I get that the Mastodon folks are out of date and don't realize how much development has happened on ATproto in the last year, but it makes them all sound silly when they repeat things as if it were still 2023.
-
@mmasnick @folkerschamel @mmasnick.bsky.social
@ricci did answer your question very specifically here:
Rob Ricci (@ricci@discuss.systems)
Hi, creator of the chart here. I'm disappointed to see people using it as a way to try to dunk on or bully various protocols. @mastodonmigration@mastodon.online and @folkerschamel@mastodon.social if you're going to share the screenshot I'd appreciate if you'd update it to the latest version, which is slightly clearer that the gauge is showing the HHI, a measure from economics that captures market concentration - not just users on the biggest servers, as are in the table. It also now has git forges as well to show that the point is to compare many forms of decentralized networks (I'm working on getting data for more); it's not just there to dunk on atproto. The heading that you cut off also (tries) to make it clear that it's showing user data, eg. it's not attempting to show things like moderation or feeds, where Bluesky likely has more diversity. @mmasnick@mastodon.social I'm working on getting Threads data in here; Meta doesn't make that data available via the standard APIs so it's not in my data sources. The most recent estimate I can find for the number of Threads users who have opted in to the fediverse is around 25k-50k as of Dec 2024: https://fediversereport.com/why-is-meta-adding-fediverse-interoperability-to-threads/ . So while I do want to get it in here for completeness, Threads doesn't really move the needle. I do want criticism of this data, the way that it's presented, and other ways I can show the decentralization that does exist in the deployment of the AT Protocol ecosystem. I've made several changes in response to feedback, some of it from Bluesky team members - in fact creating this dashboard in the first place and the way I'm getting ATProto data was the idea of a Bluesky team member. (He didn't suggest the specific use of HHI, however) But that said, this specific criticism is off-base: if we add the number of Threads users who are actually fediverse users, nothing changes. If we were to add in the 400M MAU that Threads claims to have, but who are not fediverse users, that would be kind of like asking why we didn't put Facebook on the AT Protocol side: not a meaningful thing to do. And, if, hypothetically, all those 400M Threads users *were* fediverse users, that would, in fact, centralize a *lot* of power in Meta's hands - not all of it, but a ton. We all know how networks work. This is one reason (the other being the, you know, everything, about Meta) that fediverse folks were quite worried about Thread's entrance. My goal in building this thing is so that we can watch the deployments, nothing more nothing less. Hopefully, blacksky grows and we see that reflected in the Atmosphere side. I've seen your recent post about a bunch of AT Protocol development that is not from Bluesky. Great. The point of this chart is to watch that grow. There are plenty of anecdotes, those are good and necessary. Data is part of the story too, and that's what I'm trying to provide here. Speaking of which, I would very much like to get data from Bluesky regarding the use of third-party feeds and moderation tools. As far as I can tell, I can only get this from the Appview, and I can't find any indication it's exposed yet. I hope that you understand that I'm trying to provide a valuable data source here, and if you do, I'd appreciate if you could put me in touch with the right person to ask about this. And finally, thanks for One Billion Users, I had a great game with my spouse last night :)
discuss.systems (discuss.systems)
There was also more in depth discussion of the issue. It may be that you are not seeing the entire thread across the bridge.
@mastodonmigration @folkerschamel @mmasnick.bsky.social @ricci No @ricci is making a different (totally valid!) point. I am saying IF he included Threads, would you then say that Mastodon is less decentralized?
That's the only question I am asking. If your answer is yes, I would be confused. If your answer is no, you are admitting that this is not a measure of decentralization.
Which is it?
-
@mastodonmigration @folkerschamel @mmasnick.bsky.social @ricci No @ricci is making a different (totally valid!) point. I am saying IF he included Threads, would you then say that Mastodon is less decentralized?
That's the only question I am asking. If your answer is yes, I would be confused. If your answer is no, you are admitting that this is not a measure of decentralization.
Which is it?
