I've been down on myself about this. I've been at it for coming up on 30 years. I've written scores of programs and applications. There are only a couple still in use. Rather than be upset about it, I find that it's more helpful to think of coding as tooling. As in, actual tooling that machine shops make: dies, jigs, gauges, etc. You know, the stuff that's used to make the actual stuff. Tools go out of date. Tools get upgraded. Product lines change. Tools get scrapped. It's all ephemeral. It's just the nature of it. And this applies to "the bigs" as well, like Facebook or Google. The individual components and services are just tooling in service of the overall product, and rollover and change as time goes on. The only coding that ossifies into archeological strata is COBOL on mainframes.
Over time I've come to realize that the various "IT things" in business which retain the most value are databases. Not because they never become outdated or redundant but because the data in them can always be migrated to a new schema to be used by a new application. Usually the "tooling" to do this is exactly the kind of work I might do which becomes something that is no longer in use.
(I guess here I should mention the context that I'm not coming up on 30 years; instead it's a little over 10.)
The "no longer in use" part is something I think I ultimately disagree with. It's kind of an "application of Theseus" situation. Where did this data really come from? If it was an older application, did that application ever really go away or did it just become what replaced it? Anyway, I guess I just have to hope I still have this outlook in ~20 years.
