I’m obsessed with this stuff so this was an incredibly gratifying read.
I really think we can use something like the block protocol to create something of similar utility for semantics and composition as Vue or React has for view layers. There can’t be The One True Block for every single intent, but we can make creating structured data so much easier and rewarding than it is now.
Early on in my career I was incredibly obsessed with HTML semantics and could rabbit hole on that for ages. I still love it, but as the article mentions, the relativity and subjectivity of real-world semantics inherent of ontologies it makes it hard to embrace - not only in the context of web apps and content, but as teams as well.
I’m very hopeful that just as view layers provide enough abstraction and flexibility to compose user interfaces from the same blocks, something like the block protocol can allow us to construct and compose data from the same blocks in a productive and efficient manner.
It's a bit mysterious to me how one can talk about structured data and not also discuss schemata and validation. If a schema exists, one can avoid reinventing the wheel. Name your poison: DTD, XSD, RNG, JSON schema.
This is an interesting read with concepts to keep in mind when designing some application. I've worked with open data, ontologies and semantic web, and I'm afraid a lot of the difficulty is in cross-cutting concerns, interoperability and boundaries. As some of the slides mention, there is no easy sharing of data between blocks, which is essential for flow. Providers of said blocks will not easily open their gardens to the outside world because of business lock-in or because it's just too difficult security-wise. There will be irreconcilable differences between a conceptual Meeting schema and the actual data structure provided by a block provider's API. And the idea of creating thousands of block types will make things so complex you will basically end up with what we already have, i.e. programming with classes, conditionals and all sorts of inheritance.
I understand the vision, and I think it's good to keep in mind when implementing software, but as a seasoned coder and security person, I see a lot of technical and (anti)competitive hurdles. In practice, the market does not want interoperability.