How do we create a social web using just the web?

[scripting]

I sometimes get well-thought-out questions about how this stuff works.

Until the whole package is released I imagine it could be pretty confusing.

So I like to answer these questions with pointers to where it's explained, or if it isn't explained, or if it represents something I still have to think through, that would be helpful -- point me to places where I need to do some work.

And the more I explain it, the better I become at explaining it. So if you have a question here's a good place to ask it.

We have guidelines for discourse. This is not a place to debate. ;-)

[scripting]

BTW, a frequently asked question.

How do we make this work with Twitter, Bluesky, Mastodon, etc.

Answer: We don't.

I spent 10 years trying to interop with these systems and it doesn't work. And it's the wrong way around. We work on the web, they should make the effort to work with the web too and then we never have to do anything to interop with them.

They're trying to be platform vendors, and you have to be very careful when you sign on to that.

They are imposing limits on the parts of the web we get to use in their systems. After ten years, in 2017, I realized it wasn't worth it.

I couldn't use links because Twitter and Facebook didn't support them.

I had to use titles because Google Reader made my writing look ridiculous because they would mock me for not putting a title on a 25 word blog post. I am a better writer than Google Reader.

On and on. They can't get together on what they're doing, and that's because they want the power to break us any time they want. And believe me, they do that.

We gave up too quickly.

If you want to see a list of features that to me are non-negotiable, look here.

https://textcasting.org/

[asim]

The way you speak about it makes me think we got SMTP right for email. It works, it's federated, it was a long lasting protocol. HTTP has done well but then it became a way to monetize well anything so naturally silos emerged. The protocol interop has sort of failed. Everyone has an API but they all look different, behave different and now with AI it's all a walled garden to protect the data. Efforts like ActivityPub and the AT protocol are not much better. Something about the early era of protocol design worked. Now it doesn't.

Creating a social.web using just the web is hard unless you can really go back to the protocol layer and extract, publish, consume in bites of information that are truly independent. I don't know just riffing. What do you think?

Edit: I guess that's what RSS was good for ;)

[scripting]

even better..

https://daytona.scripting.com/search?q=textcasting

that'll give you lots of ways to zero in on the idea.

[asim]

I read through a lot of that and it's amazing how so many people have similar ideas yet we never seem to get to the place we want to. A long time ago I had the idea of a "micro communication protocol" - MUCP.

It wasn't for blogging or chat specifically, it was a service layer above http. I spent a lot of time doing RPC stuff in the 2013-2015 timeframe and then wrote a framework around that for microservices development.

I think the problem isn't just posting, it's all forms of text based actions at the next layer up which the API didn't uniformly solve and that RPC only partially got right.

XML got replaced by JSON. RPC got replaced by APIs and we're no better off. Companies like Google wrote binary rpc protocols for cross service communication and standardised tools which unlocked massive value but here we are and none of it really translated to the layers above except for within their own products. We don't have a universal system of access to blogs, search, weather, etc. it's all siloed.

Maybe being focused on a niche use case allows broadening it later but I also feel like keeping that in mind when you work on something means it can be applied to any building blocks that need to interop.