Disclaimer: This comment contains some self-promotion (as we are working on a related solution) Solid overall looks very interesting, and I can very much empat

Disclaimer: This comment contains some self-promotion (as we are working on a related solution) Solid overall looks very interesting, and I can very much empathize with the "decouple data and applications" magic aspect of it (even though they're not doing a great job describing it). My personal experience has been that with a good linked-data approach, building apps can become a lot easier, as linked (more specifically semantic data) mimics traditional information gathering much better, where you start with a small information point, and then enrich the data by just adding more and more related data points to it. I also found that a linked-data approach lends itself better to adaptive UIs, where one entity might be missing data, that another one has. All that being said, I feel like Solid made some (non-)decisions that might lead it down the same path as the things that were tried in the last Semantic Web hype: - Solid is using the same old linked-data formats, that are largely URL based. The problem with URL based systems is that the content behind those URLs might at any point disappear, or change. That makes it especially hard if you try to achieve some semantic conciseness, where the data you entered keeps at one point in time is guaranteed to keep its semantic meaning forever. - In a world where most application developers think about data models on the level of database tables, RDF/JSON-LD is too low level and verbose when trying to use it for building something sophisticated. W3C also standardized a higher level format, OWL2 (which can also be expressed in RDF(S)), which provides a nice abstraction level in its native form. Those are some of the lessons I've learned while working on Rlay[0], where among other things we've built a content-addressable adaptation of OWL2[1] for the decentralized web. We had to cut some small parts out of the specification so that the semantic stability of concepts is guaranteed when adding additional concepts to the worldwide ontology everyone is sharing, but overall the expressiveness is the same. Content-addressability has many nice side-effects, but the one I like the most when it comes to Semantic Data is that it allows for a much more organic evolvement of the concepts people are using. If you decide to add a new property to a concept, you can just do that, get a new content-addressing-hash and start using that instead of the old one. No need to add it to the de-facto centralized schema repository that is schema.org. I hope I didn't go off on too many tangents here. I've been meaning to write a blog post about the topics I touched on for a long time, but never really got around to it. Well at least this comment exists now. [0]: https://rlay.com https://rlay.com [1]: https://github.com/rlay-project/rlay-ontology https://github.com/rlay-project/rlay-ontology