[gumby]

Some things are inherently transient (i.e. will be discarded and rewritten) so why not do them as quickly as possible and use the latest and greatest? Examples: a business dashboard (yes they still exist) or some back end for an IoT gadget that will probably get canceled, and if not, well, you won't be around (but today you will be rewarded for speed).

Some things will outlast your lifetime. COBOL banking systems are still undergoing active maintenance, and the aircraft reservation systems are stuck in the 60s/70s. Unmanned sacecraft can operate for decades and the hardware, at least, is not replaceable.

The trick is to figure out the non-obvious criteria that cause the thing you're working on to fall into one of the other bucket. Nobody thought the banking systems would last so long (else there wouldn't have been a Y2K crunch). And some of those quick-and-dirty business systems stick around forever.

[cognitiveosman]

Really nice dimension you highlighted - It is indeed many legacy systems are running business critical application which are quite expensive to replace, we see that in the finance, insurance, banking, and aviation domain. On important dimension to consider when learning technologies is to always invest on learning the fundamentals which rarely change with time. For example, investing on developing sound understanding of TCP/IP protocol will be helpful regardless of what technology vendor you will use and the basics remain (almost timeless)