@smallcircles If you control both ends of the pipeline, you don’t need any validation or interoperability. XML technologies are thus not needed anymore. The price you pay as a developer is a more and more heavy dependency on a HTML5-compliant web browser. The web as a whole pays an additional cost that the developer becomes the little tyrant of their corner of the web, and the browser ecosystem is collapsing into a monopoly. We can’t build anything useful on top of that.
Discussion
Loading...
@smallcircles @strypey @FinchHaven @naturzukunft2026 There are different ways to tell the story. Please consider this: the popularity of JSON stems from the promise of HTML5 that the developer would control the software at both ends of the data transmission. The web browsers taught developers to dismiss any concern for interoperability. json-ld is trying to fix that (and failing because of the external contexts, but it’s a different topic).
@smallcircles If you control both ends of the pipeline, you don’t need any validation or interoperability. XML technologies are thus not needed anymore. The price you pay as a developer is a more and more heavy dependency on a HTML5-compliant web browser. The web as a whole pays an additional cost that the developer becomes the little tyrant of their corner of the web, and the browser ecosystem is collapsing into a monopoly. We can’t build anything useful on top of that.
Yes, thank you. It is a bit of everything. Where I worked in Print production at the time we used all the XML standards to max extent and got great benefit from doing so. It all crumbled down very quickly with the take-off of Web where #JSON mapped way better to JS than #XML did, esp. when also honoring the intricate web of (de-facto) standards that had emerged, and how the XML ecosystem got captured by corporations peddling expensive products, such Enterprise service buses and the like.
It would have been great if #XHTML had stuck around. In hindsight the shift to HTML5 may be seen as a move by the WHATWG to establish their dominance over the #browser market.
As for #JSONLD it must see to rewin hearts and minds, must inspire and convince people to take the bet of adopting it.
(1/2)
@smallcircles
> As for JSON-LD it must see to rewin hearts and minds, must inspire and convince people to take the bet of adopting
Agreed. But to do that requires explaining to people what it's for, and the potential benefits over vanilla JSON that make it worth the extra dev effort.
Reviving the Semantic Web vision - the *vision* not previous (failed) implementations - seem to me to be the most obvious way to do that.
But as always, I'm open to other suggestions : )
So what is the Semantic web actually, is the real question to ask. And what is semantic social networking if we apply it to that field. What does it bring in terms of value and benefits, new opportunities that didn't exist before. To LD proponents these may be very clear and obvious, but to many others hearing it is saying nothing to them. And also, when they are enticed to take a deeper look they find an every increasing amount of inter-related standards and skills to master around them. Aggregate linked data in data store, now what? How do I query it? Oh, XQuery you say, but I already do everything in SQL.
@smallcircles @strypey The main benefit I see is, you can introduce new applications on an existing social graph, without anyone else having to care about your thing. For instance, you publish your software contributions on your microblogging platform, and your followers that have a code review application can review them. Publish your papers on a different application, and your followers can review them. You don’t need a separate graph for code and papers reviewers.
1+ more replies (not shown)
@smallcircles @strypey you can convert a SQL database into a quad store easily: just have a big table with 4 indexed string columns (subject, predicate, object, graph), and boom, you have a quad store.
(2/2)
One obvious selling point of the Semantic Web vision, in the age of #MOLE Training hype, is to point out that it's about enabling dumb software to do smarter things with data. By enabling humans to add more machine-readable metadata. Rather than trying to make software smarter, so it can make good guesses about data without such context.
Most of the snake oil people on the "AI" hype train sell as unique benefits of leasing a Trained MOLE can be done with Semantic Web tools.
I am not familiar with the term MOLE but AI is highly problematic and disruptive indeed. Ironically enough I've seen interest in LD coming from the LLM field, to help AI output be more predictable and 'smart' based on the semantics of the linked data it is trained with.