Re: WebID default serialization for WebID 2.x

Hi Jonas,

On 22.01.22 13:19, Jonas Smedegaard wrote:
>>> [and] Dan Brickley and Libby Miller [...]:
>>>> People think RDF is a pain because it is complicated. The truth is
>>>> even worse. RDF is painfully simplistic, but it allows you to work
>>>> with real-world data and problems that are horribly complicated.
>> I will try to phrase it in a diplomatic manner:  It kind of became a
>> recent trend to talk down Linked Data achievements.
> "Talk down"?  Seems we are talking past each other here, and I am truly
> sorry if I have somehow offended you or some efforts you hold dear.

Not you or anyone on the list so far. There are some papers and tweets, that are meticulously listing which datasets are not online any more. The great thing about the web and linked data is that it evolves, so things go stale and new things appear. Without meaning to offend anybody, I will list different perspectives and also which improvements, we need:

* People pointing at stale datasets: They are ignorant that it is supposed to change, but also right in a way, that we do not know how to track the changes well, i.e. how do we discover and update metadata like the ones used for the LOD Cloud. For WebID: What is the main means of discovery here? Follow foaf:knows is not in the spec. We could collect URI lists with an official validator (SHACL or otherwise) (with a consent checkbox) and then make these URIs public. Could we mandate some form of discovery mechanism? The goal here is to have a better overview of adoption and also track evolution, which is a cool feature and also makes a better point about the nature of linked data. What about license?

* People on a high horse, who have the necessary experience to work well with Linked Data/RDF: They say it is simple and try to ignore the many gaps, pitfalls and variations, instead of addressing these head on. This basically prevents innovation and adoption. I would say that simplification can simply be achieved by externalising the right kind of know-how. We can put it in the WebID 2.0 spec. People can publish their other Linked Data like they publish their WebID. An example here is spilling out how to do content negotiation for Linked Data and not shouting from the high horse that this is already covered by conneg / web architecture. As far as I see it, one of the biggest use cases for content negotiation is multilingualism. Personally, I never had to use it for I18N/Localisation, it is not common.   You can not assume that people, who get into Linked Data, know content negotiation already. I know that it is some work to engineer it down and decide on best practices. WebID has the chance to become a canonical, technical reference to Linked Data. In particular: how to use JSON-LD in a simple way, that it is totally fine to just put .ttl# files on a file server, how to cache Linked Data properly, committing to some properties/classes of FOAF in particular Agent, Person, Organisation and build a reference validator as part of the spec for people on foot that just copy the examples without reading the whole spec. Tackle RDF 1.0/1.1 problems. Decide on a definite list of 3-6 formats (3 is better than 6) and modalities and say that WebID publisher MUST pick one and MUST conform to all the MUSTs. Consumers MUST implement all formats and modalities to call themselves "conforming implementations". Write it in a way that prevents any discussion on "how to actually implement the spec".

-- Sebastian