It's an odd and funny argument to make that your product is working well if it's always on fire or broken, which you would think an HN commenter like a sysadmin

It's an odd and funny argument to make that your product is working well if it's always on fire or broken, which you would think an HN commenter like a sysadmin would understand. It just seems the margins for error get ever thinner as your system scales. Other systems have figured this out, like Tokyo where they change over a major-use track in one night. Meanwhile, MTA shuts down the L for 15 months.