Twtxt Is a Self-Hosted, Twitter-Like Decentralised MicroBlogging Platform

by harporoeder • 6 years ago

71 27 6 years ago

bachmeier• 6 years ago

"No ads, no tracking, your content, your data!"

I'm not sure that's a compelling reason to move from Twitter. A better argument would be a business/development team/research team/college class/student organization wanting a Twitter just for themselves.

app4soft• 6 years ago

No ads, no tracking, your content, your data!

... and no @elonmusk.

chrisfinazzo• 6 years ago

But isn't that the crux of the argument for most decentralized "indieweb" and microblogging things?

Unlike Twitter, the network is every single site that implements those basic principles - plus, everyone can participate, not just the users on that one site.

codesections2• 6 years ago

I'd be interested to hear how this compares with Mastodon[0] -- based on a quick look, they seem very similar.

[0] https://joinmastodon.org

Kye• 6 years ago

There's a closed but positive discussion on the project's github.

https://github.com/jointwt/twtxt/issues/86

And IndieWeb: https://github.com/jointwt/twtxt/issues/91

codesections2• 6 years ago

Key quote from the linked discussion:

Mastodon and twtxt are very similar except:
* twtxt has the notion of user owned/managed feeds (think of this as topics or personas you twt as)
* twtxt doesn't and cannot chain replies together, so there are no threads.
* twtxt polls, rather than streams for updates to both local and external feeds

mntmn• 6 years ago

btw the original twtxt is https://github.com/buckket/twtxt

rendall• 6 years ago

This confuses me. James Mills / prologic is not the author of twtxt? The prologic repo makes it seem as if he is

buckket• 6 years ago

Depends on what we’re talking about exactly. I’ve created twtxt (the format specification as well as the reference client) a few years back. Many other clients and compatible tools followed provided by other contributors. Sadly prologic chose to name his piece of software just "twtxt" as well, despite it deviating quite dramatically from the original idea.

prologic• 6 years ago

Is there an issue that I call the backend (codename) the smae thing? @buckket I _assume_ yu're okay with this, although we haven't spoken much...

monkin2• 6 years ago

It should be “No ads, no tracking, no content, only your data!”.

Key value of Twitter is community and huge user base, I don’t see any point of using any copycat if there’s no one to follow. Same thing is with mastodon, sure there are few communities of devs and other wizards, but most of them are furries and other edgy and crunchy stuff for niche communities. And, that’s fine, at least for them, but most of the world just want to follow their loved artists or other creators.

cxr4• 6 years ago

Many of the comments here are making this point and don't seem to understand twtxt. It's possible that the repo's README doesn't present what twtxt is as clearly as it could.

The value proposition of twtxt is that it's stupid easy to "add" to your server while being good enough to be used for 80% of what you'd use Twitter or Mastodon for. You can trivially host your twtxt "account" on pretty much any server, including the one where your current website already lives, even if that server only supports static sites. This is because twtxt is more like RSS than like Twitter, albeit intended to be used in such a way that it's only appropriate for the sort of dispatches you broadcast through Twitter, rather than the totality of what RSS can be/is used for. The format also happens to score better on the human readability scale, too.

Here's an example twtxt.txt feed: http://akkartik.name/twtxt.txt. That's it; if that were your feed, then that would be all your account data right there.

(NB: Just so this comment isn't taken the wrong way—imagined subtext and whatnot—this comment has been written as a twtxt outsider. I'm not doing twtxt now, and I don't plan to.)

PS: For anyone who is in the twtxt target audience, you should consider enabling CORS on your site, if possible. https://enable-cors.org/

monkin2• 6 years ago

The value proposition of twtxt is that it's stupid easy to "add" to your server while being good enough to be used for 80% of what you'd use Twitter or Mastodon for.

Maybe for HN users, but not for most of the everyday internet users. Something that people here tend to forget. :)

vidar• 6 years ago

What is the database behind this?

tyingq• 6 years ago

Appears to be bitcask. https://github.com/jointwt/twtxt/blob/ee2fdcc381710d62ae03a6...

inetsee• 6 years ago

It appears to be a version of bitcask written in Go by the author of Twtxt. There seem to be several other versions of bitcask, including one written in Erlang (the original?), one in OCaml, one written in Rust, and maybe others that don't show up on the first page of a quick search.

foxhop• 6 years ago

I've put a lot of thought into the problem with current social media and the evils of the attention economy.

