Reliability and security. Can we have those things anymore?
Back when large-scale system life cycles went on for years or even decades, we had some really solid, proven systems. Nowadays tech companies are so in love with shiny new things that it seems like technologies get ditched for something new before they have a chance to mature. I wish we would stop this make-something-new-for-reasons attitude and slow the hell down and focus on making what we have actually work as it should. The faster software changes, the less reliable and secure it becomes. If you don't believe me, update your Windows box.
And it will only get worse I am afraid. How many "dead" languages will there be in 20 years? I maintain a key program initially written in Fortran in 1981 and still gets new features. It is like playing Jenga but I am powerless to replace it. Pretty sure it will still be running after I retire.
Well, you see, languages are tools. Hammers were invented quite a long time ago, and they're still produced, used, mantained and improved. Why? Because they are very good to solve some classes of problems. And that can be said for pincers, knives, C, FORTRAN nad any other tool, IMHO.
FORTRAN is still a very good tool to solve the class of mathematical problems but you wont see many programmers making it their language of choice for that. No modern IDE, not hip, never was exposed to it, etc. many reasons I suppose. It's not still used because it's a good tool, it's still used because it would be expensive to rewrite the programs.
A significant majority of the software that manages our lives is obsolete.
According to whom? Just because software is old doesn't mean it's useless, it's as useful as it ever was, and given the fact that it's been maintained for so long it's likely more reliable. What do you want to use for a mission-critical system, an old program that has been doing the job fine for 30 years, or something just cooked up by the new kid?