It was my company's decision to discontinue the testing because it wasn't cost effective, even though there are dangerous jobs in the wafer fabs. One can make the case that somebody who consumed MJ a month ago could be a direct cause of them making a typo, but what does it cost to weed out such a person? And would that even be a case worth considering if someone isn't high when making the typo?
The problem is, there still isn't a widely available test for "Is this person high right now?" Like a breathalyzer for alcohol.
And that creates a host of problems with legalization and enforcement of things like DUI, because the line which used to be at yes/no is now reduced to subjective observation of the accused impaired driver, which can be shredded by a competent defense attorney. Other contexts may vary, but the old line of present or absent was a clear threshold. Making that line fuzzy will not help matters, just present a gray area that will feed attorneys and cost money.
Even more so, identifying limits to determine what constitutes quantifiable intoxication, as has been done with alcohol (which has its limitations, too, as do the testing methods), becomes a problem in defining when a person is impaired, and to make that even worse, unlike alcohol, where X ounces of N concentration for a given body weight will, generally and predictably yield a given Blood Alcohol Content, making avoiding a DUI technically possible, how do you determine what amount of weed smoked or ingested will yield what concentration of intoxicating cannabanoids?
More problems than it is worth, imho, but I'm neither a user nor a fan of it. Others will disagree, not only about intoxication levels, but about latency and latent effects which seem apparent to non-users--just like the odor of skunk (but that's another story).
It seems odd to me that the same people who bitched mercilessly about the odor of tobacco smoke aren't out up-at-arms about the odor of skunk, but what do I know?