I recognize lots of folks here don't trust Bluesky, but I actually find a lot of what they're doing really, really interesting and worth watching, including their experiment with "composable moderation." https://www.techdirt.com/2023/04/20/bluesky-plans-decentralized-composable-moderation/
Bluesky Plans Decentralized Composable Moderation

We just wrote about Substack’s issue with content moderation and the Nazi bar problem. As I highlighted in that piece, any centralized service is going to be defined by their moderation choices. If…

Techdirt
@mmasnick I’d like to try it. do you have an invite code to share?
@mmasnick
It's pretty interesting. @oliphant and some lovely people curate block lists for instances. That is similar to the idea of community labeling, except I don't know if anything like that exists at the user level for Mastodon. I may have just missed the user version of the same.

@mmasnick I always pictured moderation as more of the grey judgement calls of “was this really a threat or hyperbole” “is this actually pornography or a public health post” kinda thing

is moderation vs crowedsourced hashtags snd an algo designed to boost/deboost/hide going to avoid people wrongly tagging to attack folks (mass tags instead of mass reports)

Def interesting

@mmasnick I'm intrigued also. Signed up a few months ago. Still waiting for the invite code though 🙄
@mmasnick can't wait for the influence wars of what's tagged as a political hate group
@miunau i think the whole point is that under this system it's not a centralized determination. If you disagree with how one moderation system is tagging things... switch to another.
@mmasnick Yeah. Turning moderation into a marketplace is certainly an interesting phrase with its connotations. The shape of moderation is influenced by public discourse, social media personalities with a large following, etc. Some of which have less than stellar takes on people that don't look like them and many lack in the department of empathy and patience required to make moderation decisions, so I have some doubts
@mmasnick To give an example, from what I've seen in my network on here, IRC, Matrix etc. communities work fine with blocklists that simply block chuds that want to harm them based on their identity. How many of those lists would be necessary or even possible to construct? Maybe a handful at most is my guess
@mmasnick @miunau isn’t that functionally the same as any Mastodon server, then? If you don’t like the moderation on one server, you move.
@Brendanjones @miunau yes, but with bluesky you don't need to move. You can. Or you can just change which moderation system you subscribe to. Or adjust the settings on the one you subscribe to. It's much more putting power in your hands.
@mmasnick @miunau what I can’t wrap wrap my head around is how the server will ever be independent of moderation. If you host a server then you’re at some level responsible for what’s hosted there. So server admins still have to do moderation … in which case moderation is still not independent of server. And relying on others for moderation tags kinda sounds like advanced filters, not moderation per se. I think I’m not understanding something.
@mmasnick I’d like to understand the business model. The wrong business model creates the wrong incentives.
Will Elon Musk keep funding Twitter’s most interesting side project?

Bluesky, the decentralized social networking project currently funded by Twitter, faces an uncertain future after Elon Musk’s acquisition.

The Verge

@labelsoren it's wholly separate and independent from Twitter, and as I understand it has no interest or desire to continue getting any funding from Musk or Twitter.

These days, it's literally a competitor.

@mmasnick It’ll be genuinely interesting to see how this plays out across jurisdictions—nationally, but also maybe at the state level, if things in the US carry on as they are. Honestly, I can see the appeal of a system like this for a legal team.
@[email protected] Good article. I broadly agree though I think it will remain to be seen exactly how the base moderation decisions will affect the perception of BlueSky at the start. Particularly for the hate groups labeling, I expect a lot of angry folks from one circle or another.
@mmasnick I'd like to try it out, sling an invite code my way :)
@jbaert i'm all out of invites right now... sorry.
@mmasnick
l kept on reading compostable.
@mmasnick The way they break up Speech and Reach really resonated with me (image from the AT protocol site)

@jmalone @mmasnick Can you say more about what you find appealing here?

One key presumption for a marketplace of label/throttle models of moderation is that posts are broadly available (to labelers) but narrowly distributed (to individuals).

Over time, doesn't that foster yet another form of surveillance capitalism (with all its attendant problems)?

@jmalone @mmasnick How this plays out may depend on architectural decisions about when/how:

- posts are exposed to labelers
- labels are exposed to distribution algos

That may be a JIT decision (hey this post just arrived at my instance or client, what's inside?).

Alternatively, posts could arrive pre-labeled (in some trusted/attributable way) because it's only economical to ask labeling services to make that decision once, upstream.

@jmalone @mmasnick I assume the #BlueSky devs will make pragmatic choices so the protocol can function (at least initially).

But zooming out, wouldn't there be strong incentives to scale up + control chokepoints at/near the boundary where distribution algorithms consume labels? Hmm...

@pevohr @mmasnick For me the appeal is separation of concerns between services, so each can focus on what they do best.

Mastodon today both hosts your content and is responsible for the reach of your content, which is convenient. The down side is then the server you join is responsible for hosting your identity, your content, moderation and composing your feed.

I like the idea of offloading some of that work to other open source, opinionated algorithms, for me to consume content through.