@mmasnick @mastodonmigration @folkerschamel @mmasnick.bsky.social I think a better question here, Mike is *if* Threads had 400M users who were active on the fediverse, would the fediverse be more centralized?
I would say yes.
-
@mmasnick @mastodonmigration @folkerschamel @mmasnick.bsky.social I think a better question here, Mike is *if* Threads had 400M users who were active on the fediverse, would the fediverse be more centralized?
I would say yes.
@mmasnick @mastodonmigration @folkerschamel @mmasnick.bsky.social Of course, it does not. It has like 50k.
-
@mmasnick @mastodonmigration @folkerschamel @mmasnick.bsky.social Of course, it does not. It has like 50k.
@ricci @mmasnick @mastodonmigration @mmasnick.bsky.social
But even if all of #threads would be part of the #fediverse and make the fediverse practically centralized, it wouldn't change the the factual situation that
a) #mastodon is #decentralized,
b) #bluesky as being controlled by a single company is completely centralized, and
c) #atmosphere as being dominated by #bluesky is practically centralized. -
@ricci @mmasnick @mastodonmigration @mmasnick.bsky.social
But even if all of #threads would be part of the #fediverse and make the fediverse practically centralized, it wouldn't change the the factual situation that
a) #mastodon is #decentralized,
b) #bluesky as being controlled by a single company is completely centralized, and
c) #atmosphere as being dominated by #bluesky is practically centralized.@folkerschamel @mmasnick @mastodonmigration @mmasnick.bsky.social
Centralization is not binary the way you're presenting it here.
Let's say we have a 400M-user Threads and a 1M-user Mastodon. If they defederate or Threads dies, Mastodon users loose access to 99.8% of the people they could communicate with. Ouch. But Mastodonians still have 1M people in their network so maybe it'll survive. Definitely not certain, though, that's a big cut. That's why I say a version of the Fediverse with 400M actual federated Threads users would be quite centralized.
Now, people on one side may not actually give a shit about communicating with people on the other side. Fine, the people in both networks are not gaining a lot from federation. This seems to be more or less the status of most of the Fediverse and Threads, and why I think the correct thing to do is count the number of Threads users who have actually turned on federation, not the rest of them. If they defederate (as a lot of the Fediverse has done already), not a lot of connections are cut. This is why the existence of Threads does not increase the centralization the Fediverse today. This could, of course, change. This is why one should *keep* counting the number of Threads users who federate.
Now, let's do this for Bluesky and Blacksky (in its role as a PDS, appview, and maybe soon relay). If, today, they split (say, the Bluesky relay stops talking to the Blacksky PDS and appview) or Bluesky dies, the 718 people on the BlackSky PDS lose access to the 38M people on the Bluesky PDSes: 99.999% of the people in the network. Again, maybe people on Blacksky could care less about people on Bluesky. But, given the even vaster difference in size, I'd wager that Blacksky users are pretty strongly interconnected with Bluesky users. And again, maybe this changes; in the week I've been watching, the number of users on the Blacksky PDS has gone up by about 200. Maybe it continues to grow and gets a lot bigger, that would change the dynamic. So again, this is why it's worth measuring and watching.
-
@folkerschamel @mmasnick @mastodonmigration @mmasnick.bsky.social
Centralization is not binary the way you're presenting it here.
Let's say we have a 400M-user Threads and a 1M-user Mastodon. If they defederate or Threads dies, Mastodon users loose access to 99.8% of the people they could communicate with. Ouch. But Mastodonians still have 1M people in their network so maybe it'll survive. Definitely not certain, though, that's a big cut. That's why I say a version of the Fediverse with 400M actual federated Threads users would be quite centralized.
Now, people on one side may not actually give a shit about communicating with people on the other side. Fine, the people in both networks are not gaining a lot from federation. This seems to be more or less the status of most of the Fediverse and Threads, and why I think the correct thing to do is count the number of Threads users who have actually turned on federation, not the rest of them. If they defederate (as a lot of the Fediverse has done already), not a lot of connections are cut. This is why the existence of Threads does not increase the centralization the Fediverse today. This could, of course, change. This is why one should *keep* counting the number of Threads users who federate.
