Jack Dorsey, Twitter's former CEO and founder launched his own social media platform today, called Bluesky. Bluesky is based on the principle of allowing users to build a shared and open social media platform.
Bluesky is based on the principle of allowing users to build a shared and open social media platform.
Bollocks. They are using a centralized “DID Placeholder” thingamajig, claiming they want to “replace it later”:
We introduced DID Placeholder because we weren’t totally satisfied with any of the existing DID methods. We wanted a strongly consistent, highly available, recoverable, and cryptographically secure method with cheap and fast propagation of updates.
We cheekily titled the method “Placeholder”, because we don’t want it to stick around. We’re actively hoping to replace it with something less centralized. We expect a method to emerge that fits the bill within the next few years, likely a permissioned DID consortium.
That currently they are using the DID Placeholder, which seems to be centralized. They can use other DIDs, but they don’t. And once you build a system on something it’s really difficult to replace it.
I’m sure they did not b0rk their implementation and it will not turn out that for some weird reason BlueSky can only work with some very specific DIDs. That would be unheard of, even though it would be clearly in their best financial interest. 🤣
On a more serious note, BlueSky seems to me to be designed in a way that allows its creators to claim decentralization while at the same time have similar secondary centralization characteristics as, say, cryptocurrencies. The separation into “layers” means it doesn’t matter which node you happen to use for storage of your data, the important, user-locking-in stuff happens in a higher layer.
The higher layers provide search and discoverability. It just so happens that in case of these, scale matters. The bigger the provider, the “better” the service. Look at Google Search. Web is open and decentralized, but search got all-but monopolized because of economies of scale and because of the intrinsic way being Big in search is an advantage.
BlueSky is trying to replicate that, it seems to me. “Here, have your decentralized and irrelevant storage layer and be merry, and we’ll just build the biggest search/discoverability provider and own the game anyway”.
Look how different this is from the Fediverse. Fediverse is not layered. It is decentralized from instance to instance, and it seems that scale is not that much of a benefit — or might even be a hinderance. This promotes loads of smaller instances, with no single entity controlling any important facet of the network.
This is completely different from how BlueSky is designed. Again, to me, BlueSky’s (BS, I like that acronym) design is made specifically so that scale is a benefit and the winner takes all — in the search and discoverability layer.
It’s a trojan horse, from people who brought you Twitter.
I will add one more thing to this, to make it maybe a bit more clear: the question is who has agency over what.
In Fediverse, communities large and small can set up their instances and have full agency over moderation decisions, registrations, blocking, defederation. They can build and maintain their garden while allowing people in and out as they choose.
In BlueSky, that kind of agency is gone. The storage/hosting layer is irrelevant — the idea is that you can move your account and content anywhere else with exactly same access to everything else retained, so that’s not where moderation decisions can happen.
The “search and discoverability” layer, on the other hand, is where “providers” reign supreme. And they are incentivised to slurp as much data as possible to “offer a better service”, and are not as closely connected to specific communities or people. The power dynamics are completely different.
This de-coupling of storage/hosting and search/discoverability is billed as a great advantage, but it’s not an advantage for the users — it’s an advantage for the providers.
In fedi there is no single entity that can have full visibility into the whole network. In BlueSky, that kind of visibility for “search and discoverability” providers is the basic assumption the protocol is built on.
And let’s not even start delving into the question of consent here… 🙄
@rysiek first of all I don’t see any problems in Twitter technically. It worked well for long time and only now with Musk as CEO we see that not only architecture matters but people supporting it also.
Regarding BS: just take a look at how centralized fediverse is in reality. Eugene just happened to be always talking about mastodon instead of fediverse, everyone is trying to register on mastodon.social when they herd of it first time. I mean they whole fediverse for people right now is only what a man with a German company did.
Once again it’s not technicality but also people who support it matters. We can take at-protocol and do many integrations by our own to create own high level crystallization. Or we can give up on current AP and move on to the next one some time. The more ideas you give to the world the more it gives back.
If they are to make their own evil corp then you and I are allowed not to be it’s customers. Decentralization is about community and people
@rysiek first of all I don’t see any problems in Twitter technically. It worked well for long time and only now with Musk as CEO we see that not only architecture matters but people supporting it also.
Regarding BS: just take a look at how centralized fediverse is in reality. Eugene just happened to be always talking about mastodon instead of fediverse, everyone is trying to register on mastodon.social when they herd of it first time. I mean they whole fediverse for people right now is only what a man with a German company did.
And yet when somebody forks Mastodon (like GlitchSoc or Hometown), they are not beholden to Eugene. Moving between instances works. Moving between forks works. Yes, Mastodon has an outsized presence in fedi, but it has nowhere near as much control as Google Search has over website traffic, or as the biggest BS’s “search and discoverability layer” provider will have over BlueSky’s users.
Fedi is simply a different architecture on a very basic level, built in a way that is not as supportive of economies of scale and “winner-takes-all” model, as BlueSky is.
Would I want fedi to be more diverse in instance software offerings? Totally. Is it fair to compare this to secondary centralization that BlueSky will support by design? Absolutely not.
Once again it’s not technicality but also people who support it matters. We can take at-protocol and do many integrations by our own to create own high level crystallization. Or we can give up on current AP and move on to the next one some time. The more ideas you give to the world the more it gives back.
Differences in protocol design define what will happen in a network run on it. Differences in design between BlueSky’s protocol and ActivityPub mean that secondary centralization will be way, way easier in BlueSky than it is on fedi — to a point that it seems this is by design.
If they are to make their own evil corp then you and I are allowed not to be it’s customers.
