Even worse: An FAA certified system used on commercial jetliners for more than five yaers. The manufacturer started using a different date code of an FPGA, and it was found during testing that the system fire sensor input failed when it got hot (no, I'm not making this up). I was contracted to review the design, and found that incorrect clock distribution practice had been used. This was not caught during design reviews, validation, verification, or certification. This was arguably the definition of a proven product, and all it took was a new batch of parts to render it unreliable and potentially unsafe. Specification limits and design practices are there for a reason. Violating those limits and practices is risky and potentially unsafe. The consequences may not become evident for years.