The feed view is the problem, I would much prefer pintrest style view where each person I follow will have a card with there latest txt update. The cards would be in the order of newest to oldest. Follow 100 people, have 100 cards on a single dashboard.

Cards will not be lost in the ocean of unordered timelines/feeds.

Clicking a username will show that account's timeline of all posts (paginated if needed) with the ability to sort by ascending or decending, and keyword search a timeline (simple full-text-search).

The workflow: as a user I check the main homepage dashboard with pintrest style cards, the newest posts front and center at top.

Cards could be clicked to view responses and usernames could be clicked to view an individual's ordered timeline.

At a glance the reader could quickly determine if anything new is posted.

If you want to hear more about my idea let me contact me (details in my profile).

cxr4• 6 years ago

I think you're halfway there. The problem is not the feed view—that's a matter the client can handle. The problem is the feed itself. It's the wrong data structure.

I wrote about this approx 2 years ago: https://www.colbyrussell.com/2019/02/15/what-happened-in-jan...

Around the time that the phrase "social media" started getting bandied about, the Internet went from topic-focused indexing to person-focused indexing. Makes sense, because topic indexing mirrors the real world. Anywhere you're physically present can be considered a small topic-indexed island. When you're at the grocery store or the post office, the collection of people in that space is an (ephemeral) aggregation over a topic/interest. So is a house party.

When you're at a house party, there is no person-based index that anyone who's there can consult to cross reference the other house parties that the host or another guest has been to. (And there's certainly no index that people who are not even at the party can consult.) At best, you're limited to a view that consists of the house parties where you've seen that person in the past or heard that they attended.

Like I said, topic-based rather than person-focused indexing is the way the Internet used to work. The only aggregation of all my posts across all mailing lists is the one in my mail client, on my machine. It's one not available to the general public. This is a good thing, because person-based indexes are an invasion privacy. We'll probably eventually find proof one day that they cause cancer (of the conversation).

I think people building out services are stuck in a metaphorical rut that retards their vison. As projects crop up to try to address the "surveillance" and "tracking" that the media reports Facebook is doing, these would-be pioneers keep reaching for the same wrong data structure that ultimately sabotages the intent. This happens because they look at their forebears that they're trying to displace, but Facebook, Twitter, etc. have poisoned the well of thought, so people end up recreating alternatives built on the same flaws. I look at this and see it in the same way that I see people who have only started programming in the GitHub era—and a bunch of people who predate it but have had their vision clouded by what they've been keeping in their immediate environment most recently. They cannot conceive of a way to host code repositories or any development model that doesn't work the way GitHub does. (Which happens to record activity in the same type of data structure—although it does so in a way that it's presented to anyone who asks for it, with even fewer privacy controls than Facebook or Twitter.)

zpeti• 6 years ago

You are ignoring the why, though. Person centered feeds simply make us spend more time on them. They are more engaging.

And unfortunately until people start being a lot more conscious about whether this is a good thing, companies with person centered feeds will win in the market. Twitter was going bankrupt before it introduced their recommended feed, now its not.

This will be like smoking. It's great, and cool, and also addictive. People will have to make a decision to stop doing what their emotions tell them to do. I think eventually it will happen, but smoking is still around, most people have smoked a bit in their lives. You just stop eventually because you know its bad for you.

seniorivn• 6 years ago

look at telegram channels, might be close enough to what you want

dsparkman• 6 years ago

Is it just me or are all of these "new" micro blogging platforms just trying to recreate RSS feeds. You take a few pieces of relatively simple tech and wire them up and you have solved the "decentralized social media" problem. RSS, Webmention, Webfinger, and WebSub (formerly PubSubHubbub).

Create a better "RSS reader" that is more Twitter timeline interface and less email "folders" and you have it.

Also for the love-of-god can we dump the hokey @username@hostname mentions and just use username@hostname like the internet at large has used for decades. username@hostname is NOT only for email addresses.

remram• 6 years ago

I for one do want folders in my microblogging timeline. Mastodon mostly has that, Twitter lists don't really fill the gap though.

stanislavb• 6 years ago

I’m working on a new RSS reader. Hopefully I’d be able to show something in the beginning of next year.

hopesthoughts• 6 years ago

Nope, I can't stand RSS readers that have a Twitter timeline interface!! I'd rather have the ability to display one single feed, than all feeds at once.

Twtxt Is a Self-Hosted, Twitter-Like Decentralised MicroBlogging Platform

Related Stories