# Understanding AP and RSS
I've been trying to figure out what ActivityPub does that RSS doesn't.
Off the top of my head, it's not the ability to syndicate, RSS already does that. I can follow anyone on any server.
I think it's the timeline? And the ability to delete posts. Keeping all that in sync is a lot of work, and presumably a lot of traffic?
Also replies. If I reply to a post when viewing it on another server, the reply should show up under any other view of that post.
What else??
PS: Only thinking about features that are used by Mastodon to federate, not potential future features. If that were the rule, then RSS could do anything AP could do, right -- because you can always add the feature. Just the features that are in use.
@davew They could've implemented the outbox as an RSS feed. Outbox has accessible URL like
https://mastodon.social/users/davew/outbox?page=true
Instead it has own new format.
@davew I could be mistaken, but I don't think you see *all* replies on another server - just ones your instance is aware of (e.g. because someone on yours follows the replyer?)
@rscottjones -- interesting. it feels like that to me too.
so why not just open a discussion group of some kind, using discourse for example.
or is that just another approach to what AP and Mastodon are doing??
@davew One advantage that RSS has is it doesn't require a special server, it works off regular HTTP servers which everyone has. This makes me wonder if there's a useful subset of ActivityPub that could be served via HTTP.
@davew I think RSS + Webmention (for sending replies) gets you 90% of the way there. ActivityPub does provide a comprehensive framework for the rest, though, and perhaps follows modern social network conventions more closely, e.g. liking posts, approving follows.
Thanks for the confirmation.
You're the only person I know who's gone really deep into both technologies.
@davew a few things that come to mind (some artifacts of the user experiences)
1) RSS is how I follow a site/blog (which may not map to an individual - some do, many don’t, and the posts feel less personal (they may not always be signed by an individual). In contrast most accounts I follow on the fediverse (via ActivityPub) are individuals and speak with a personal voice (some aren’t but they are the exception).
Apps like Ivory or Mastadon’s web reinforce this with profile photos and names
@davew 2)comments on blog posts feel different than replies on the fediverse (and don’t usually show up in the same rss feed). Posts linking to other posts likewise show up on my site perhaps a link/mention on another site but I haven’t seen them mix in a main feed from a site. As a reader following an rss feed feels more like subscribing (and I hate non-full text feeds that force me to click in to see the whole post) whereas the fediverse feels more two way and conversational with boosts also
@davew 3) boosts feel different than links via rss. Quoted posts (which is eh partially here on the fediverse but not fully) likewise feel different from what I usually see via raw feeds I’ve used.
Related edited posts on the fediverse feel functionally different than via rss - I appreciate the ability to edit a post (and the value of being able to see the version history if I want)
Again some of these are based on how the experience typically feels I’m sure some rss feeds could feel closer
@davew At one point I had an RSS reader/writer that looked like Twitter. It was just RSS Cloud. Only thing needed was an easier follow button, which at the time I thought should be a standard path like `/feed` but now if we just incorporated Webfinger in to the mix the clients could just look up profile and feed information
this package usually finds the feed.
https://github.com/scripting/reallysimple/tree/main/demos/feedhunter
@davew I think for clients to have a simple one click follow button the feed info needs to be embedded in the item or the profile, or we need identity as you've pointed out
@davew
The big one is push. My AP server sends a notification to subscribers when new content is published.
RSS requires a pull.
Update notifications are also useful.
HTTP Signatures have some uses, but I'm not yet convinced I understand them.
Some more semantics. AP technically supports different verbs in a well defined manner. But very few sites use them.
Also, new and shiny JSON rather than old and boring XML 😁
@davew People mentioned social interactions, which are core in AP. I've seen comment submission facilites grafted onto RSS but it didn't seem like that ever caught on much.
Then there's the push-based model, which allows AP to be fast. When a friend posts on social media, I want to see it in seconds, not after 15 minutes or whatever RSS pull interval.
But yeah, "kinda like RSS" is a useful shorthand for explaining AP, similar to "kinda like email" but with a different venn overlap.
rssCloud which is part of rss 2.0, gives instant updates. Not even a second between update and it appearing on readers screen.
@davew Oh neat! I hadn't seen that. 🙂 Yeah, at that point you have something very close to ActivityPub.
From a first glance, one difference seems to be that with AP your follower list is based on individual accounts and not servers/aggregators. Multiple accounts can have shared inboxes to reduce redundant traffic, but they're not mandatory. So arguably AP gives you more knowledge of who your followers are. Now, whether that's on the whole a good thing? 🤷
@davew What I'm getting at is, as I understand rssCloud (like I said, very cursory research just earlier), when someone follows you you get an IP address. Which can be a server somewhere (potentially giving multiple people access to the feed) or the person's home or their VPN or whatever.
With ActivityPub, when someone follows you, you get their account. Which of course doesn't need to correspond to an established identity, but it still seems like a qualitative difference.
a simpler way of saying that is that with AP you know who's following you and in RSS you don't.
which sounds like a negative until you think about it from the user's standpoint, no spam, spyware, etc.
which is why google didn't like it btw.