How’s that working for us all regarding Google Search? We’re not even paying Google for search and yet we cannot escape it. I use DuckDuckGo daily, and yet I still am sometimes forced to use Google Search. Websites that want any real traffic need to optimize for Google Search.
“Voting with your feet” is a naïve myth unless the underlying system actually empowers people to do it effectively. Fedi does. BlueSky very definitely does not in the layer that matters.
Decentralization is about community and people
Totally. No wonder, then, that BlueSky’s design carves out a space for secondary centralization exactly where communities and people can be found: search and discoverability.
Bollocks. They are using a centralized “DID Placeholder” thingamajig, claiming they want to “replace it later”:
What’s that?
@rysiek where centralization is stated?.. I don’t really see, only requirements for server to process these dids
@jabberati
That currently they are using the DID Placeholder, which seems to be centralized. They can use other DIDs, but they don’t. And once you build a system on something it’s really difficult to replace it.
@rysiek well DIDs are designed to be easily replaceable so I don’t really see a problem.
I’m sure they did not b0rk their implementation and it will not turn out that for some weird reason BlueSky can only work with some very specific DIDs. That would be unheard of, even though it would be clearly in their best financial interest. 🤣
On a more serious note, BlueSky seems to me to be designed in a way that allows its creators to claim decentralization while at the same time have similar secondary centralization characteristics as, say, cryptocurrencies. The separation into “layers” means it doesn’t matter which node you happen to use for storage of your data, the important, user-locking-in stuff happens in a higher layer.
The higher layers provide search and discoverability. It just so happens that in case of these, scale matters. The bigger the provider, the “better” the service. Look at Google Search. Web is open and decentralized, but search got all-but monopolized because of economies of scale and because of the intrinsic way being Big in search is an advantage.
BlueSky is trying to replicate that, it seems to me. “Here, have your decentralized and irrelevant storage layer and be merry, and we’ll just build the biggest search/discoverability provider and own the game anyway”.
Look how different this is from the Fediverse. Fediverse is not layered. It is decentralized from instance to instance, and it seems that scale is not that much of a benefit — or might even be a hinderance. This promotes loads of smaller instances, with no single entity controlling any important facet of the network.
This is completely different from how BlueSky is designed. Again, to me, BlueSky’s (BS, I like that acronym) design is made specifically so that scale is a benefit and the winner takes all — in the search and discoverability layer.
It’s a trojan horse, from people who brought you Twitter.
I will add one more thing to this, to make it maybe a bit more clear: the question is who has agency over what.
In Fediverse, communities large and small can set up their instances and have full agency over moderation decisions, registrations, blocking, defederation. They can build and maintain their garden while allowing people in and out as they choose.
In BlueSky, that kind of agency is gone. The storage/hosting layer is irrelevant — the idea is that you can move your account and content anywhere else with exactly same access to everything else retained, so that’s not where moderation decisions can happen.
The “search and discoverability” layer, on the other hand, is where “providers” reign supreme. And they are incentivised to slurp as much data as possible to “offer a better service”, and are not as closely connected to specific communities or people. The power dynamics are completely different.
This de-coupling of storage/hosting and search/discoverability is billed as a great advantage, but it’s not an advantage for the users — it’s an advantage for the providers.
In fedi there is no single entity that can have full visibility into the whole network. In BlueSky, that kind of visibility for “search and discoverability” providers is the basic assumption the protocol is built on.
And let’s not even start delving into the question of consent here… 🙄
@rysiek first of all I don’t see any problems in Twitter technically. It worked well for long time and only now with Musk as CEO we see that not only architecture matters but people supporting it also.
Regarding BS: just take a look at how centralized fediverse is in reality. Eugene just happened to be always talking about mastodon instead of fediverse, everyone is trying to register on mastodon.social when they herd of it first time. I mean they whole fediverse for people right now is only what a man with a German company did.
Once again it’s not technicality but also people who support it matters. We can take at-protocol and do many integrations by our own to create own high level crystallization. Or we can give up on current AP and move on to the next one some time. The more ideas you give to the world the more it gives back.
If they are to make their own evil corp then you and I are allowed not to be it’s customers. Decentralization is about community and people
Check out Mudge’s whistleblowing report:
https://techcrunch.com/2022/09/13/twitter-whistleblower-mudge-congress/
I read the whole 200+ pages, here are some excerpts that I found the most damning:
https://rys.io/static/TwitterWhistleblowerRevelationsExcerpts.html
And yet when somebody forks Mastodon (like GlitchSoc or Hometown), they are not beholden to Eugene. Moving between instances works. Moving between forks works. Yes, Mastodon has an outsized presence in fedi, but it has nowhere near as much control as Google Search has over website traffic, or as the biggest BS’s “search and discoverability layer” provider will have over BlueSky’s users.
Fedi is simply a different architecture on a very basic level, built in a way that is not as supportive of economies of scale and “winner-takes-all” model, as BlueSky is.
Would I want fedi to be more diverse in instance software offerings? Totally. Is it fair to compare this to secondary centralization that BlueSky will support by design? Absolutely not.
Differences in protocol design define what will happen in a network run on it. Differences in design between BlueSky’s protocol and ActivityPub mean that secondary centralization will be way, way easier in BlueSky than it is on fedi — to a point that it seems this is by design.
How’s that working for us all regarding Google Search? We’re not even paying Google for search and yet we cannot escape it. I use DuckDuckGo daily, and yet I still am sometimes forced to use Google Search. Websites that want any real traffic need to optimize for Google Search.
“Voting with your feet” is a naïve myth unless the underlying system actually empowers people to do it effectively. Fedi does. BlueSky very definitely does not in the layer that matters.
Totally. No wonder, then, that BlueSky’s design carves out a space for secondary centralization exactly where communities and people can be found: search and discoverability.