Now, let's do this for Bluesky and Blacksky (in its role as a PDS, appview, and maybe soon relay). If, today, they split (say, the Bluesky relay stops talking to the Blacksky PDS and appview) or Bluesky dies, the 718 people on the BlackSky PDS lose access to the 38M people on the Bluesky PDSes: 99.999% of the people in the network. Again, maybe people on Blacksky could care less about people on Bluesky. But, given the even vaster difference in size, I'd wager that Blacksky users are pretty strongly interconnected with Bluesky users. And again, maybe this changes; in the week I've been watching, the number of users on the Blacksky PDS has gone up by about 200. Maybe it continues to grow and gets a lot bigger, that would change the dynamic. So again, this is why it's worth measuring and watching.
@ricci @mmasnick @mastodonmigration @mmasnick.bsky.social
I didn't present it binary - I deliberately chose the words "completely centralized" versus "practically centralized" versus "decentralized".
And while of course centralization is not binary, and there are many discussions about details and about the future, we should not muddle the water and forget the overall big picture that for all practical purposes today bluesky is centralized and mastodon is decentralized.
-
@ricci @mmasnick @mastodonmigration @mmasnick.bsky.social
I didn't present it binary - I deliberately chose the words "completely centralized" versus "practically centralized" versus "decentralized".
And while of course centralization is not binary, and there are many discussions about details and about the future, we should not muddle the water and forget the overall big picture that for all practical purposes today bluesky is centralized and mastodon is decentralized.
@folkerschamel @mmasnick @mastodonmigration @mmasnick.bsky.social
Here's an illustration of the difference (in agreement with your basic point).
Threads is currently blocked by servers representing 31% of Fediverse MAU and muted by another 2.5%. (Data from https://fedipact.veganism.social/ )
That's the independent decisions of 3,280 instances. Several thousand others chose to federate.
Let's say the atmosphere had the same decision to make. One organization's decision re: whether or not Thread's PDSes on their relay would affect >99% of atmosphere users.
Mike will probably say that Bluesky might fragment, with people who don't like its decision, whichever way it goes, moving to services that make a different decision. Yes, it might. And they would get to take their data with them, which is great, and they can still communicate with the people who stay behind.
But I gotta say, these seem to me like very different decision making processes with respect to whether power is held centrally or distributed to independent actors.
-
@folkerschamel @mmasnick @mastodonmigration @mmasnick.bsky.social
Here's an illustration of the difference (in agreement with your basic point).
Threads is currently blocked by servers representing 31% of Fediverse MAU and muted by another 2.5%. (Data from https://fedipact.veganism.social/ )
That's the independent decisions of 3,280 instances. Several thousand others chose to federate.
Let's say the atmosphere had the same decision to make. One organization's decision re: whether or not Thread's PDSes on their relay would affect >99% of atmosphere users.
Mike will probably say that Bluesky might fragment, with people who don't like its decision, whichever way it goes, moving to services that make a different decision. Yes, it might. And they would get to take their data with them, which is great, and they can still communicate with the people who stay behind.
But I gotta say, these seem to me like very different decision making processes with respect to whether power is held centrally or distributed to independent actors.
@ricci @mmasnick @mastodonmigration @mmasnick.bsky.social
To add, I think the credible exit argument falls apart in reality anyway. See reality check Mississippi blocking: Many people on bluesky seem to not like the decision, but at the same time I have seen nobody acting on it and moving to services that make a different decision.
BTW, using a different analysis only 7% of the fediverse seem to be known to block #threads, down from 9% last year, see also https://mastodon.social/@folkerschamel/111612927848322776.
-
@ricci @mmasnick @mastodonmigration @mmasnick.bsky.social
To add, I think the credible exit argument falls apart in reality anyway. See reality check Mississippi blocking: Many people on bluesky seem to not like the decision, but at the same time I have seen nobody acting on it and moving to services that make a different decision.
