Old Languages Like COBOL Stick Around for Good Reasons

It blows my mind that we haven't solved this problem yet given half a century. That we're still so afraid of touching our code. COBOL may be COBOL, but who among us can honestly say that things are categorically different in 2020?

It got me thinking about languages like Haskell that resist this kind of breakage. And that got me thinking about why people don't use those languages for mission-critical systems like these. Probably it's because they tend to be unapproachable and their communities populated by language nerds, instead of people who just want to get things working and keep them that way. And that reminded me of this thing I saw a while back about finding a "simple subset" of Haskell that could give you all the functional and type-checked benefits without the dizzying theory: https://www.simplehaskell.org/

I don't know what point I'm trying to make exactly; I just find it bizarre and depressing that our business logic still has to be rewritten just because we want to change platforms. That we still can't build systems that can safely be upgraded piecemeal. I just don't get it. These feel like solvable, incredibly valuable, problems. It can make a programmer feel nihilistic.