But on the other side of the coin, do you really want to be locked into a decades-old solution to a "solved problem" forever? Or are there still improvements that can be made to reduce friction and human error?
"Machine/human readable data exchange format? Yeah, that's a solved problem - we use XML for that. What's that? You want to reinvent the wheel? JavaScript Object Notation, are you insane? Didn't you bother seeing which exchange formats are already out there? XML can already do everything JSON can do and more! Plus, JavaScript is the exact opposite of what we value here, get that crap out of my face; like I said, XML is the one true solution to data exchange and forever will be."
S-expressions also have no standard for comments, and can't distinguish maps from lists.
A S-expression based serialization standard would be a bit cleaner than JSON, but it's enough of a change that it's worth redoing anything.
People don't discuss solutions in a vacuum. Ideally they'd think about what the potential fallbacks of using XML alternatives are before jumping into a new tech. Which they will, unless they're really really new
[1] https://json-schema.org [2] http://www.jsoniq.org [3] https://www.p6r.com/articles/2008/05/06/xslt-and-xpath-for-j...
Easy to say with hindsight now that it is massively popular and has a huge ecosystem of parsers for every conceivable language. At the time it was invented though, it was yet another data exchange format and I'm sure there were a lot of grey beards pooh-poohing it.
The job market is smaller, but so is the pool of developers. Companies tend to be more flexible because of that and are often open to remote work. I'd much rather work in a sane niche market than deal with the mainstream churn.
We have local Clojure meetup in town, and when I first started going there pretty much everybody was using Clojure as a hobby. Today, we have a bunch of companies using it in production, and all of them are actively hiring. The last three coop students I had all ended up getting Clojure jobs. I imagine this varies based on where you live of course, but another option is to simply introduce Clojure at a place that's using something else. That's where Clojure jobs come from in the first place at the end of the day.
You don't see in other disciplines such as physics or chemistry where people spend years learning about existing research before actually starting to contribute.
With programming the barrier to starting to write code is much lower than to actually learning the background research. People don't bother looking at what's been done already, and just start "inventing" things.
More often than not this result in half baked ideas because the authors of projects don't really think about the full scope of the problem. Then once the project starts hitting limits in practice, people start kludging things on top of it, and eventually it becomes unwieldy to use. Then somebody comes by and does the same thing simplistically again, and the cycle repeats. Nothing new is learned in this process, and you get churn for the sake of churn.