BTW, using a different analysis only 7% of the fediverse seem to be known to block #threads, down from 9% last year, see also https://mastodon.social/@folkerschamel/111612927848322776.
@folkerschamel @mmasnick @mastodonmigration @mmasnick.bsky.social
Interesting post, thanks. I didn't cross-check the fedipact data against another source, that's probably part of our discrepancy, but the data from them does add up to about 900k MAU, so I don't know why it's such a big discrepancy. Anyway, we agree that this represents highly-decentralized decision making in action, right?
I think I'm more optimistic about credible exit than you are, part of the goal is not just that people will move, but that it will discourage individual platforms from taking actions that will piss users off. That said, I think it absolutely does not *guarantee* decentralization or user-positive behavior. The web search market has great credible exit, yet Google is staying around 90% (depending on source) even while making a lot of people quite angry about various things.
-
@folkerschamel @mmasnick @mastodonmigration @mmasnick.bsky.social
Interesting post, thanks. I didn't cross-check the fedipact data against another source, that's probably part of our discrepancy, but the data from them does add up to about 900k MAU, so I don't know why it's such a big discrepancy. Anyway, we agree that this represents highly-decentralized decision making in action, right?
I think I'm more optimistic about credible exit than you are, part of the goal is not just that people will move, but that it will discourage individual platforms from taking actions that will piss users off. That said, I think it absolutely does not *guarantee* decentralization or user-positive behavior. The web search market has great credible exit, yet Google is staying around 90% (depending on source) even while making a lot of people quite angry about various things.
@ricci @mmasnick @mastodonmigration @mmasnick.bsky.social
Yes, agreed that fediblock is a text book example how decentralization works well in practice.
Well, then #bluesky credible exit is the same as decentralization support of #atproto: cool tech, nice theory, clever marketing trick, but largely irrelevant in reality. Btw, #twitter has a credible exit too: just export your data and import it into something new.
PS: any chance of an explanation of
https://mastodon.social/@folkerschamel/115184281756139935 ? -
@ricci @mmasnick @mastodonmigration @mmasnick.bsky.social
Yes, agreed that fediblock is a text book example how decentralization works well in practice.
Well, then #bluesky credible exit is the same as decentralization support of #atproto: cool tech, nice theory, clever marketing trick, but largely irrelevant in reality. Btw, #twitter has a credible exit too: just export your data and import it into something new.
PS: any chance of an explanation of
https://mastodon.social/@folkerschamel/115184281756139935 ?@folkerschamel @mmasnick @mastodonmigration @mmasnick.bsky.social
I think the contention here is that Mastodon GMBH is to the Fediverse as Bluesky PBC is to the Atmosphere: both individual companies develop software, run services with users, and participate in a larger ecosystem that they don't control[*]. Each company can be called a "central" entity, and the whole can be said to contain parts that are outside that center. This is the point that I understand Mike to be making that he thinks you are missing or confused about.
At some level, this is correct. But I think this is a very shallow reading. If you look at the actual structure of these networks, they are very different. A large majority of the Fediverse is not subject to Mastodon GMBH's operational decisions (TOS, federation decisions, etc. ). There are other projects that rival the software produced by Mastodon GMBH in both scope and userbase. The pressures that shape the Fediverse are much larger *outside* of Mastodon GMBH that the ones that come from the company. There are tens of thousands of people in the Fediverse who, for better or worse, are social network administrators. Decisions are made by a huge set of people. atproto envisions a world where these things *could* happen - but so far as I know none of them are true of it today.
I think it is a major mistake to gloss over these differences.
[*] I think we will all concede that Bluesky does currently have significant control over the larger ecosystem through its control over atproto, but I am willing to give them the assumption of good faith in terms of handing off that control; they've started, and I assume they will complete the process.
-
J Jürgen Hubert shared this topic on
