from Hacker News

"This process only seems straightforward. In reality, the entire operation has about as much built-in voodoo and superstition as a major league baseball team. No one is even quite sure why "yield rates" (the percentage of good chips) are what they are. Thus, when they suddenly fall, there is general panic. When they rise, fab managers scurry about trying to find the magic key. Short of a scientific solution, the managers revert to an almost primitive faith in retaining every vestige of status quo. One of the women wore a silk blouse that day? Then she'll wear it every day. Someone else used his left hand rather than his right to dip the wafer in acid? Then from now on he's a southpaw.

These examples are a bit extreme, but every fab lab in the Valley can come up with a similar example. In the old days, before clean rooms and laminar-flow ventilation, making chips was even more unpredictable. Yield rates would plummet for no apparent reason. For example, one company found its yield dips corresponded to phases of the moon: It seemed that high tides also raised groundwater levels and thereby increased the moisture in the lab. Another lab found that during their menstrual periods female lab workers secreted extra oil in their hands, ruining wafers.

The lab at Signetics nearly shut down in the spring of 1965—until it was discovered that the crop-dusting taking place at the farm next door was wrecking chip surfaces.

But perhaps the most unusual case of early low yields was discovered by a local firm faced with a disconcerting and destructive appearance of organic crystals on wafers. Thorough chemical analysis finally came up with an answer: the crystals were formed by droplets of urine left on male workers' hands after they used the restroom.

One obvious result of all these early problems is the now mandatory use in labs of plastic gloves or fingertip protectors (nicknamed "rubbers for midgets" by workers)—and, of course, the growing use of automation and specialized robots."