproducibility is taken as a criterion to reject results, you must reject the top quark, many plasma fusion results, cloning mammals (which works less than 0.1% of the time), and semiconductor production before 1955, which failed roughly 90% of the time for many types of devices.