I'm still fairly optimistic about ActivityPub since a lot of the problems there are in theory solvable, but we'll see.
* it has by far the highest-quality conversations I've ever seen
* no right-wing freaks disparaging trans people in the replies to everything
ActivityPub does have a migration story -- people move servers all the time. But it's not as good as it could be, and there are rough edges.
Correct me if I'm wrong (and in this case I would love to be), but you can migrate your account to a new server, but you have to create a new identity (ie profile URI) on that server. This would be fairly easily solved by servers allowing users to bring their own domain names. In the age of Let's Encrypt it's much easier to manage certs for many domains. They've known about this for 6+ years[0]. The fact it's not a high priority is concerning.
Fortunately this isn't an inherent limitation of ActivityPub, and I'm hopeful we'll see implementations with better custom domain support in the future.
Oh thank god I'm not crazy and not the only one. I genuinely don't understand what either broke these people or what propaganda/troll bot network is responsible but the replies randomly ranting about trans people in threads that neither involve an actual trans person in the conversation nor are about anything even remotely related to lgbt issues is baffling.
I mean, they most certainly are on the Fediverse/Mastodon, but they are also almost certainly not federated with your instance.
BlueSky’s protocol depends on large scale indexers or relays which index the data from the PDSs and transform the nominal decentral streams into a firehose. That firehose then get’s labeled and filtered – and those services again must be somewhat Twitter-scale just to survive that firehose.
Now there seem to be alternative relays (and alternative labelers/filterers), but my concern is less with the man in the middle but the needed scale for the man-in-the-middle.
I think because of this need for scale inherent in the protocol there will only be very few of them, only by very well capitalised companies. If Twitter is a monopol, an AT protocol network will be an oligopoly, possibly a cartel.
I’m astonished that so many technologists who grew up in the more indie 90s and 2000s, in the time of the blogosphere, of individual servers, loosely joined via protocols, don’t seem to see that – or don’t seem to care.
Example new features/improvements:
- Community Notes (huge feature).
- Grok AI.
- Long form messages.
- Open Sourced ranking algorithms.
- Monetization for creators.
- Audio/video calls.
- Job listings/search portal.
- Passkey support (at least for iOS)
- Upcoming features like money transmissions/payments.
- Tweets from a lot of the people I really cared about dried up as people left
- Discourse in replies became noticeably worse as low-quality replies were brought to the top because the poster had paid for “verification”, when most of these posts would probably have been (and rightly so) under the ‘more replies’ section before due to how low-quality they were.
Twitter’s Community Notes may sound nice in theory, but with the current botched implementation it’s just another vector to spread misinformation, now labeled as “context” and made harder to disregard.
Of course, Birdwatch existed since before Musk, so I don’t blame him for inventing it, only for pushing ahead with it.
What Musk could’ve done is apply the same algorithm (that Twitter now uses to choose which community note to show) to sorting replies. Of course, that would defeat the point of blueticks paying 8 bucks to get on top of replies, and Musk wouldn’t be able to keep the lights on now that the big advertisers left.
Just yesterday I get a notification of a tweet I "liked" getting community-noted. I checked it, and found the original tweet was inaccurate.
I've seen some notes I don't agree with, and I simply vote accordingly.
Twitter never had such a widespread feature before where users had the power to fact-check others, even Elon Musk and Joe Biden themselves.
https://twitter.com/MOSSADil/status/1745921315498811752
The original note (see screenshot at https://www.taiwannews.com.tw/en/news/5080048) was up for a while as trending tweet gathered 1-2 million views by users worldwide who saw “context” and were trained to assume it’s the truth (it’s context, after all; it’s the meta above the facts).
Since the algorithm for which note to show is more or less a popularity contest, whoever has the most bots wins, and it took a while until people managed to correct the note. But of course the way Twitter notes work, there is no evidence of this controversy and attempted misinformation, or how many people exactly saw the false “context”.
How many notes like that can you honestly estimate you have seen and just automatically believed? For me, the answer is “no idea”. (Note that if the note agreed with your preexisting opinion it doesn’t make it true.)
How many more millions of viewers were lied to re: COVID vaccine safety when the doctors and scientists were outright silenced and replaced with one-sided propaganda?
And so many other topics.
At least now people can continually fact check and expose the bias to a greater degree.
People could continually fact check tweets without false notes masquerading as “context”, that is why I said Twitter could just use the same algorithm for sorting replies in the first place.
Community notes are not more trustworthy than tweets, but they are painted to be. This makes them a valuable target for misinformation campaigns.
They're just notes, usually with links to contrary/correcting evidence.
I can vote to agree after reviewing the evidence, or disagree and specify why.
You know there's AI and algorithms that moderate these things? You can't just go crazy and doing some community note disinformation campaign all over the place.
Fact-checking companies don't even allow for disagreement.
Galactic brain: use your “AI” to prioritize tweets with evidence.
> You can't just go crazy and doing some community note disinformation campaign all over the place
You know, I literally just showed how it’s done.
I've never seen Grok AI being used for anything on Twitter and long-form messages is a downgrade of the entire platform, not an upgrade.
Open-sourcing the algorithms has also not really done anything to actually improve the platform. What he should open-source instead is the way they pay out to creators, since Mr. Beast seems to be earning way more for impressions than anyone else on the platform. His videos also seem to be served as non-disclosed ads on the platform (https://twitter.com/SHL0MS/status/1748337607694574045), which also seems like a thing that would be better laid out in the open.
Audio calls were also a feature before Elon took over since Twitter Spaces has existed for years.
And here's waiting for money transmissions and payments. Elon says that it'll be out by 2024, much like your Tesla could operate independently as a robo-taxi by 2020.
Launched for testing in June 2022 4 months before Musk was forced to honour his offer to buy Twitter. I don't think you can claim it as a "new feature" under his leadership.
The community I've found on Mastodon has been incredible—I'm reminded of early Usenet days.
If we could choose Twitter instances everyone who cared would have simply switched to the one without Elon and there would have been no problem.
I like having a place to post little things. I like keeping up with other people posting little things.
It isn't Twitter, but we don't have Twitter anymore, so. Just kinda giving things a go. I don't think we really can have "twitter" anymore, those days are past.
But yeah, that's just BlueSky. AT isn't geared towards a specific thing, like it is not a social networking protocol. It's kind of like if RSS was Git, if I squint.
What do you mean it's totally collapsed? I use it daily.
There's tons of people on Twitter who don't like Elon, so that part of your statement is false.
If you wouldn't mind reviewing https://news.ycombinator.com/newsguidelines.html and taking the intended spirit of the site more to heart, we'd be grateful.
Doesn't mean it collapsed, but I get the point of your language now.
X just rated #1 news iOS app in 163 countries.
That's my argument, #1 news iOS app means nothing