My blog is on Mastodon! Check out @blog. It would be a bit hypocritical of me to not use #Pterotype when I am asking others to use it... 🤓

@jdormit @blog Congratulations! I'm absolutely with you to open up the "entire internet" and I see that ActivityPub might be a good way to go.

One question is how we get it going for a critical mass and how we avoid that it cannibalizes other decentralized protocols as @fluffy pointed out.

I would say that together with new protocols that foster decentralization we need to promote the benefits of self-hosted blogs and instances although they are not as convenient as the existing services.

@xldrkp @jdormit @blog I would love to see a world where ActivityPub coexists with Atom/WebSub/WebMention, not the least of which being that the latter is a lot easier to build software around at scale (and in particular they do a better job of scaling around failures, and also work way better with static or low-resource hosting).

@xldrkp @jdormit @blog also I should add that the on,y reason I have any interest in ActivityPub is that it at least has a model for private-ish communications (albeit not a very good one); atom currently lacks any mechanism for that which is satisfactory to me. It would be possible to combine atom with some other auth mechanism but then the problem is getting user agents to buy into it, which is... difficult at this point.

@xldrkp @jdormit private per-reader feeds with essentially a session ID in the URL is workable but then there’s issues with discovery/subscription and it becomes way too easy for that to “leak” in e.g. reblogs or whatever. Another idea is having all entries in the feed but private ones being encrypted AACS/DVDCSS-style, which feels like a better middle ground to me except it raises the burden on end users who now have to manage keys somehow. I should write the idea up in more detail though.

@fluffy @xldrkp

I'd like to read a post like that. To your point that ActivityPub doesn't scale efficiently, I still agree with you but I'd rather have an inefficient broadly-adopted protocol than an efficient one that nobody uses.

Raising the burden on end users is never a good idea for projects that are only as valuable as the number of people using them. It's why GPG remains limited to a small niche of tech-savvy programmers, while Signal/WhatsApp/etc. have brought encrypted messaging to the masses

@jdormit @xldrkp I don’t want to see a situation where the protocol is so difficult to implement correctly that we end up with a monoculture of a handful of implementations and UXes though. And especially the high hosting requirements for both the publishing and subscribing side of things.

There’s also a lot of inherent fragility in what’s emerging and that worries me a lot.

@fluffy @xldrkp

The fragility could be solved with a suite of well-implemented libraries, though. Then implementations could just use a proven library in the language of their choice and not have to worry about messing up the spec.

@jdormit @xldrkp The fragility is more than an implementation detail though. ActivityPub directly ties your identity to the URL from which your client hails, it requires you to have a presence on a specific instance to exist on the network, migrations are painful, etc.etc.etc.

Also the data fragility is a problem; migrating content between implementations becomes quite chaotic in ways that don't exist for a website with a sideband feed.

Sign in to participate in the conversation
Queer Party!

A silly instance of Mastodon for queer folk and non-queer folk alike. Let's be friends!