Birb + Fossil: An RSS Revival?

A couple of days in the past, @twilliability introduced Birb, a Mastodon bot the place you possibly can ship it
a URL of any RSS feed, Atom feed, podcast, Substack, and so on. and it’ll create a Mastodon account for it that you would be able to observe.
This successfully meshes social media and the blogosphere. That is nice! However Mastodon
has been infamous for sticking with chronologically-ordered timelines, so except you may have time to have a look at each single publish, you’ll
probably miss one thing.
Enter fossil. I announced it earlier than New Years. It’s a Mastodon consumer I made that permits experimenting
with timeline algorithms. In contrast to a full Mastodon server, it don’t deal with any form of firehose of posts, it merely reformats
my residence timeline in a means that helps me discover the attention-grabbing stuff and ignore every little thing else. Proper now, it teams posts
collectively based mostly on similarity and generates a label.
I’ve lots of concepts for how you can format a timeline, however frankly, I’m unsure they’re good concepts. It’s arduous to know
with out making an attempt them out. Within the final week, I’ve begun pivoting fossil to be extra extensible, through plugins to be able to
construct your personal timeline algorithm or customise the view, with out having to clone my repo or ship pull requests. Hacking
is nice! We should always make hacking even simpler!
So between Birb & Fossil, it looks like we’re seeing an RSS revival.
I put an RSS feed on my weblog again when RSS was the new factor. You may see it right here,
https://timkellogg.me/blog/atom.xml
Alright, advantageous, it’s truly Atom, however most individuals use “RSS” and “Atom” interchangeably since they each work the identical.
It’s an XML doc that accommodates an array of entries, one per weblog publish. Every entry has a title, hyperlink, date, ID, and a
brief paragraph that summarizes it (or the complete publish, in my case). An RSS consumer periodically downloads the XML doc
and makes use of the ID area to determined if a brand new publish has been printed.
RSS is simple to parse, which makes it nice for constructing instruments, integrating with different techniques, or constructing interest
initiatives. (I’ve even see folks use it for propagating server configurations, uncertain how that went, however nonetheless it’s a cool concept)
Again within the ’00s you’ll obtain a feed reader and subscribe to feeds. This felt lots like an early model of social
media. Google Reader was killed in 2013, which was largely seen because the demise of RSS. I feel social media
typically changed RSS as a result of it took far fewer technical abilities to setup a Fb account versus an RSS-enabled weblog.
I consider we’re seeing a rebirth of RSS, and it’s pushed by a couple of surprising tendencies.
Development 1: Loss of life of Twitter
I get it, Twitter could be very a lot alive, but it surely’s clearly not the identical anymore. I left Twitter after Elon took over, and each
time I’m going again to go to it appears ever extra overseas to me. I attempt to login each few months to maintain my account energetic, however actually,
I could overlook as a result of the positioning has retained so little of the character that drew me there within the first place.
Development 2: Rise of the Fediverse
I get it, by the numbers it’s nothing in comparison with Instagram, TikTok and even Twitter/X. However comparatively, it’s progress has
exploded over the past 12 months. Extra necessary, it actually feels just like the open Web that social media all the time ought to have been.
When Meta lastly finishes federating Threads with the remainder of the fediverse, it means you’ll be capable to observe
and work together with Threads accounts & posts from Mastodon and vice versa.
Development 3: Rise of Syndication
Podcasts run on RSS. Notifications of recent episodes are dealt with via an open web normal, RSS. Newer websites
have been enabling RSS. Some examples:
There’s clearly content material being uncovered through RSS, however lots of the feed readers died or nonetheless really feel like they have been born
within the ’00s.
Development 4: Plummeting Complexity of NLP
With the rise of ChatGPT, the world has grow to be acutely conscious in regards to the potential of AI. Successfully, any dummy can
now throw collectively some utility that “understands” textual content and reply in an intelligent-sounding means.
Skeptical of AI?
Consider the 1000’s of idiotic “AI powered” concepts folks have provide you with in the previous couple of months. A couple of years in the past
none of that will have been even remotely attainable exterior massive tech firms like Fb, Google or Netflix. The
incontrovertible fact that dumb concepts can flourish is proof that the complexity has clearly plummeted.
Nonetheless, embeddings are where it’s at. In contrast to full LLMs, their output could be very cacheable, aggregatable,
and you may simply do math on them in ways in which we’re nonetheless understanding:
- Clustering (e.g. “group these posts by comparable content material”)
- Classification (e.g. “is that this publish about kittens or puppies?”)
- Search (e.g. “discover all posts about kittens working into issues”)
- Similarity (e.g. “is that this publish much like that one? how comparable?”)
I exploit embeddings for clustering (and shortly for classification) inside fossil. It’s really easy.
Between the frequent availability of LLMs and embedding fashions, a classy pure language processing (NLP) mission
takes only some minutes to undertake, the place a couple of years in the past it probably wasn’t even attainable for a hobbyist.
It’s arduous to make predictions, but it surely certain looks like a serious theme of 2024 goes to be open requirements and open supply.
From the provision of supply knowledge to the delicate instruments to work with the info, we’ve bought a ton of potentialities
in entrance of us. I’m actually excited in regards to the instruments we’ll see constructed this 12 months.
If you wish to take part extra within the syndiverse, try these items:
- atom.xml — I exploit Github Pages to host this
web site. This Jekyll template is how I’m producing an Atom feed for the weblog portion. It’s actually very straightforward, principally
cut-n-paste. - Fossil — My Mastodon consumer. I’d like to see folks use it, however I’m particularly excited to see what folks make
out of it. Ship pull requests, create points. Even should you write your personal competing software, inform me about it, I’d gladly
promote it. - Birb — Go fedify an RSS feed and observe it! Create a mastodon account (or threads!). Take part within the syndiverse.