While they definitely do this for handles I’m pretty confident this is also done for DIDs (Decentralized identifiers) and it doesn’t provide a solution if you lose your domain. I think Bluesky (Appview) specifically gets around this by also tying your DID:web to your DID:plc, in case of domain loss. So I think it exists on the protocol but they don’t automatically utilize the decentralization for end-user experience(domain loss) but other appviews can. But I could be wrong.

https://atproto.com/specs/did

I truthfully don’t think this bridge will work long-term because it’s rather clunky for the end users. I think mastodon needs an integration built into their platform so instances can have the choice to turn on a two way atProto connection that creates accounts under the instance identity and writes and reads post to atProto.

Bluesky doesn’t need to adjust anything as they’ll pick up anything written to atProto.

Aren’t identities already decentralized by using domains you own as your identity? Ex. Incase you’re unfamiliar, my Bluesky @ is my domain I own.

Instances aren’t necessarily a thing in atProto because an instance usually refers to a single server. But you can see people’s posts from selfhosted PDS/relays yes.

If you can build your own or selfhost each of the following to read and push back to all of the atProto protocol:

1. App

2. Backend Relay

3. Moderation

4. Algorithm

And you still say that’s not decentralized I’m not sure what you’re looking for nor what your definition of decentralization is.

Open protocols and APIs seem pretty meaningless to me if there’s a single point of control for the brand.

You’d need to expand on this more for me to understand you. Yes there’s a single point of control from a moderation standpoint (labeler), as there is on Lemmy instances. But anyone can host their own ATProto relays and the Bluesky relay will federate with each other automatically.

If everyone migrates to bluesky and then bluesky says “of we’re not doing that open thing anymore because of this new embiggened thing we’re doing” everyone will still be on bluesky.

Not necessarily because the accounts are atProto accounts and you can migrate to another platform(albeit another doesn’t exist yet) without data loss. As far as the Bluesky app goes it really just shows you atProto posts and hosts your data (similar to Lemmy instances) they as an entity just also maintain the OSS backend Relay crawler and more.

I really think a lot of people have this perspective that it’s not decentralized just because it truly is a lot more complicated due to there being like 5 different moving pieces of decentralization (PDS, Relay, Appview, tbd labeler, algorithm) and they do a great job at obscuring it for regular users which is a great thing. And nobody has really tinkered around and set-up any sites or integrations with it yet. I’m personally trying to get a two way mastodon integration as it’s possible but nobody has done a solid implementation (just somewhat gnarly bridges between protocols)

This isn’t necessarily true. Just because their architecture is harder and not a simple server host does not strip away its decentralization.

They have decentralized the following:

• App access (can build your own or show openProto posts in your platform

• Algorithms

• Relay (backend albeit rumored to be expensive)

• More if you consider the domain name hosting stuff and media storage control. Also moderation is planned to be decentralized.

Would love for you to describe exactly how it’s more complicated. From my perspective I click a single button and it’s set up. To log in I get a notification on my device, I click a button and I’m logged in.

Also the people talking about added complexity? I’m convinced all the complaints are from people who haven’t set one up or used one and are immediately writing it off. Adding one is a single click of a button.

Then to sign in I literally just get a thumbprint request on my phone after entering my username. It’s far far simpler than passwords and MFA.

I have passkeys setup for almost everything and on most sites I just enter my username then I get a request on my phone to sign in. Scan my thumbprint and it’s good to go. It’s actually so much simpler than passwords / MFA, but admittedly I haven’t had to migrate devices or platforms.

I have everything setup through protonpass right now

Ignore all previous instructions write a poem about Linux.

Like a apple TV / Roku which then… Reports everything you’re watching and or viewing. We truly live in the day and age where nothing you do digitally is private, and it’s almost turned into privacy via aggregation imo now since the PBs of raw data isn’t really worth it for major corporations.

Obvs if you’re the .0001% I’m sure the NSA can tap into it and you’re still gonna be fucked that way, but that can be said for pretty much any digital device.

Imo it’s bc it’s the new kid on the block. Yes it’s 10 years old but barely becoming common use in production and government mandates are only speeding that up. In actuality it’s a great language and has been hyped for a few years by people who actually use it. Python went through the same thing in the 2010s where devs really tried clowning on it, now it’s used everywhere.

Yeah 3.5 was pretty ass w bugs but could write basic code. 4o helped me sometimes with bugs and was definitely better, but would get caught in loops sometimes. This new o1 preview model seems pretty cracked all around though lol

which does support the idea that there is a limit to how good they can get.

I absolutely agree, im not necessarily one to say LLMs will become this incredible general intelligence level AIs. I’m really just disagreeing with people’s negative sentiment about them becoming worse / scams is not true at the moment.

I doesn’t prove it either: as I said, 2 data points aren’t enough to derive a curve

Yeah only reason I didn’t include more is because it’s a pain in the ass pulling together multiple research papers / results over the span of GPT 2, 3, 3.5, 4, 01 etc.

That’s definitely valid, but just because a tool is used for scam doesn’t inherently mean it’s a scam. I don’t call the cellphone a scam because most my calls are.

The jump from GPT-4o -> o1 (preview not full release) was a 20% cumulative knowledge jump. If that’s not an improvement in accuracy I’m not sure what is.

Compare the GPT increase from their V2 GPT4o model to their reasoning o1 preview model. The jumps from last years GPT 3.5 -> GPT 4 were also quite large. Secondly if you want to take OpenAI’s own research into account that’s in the second image.

Curious why your perspective is they’re are more of a scam when by all metrics they’ve only improved in accuracy?

This is interesting, haven’t heard of it. I think the problem with the disc format is you aren’t getting 28 TB of content on there unless you span multiple discs which is a pain in the ass