So, after too many hours of digging through specs and primers and whatnot, here's a brief summary of why Linked Data and RDF and all are interesting:
It allows to link "databases" (datasets) across different sites, without needing centralized coordination. It's like federation for datasets, and there is a whole ecosystem of tools for merging and relating between such datasets.
It was essentially an attempt to extend the 'open web' to machine-readable data, not just human-readable webpages.
@joepie91 do you know of any practical uses other than in search engines?
@selfisekai I think that even within search engines, there is still a lot of room to do cool stuff with it; semantic/specialized search, for example.
But outside of that, it also seems quite useful to answer arbitrary statistical questions about Things, for example.
Like, taking food as an example: "show me all the milk chocolate bars that include ingredients which were at some point handled or processed by a company with known human rights violations, and those that weren't" is an entirely plausible query.
But it's unlikely that a singular entity would be maintaining both a complete database of chocolate products *and* a complete database of corporate human rights violations (those are very different domains of specialization), so the ability to do a query *across* two such disparate datasets would be very helpful.
There's a lot of technical details, and a lot of issues that hampered its adoption, and a lot of specialized tools, but like, once you ignore all the details, that's the pitch and that's what it does (and the design *does* broadly seem to make sense).