The New York Times also had an article yesterday [1] looking at some of the regular pro-Brexit people who just plain do not believe any warnings about no-deal. One quote though really jumped out at me since it was related to one of the earliest projects I ever did, working on a small bit of the Y2K problem, and I think it really serves as an example of a fundamental difficulty when dealing with public facing threats:<p>><i>Mr. Ridley compared the anticipation mounting before the Brexit deadline to the run-up to forestall the Millennium Bug, also known as Y2K, in which companies worldwide scrambled to avert technical breakdowns when digital systems switched from 1999 to 2000.</i><p>><i>“It’ll be like Y2K,” he said. “Remember that one? They were like panic, panic, panic, the world’s going to end, the electric grid’s going to go down,” Mr. Ridley said. “None of it is going to happen.”</i><p>I knew that this was the impression left with the general public after that, but it's really unfortunate because in my recollection Y2K is quite possibly one of the most unified, successful responses to a serious tech problem our industry has ever managed to pull off. Yeah, it didn't amount to much... after hundreds of thousands to millions of man YEARS of work. I know people who had gone full time digging through ancient code bases back by 1998 or so, wouldn't be surprised if some had started in 97. Initial issues began bubbling up well before 00 after all, stuff like credit card expiration dates that were a few years in the future. Enormous amounts of resources were sunk into working on it, sometimes for clean fixes and sometimes for hacks [2], and overall it worked. But of course then we ended up with "well what was the point of all the hype or effort, everything was fine!" and the public taking literally the opposite lesson.<p>I can't remember if there is a technical term for this class of problem (beyond "life in ops"), where it's like air, the general "success" state is "nobody even thinks about it most of the time" but the failure state is catastrophic. It's an issue with security too of course. Ops and security are non-revenue generating, but their absence can certainly be revenue destroying. It's hard to get budget support there.<p>It seems like a really hard problem, particularly when future testing isn't available and results are irreversible.<p>----<p>1: <a href="https://www.nytimes.com/2019/01/14/world/europe/no-deal-brexit-britain-uk.html" rel="nofollow">https://www.nytimes.com/2019/01/14/world/europe/no-deal-brex...</a><p>2: Old systems are still in service that are still using 2-digit dates, but just were patches so that 00 to 30 were assumed to be 2000s while >30 were 1900s, so there is actually still lurking remnants of Y2K that will come up again if not dealt with by 2030 or so. And of course there is the 32-bit unix time issue for 2038, examples of which have also cropped up a few times already too.