@pevohr @jmalone @mmasnick Same thing with their non-chronological content recommendation algorithms as far as I can tell. Including no ability (that I have heard of) to say "I don't want that one particular nasty algorithm to see/recommend my stuff."
@mmasnick the crackhead downtown is interesting, too, but I’m not giving him my credentials
@mmasnick People should create at least some content there, so they get a sense of how it works, first hand, so they can make sense of the inevitable *takes*.
@mmasnick Mike, this is a great write up. These ideas (composable moderation combined with a marketplace for algorithms) can and should be implemented via #ActivityPub. There's no technical reason why this can't happen via the developer community here in the Fediverse.
@mike @mmasnick I've beat this drum a lot. Everything that Bluesky is proposing can be done right now in Mastodon via the client. Open API and open source allow you to theoretically have a client that could offer you algorithmic curation/moderation of your feeds. You could even have a market place of interchangeable algorithms between clients. Open beats black box.
@mike @mike yup. I hope that people take up the opportunity to do the same here, but every time I've suggested similar things here in the past, people get... shouty... about how algorithms are bad. I love it here, but some of the rigid norms get frustrating.
@mmasnick @mike That's the Buety of doing it client side. Don't want algorithms use a different app. Mastodon is the engine but all the good stuff can happen in the app.
@mmasnick Unlike others in this thread I did not read "composting moderation". But I *did* read it with Monty Python's "Decomposing composers" playing in the back of my head.
@mmasnick Have you got an invite code for me? I'd like to check it out.
@admin currently all out, sorry.
@mmasnick whoever gets moderation right will create a monopoly (in a good way).
@mmasnick This felt off as I was reading about it and I wasn’t sure why. It actually comes across as a terrible idea. It sounds like they’re actually washing their hands of moderation, it’s up to you to put on blinkers and choose a provider to label stuff you don’t want. The terrible stuff is still there, but you can’t see them in your replies and mentions. BUT EVERYONE ELSE CAN. It’s an invisibility cloak for trolls.
@mmasnick Nazis are welcome at this bar, and if you don’t like Nazis you are kindly requested to wear special glasses so you won’t see them as they encircle you and your friends.
I don’t think Jack gets it. At all.
@MetalSamurai @mmasnick that's literally the ideal tho, you choose what you want to put up with but don't make the decision for anyone else.
@palecur @mmasnick Nazis in the bar with you, but behind one way mirrors?
Free speech is great, but we should be intolerant of intolerance.
This is Jack trying to run Schrödinger’s Nazi bar.

@mmasnick I’m a big fan of what’s happening at #bluesky

It’s effectively the web3 alternative to #activitypub . It’s a good North Star for the fediverse to aspirationally build towards.

But I’m only interested in BlueSky insofar as it interoperates with the existing #fediverse .

And I think they recognize that the the fediverse movement & community would turn on them if they cannibalize the existing network instead of adding to it as a friendly & compatible node.

@erlend @mmasnick It's a web3 platform? If so I've immediately lost any and all interest.
@cdp1337 @erlend it's not. i mean, web3's definition is sorta fuzzy, but if you're asking if it's based on cryptocurrency, the answer is "no."

@mmasnick @erlend Ah, ok. It seems that everytime I see 'web3' it's some half-baked crypto/blockchain scam being pushed by a company trying to exploit the community to make a quick buck.

So when are we getting the source code to this bluesky thing?

Comparing main...bnewbold/publish-lex-script · bluesky-social/atproto

Social networking technology created by Bluesky. Contribute to bluesky-social/atproto development by creating an account on GitHub.

GitHub
@mmasnick @cdp1337 the web3 aspects of it are things like UCAN identity/login, which is more focused on the p2p and offline-first aspects of web3.

@mmasnick @erlend

Awesome, thanks for posting! I'll totally check it out and see what they're up to. (I was completely expecting a response of the typical private trademarked intellectual property corporate BS btw. Legit awesome that they're contributing _anything_ under MIT.)

I'll make my leave by contributing this _totally unrelated_ link in here... ;)

https://xkcd.com/927/

Standards

xkcd
@cdp1337 @erlend i recognize there's a lot of distrust given its founding. But the people behind it are legit good people, and super smart and thoughtful.
@liztai if you want a look I can give you a code. It really is an early Twitter clone at this point, but you can test out the dns based verification for example and see some of the bots people have built on top of atproto. It is just the beginning

@mmasnick honestly, I remember looking at bluesky when it first launched.

to me it always read as a kind of absurdly convoluted system that wants to be decentralized while also just recentralizing everything that’s deemed overly computationally expensive (like a search engine), while slapping an authentication system on top of it that isn’t required to be crypto nonsense, it really does want to be crypto nonsense.

basically, freedom within a walled garden of centralized components (which of course can only be operated by a sufficiently scaled tech company).

the moderation thing seems interesting though.

wonder if you could mimic something similar with the AP protocol - make a sort of external, independent, tagging server where users can bind status links to notes/warnings. it should in theory be possible to then build that into fedi software (ideally with allowing admins to add any number of preferred “note sources” so that an admin can choose which source they trust) on the display side.

broadcasting tags as just AP statuses is assuming there’s no malicious actors in the system, which is a very bad/dumb assumption to make, so it shouldn’t be made. best to keep that as a sort of external system.

@mmasnick That sounds super interesting. Should be slapped on Mastodon.
@mmasnick I look forward to exploring it when it becomes more widely available.

@mmasnick The below is not true!

Platforming hate speech under a "freedom of speech, not freedom of reach" banner is what's turned Twitter into a cesspit of hatred under Musk and *not* platforming it is what makes Mastodon so much nicer and safer.

Hate speech is not free speech, and censoring it *does* work—the moderation policy on this server and the continuous efforts of moderators is exactly why you hardly ever see it here.

"You'll be able to find it anyway" is not a reason to platform it.

@hughster i think that's misrepresenting what i'm arguing.
@mmasnick OK; I do apologise if that's the case. My understanding from what you wrote is that Bluesky's moderation principle will be not to remove "political hate groups" but to allow users to hide them instead; is that not how it'll be?