inevitably some of your learning efforts are sunk into the tools, apis, and platforms you use. those can compound over time, making you more effective for the rest of your life, or simply be a pleasant memory. spending time on tools, apis, and platforms that are not dependent on some company's goodwill is helpful here; none of vulkan, x86, risc-v, linux, freebsd, emacs, perl, python, angular, or flask is dependent on any single company
(x86? yes, without amd, x86 would have died around 02000 like most other 80s architectures did, killed by intel's itanic blunder.)
in the 90s i invested time learning a lot of different tools and platforms. most of them are no longer useful to me because they were dependent on some company or other: vax, vax/vms, informix, netapp, mfc, sunos 4, purify, clearcase, pv-wave idl, erdas imagine, khoros cantata, irix, windows nt 3.51, cde. but, for the most part, those that were open standards (or, better still, free software) remain useful to me to this day: diff and patch, shellutils (now coreutils), tcp/ip, perl, netbsd, emacs, python, html, css, the i386 architecture, bash, vim, c, c++, sql, gcc, gdb, samba, ssh, postscript, tcl, tk, etc.
admittedly x.25, smtp, ftp, perl, tcl, twm, and fvwm are not as useful these days, and visual c++ and win32 are still somewhat useful. so it's not 100% black and white. but the longevity difference is extremely striking
however, see https://news.ycombinator.com/item?id=41695702
inevitably some of your learning efforts are sunk into the tools, apis, and platforms you use. those can compound over time, making you more effective for the rest of your life, or simply be a pleasant memory. spending time on tools, apis, and platforms that are not dependent on some company's goodwill is helpful here; none of vulkan, x86, risc-v, linux, freebsd, emacs, perl, python, angular, or flask is dependent on any single company
(x86? yes, without amd, x86 would have died around 02000 like most other 80s architectures did, killed by intel's itanic blunder.)
in the 90s i invested time learning a lot of different tools and platforms. most of them are no longer useful to me because they were dependent on some company or other: vax, vax/vms, informix, netapp, mfc, sunos 4, purify, clearcase, pv-wave idl, erdas imagine, khoros cantata, irix, windows nt 3.51, cde. but, for the most part, those that were open standards (or, better still, free software) remain useful to me to this day: diff and patch, shellutils (now coreutils), tcp/ip, perl, netbsd, emacs, python, html, css, the i386 architecture, bash, vim, c, c++, sql, gcc, gdb, samba, ssh, postscript, tcl, tk, etc.
admittedly x.25, smtp, ftp, perl, tcl, twm, and fvwm are not as useful these days, and visual c++ and win32 are still somewhat useful. so it's not 100% black and white. but the longevity difference is extremely striking