@ricci You mention elsewhere that for your plots quantifying Fediverse decentralization, you (want to) count only those Threads users that federate with other instances running ActivityPub-based software.

[feralthoughts@union.place]

@ricci

You mention elsewhere that for your plots quantifying Fediverse decentralization, you (want to) count only those Threads users that federate with other instances running ActivityPub-based software. In contrast, Eugen Rochko seems to count any instance running software that natively speaks ActivityPub as part of the Fediverse.

Hence my curiosity questions below about some other software.

#ActivityPub #ATProto #Fediverse #Bluesky #Threads #Hubzilla #(streams) #Forte

(continues)

[feralthoughts@union.place]

@ricci

Are you counting the user base from instances which run software such as Hubzilla or (streams)—specifically, software that is NOT based on ActivityPub—but which federate with all the instances running ActivityPub-based software such as Mastodon?

If yes/no, why?

For context, the present mandarins of the ActivityPub world do not count Hubzilla or (streams) as parts of the Fediverse.

#ActivityPub #ATProto #Fediverse #Bluesky #Threads #Hubzilla #(streams) #Forte

(continues)

[feralthoughts@union.place]

@ricci

This also raises another question. Hubzilla and (streams) both have nomadic identity, so a user can have multiple clones of a channel across different instances, and activity on any one of those gets mirrored across all the clones. And an account on an instance can have multiple channels—a channel is what participates in the Fediverse.

#ActivityPub #ATProto #Fediverse #Bluesky #Threads #Hubzilla #(streams) #Forte

(continues)

[feralthoughts@union.place]

@ricci

So if you are including Hubzilla and (streams) users, you will have to worry about counting each channel for an account, but not counting any of the clones of a given channel. Are you doing that?

#ActivityPub #ATProto #Fediverse #Bluesky #Threads #Hubzilla #(streams) #Forte

(continues)

[feralthoughts@union.place]

@ricci

The author of Hubzilla and (streams) also developed the ActivityPub-based Forte, which has nomadic identity, with all of the properties described above.

https://codeberg.org/fortified/forte/

#ActivityPub #ATProto #Fediverse #Bluesky #Threads #Hubzilla #(streams) #Forte

(continues)

[feralthoughts@union.place]

@ricci

Forte certainly qualifies as part of the Fediverse, by any criteria. So you will definitely have to worry about counting each channel of any account on a Forte instance, and about not counting clones of that channel.

#ActivityPub #ATProto #Fediverse #Bluesky #Threads #Hubzilla #(streams) #Forte

(continues)

[feralthoughts@union.place]

@ricci

To be clear: the active user base of Hubzilla probably numbers less than 1000, and those of (streams) and Forte far far smaller than that; so in your plots, their exclusion/inclusion is not going to be visually discernible to anyone. At present, it will not affect any inferences one may draw from your plots.

But for the sake of completeness, you may want to consider these questions.

#ActivityPub #ATProto #Fediverse #Bluesky #Threads #Hubzilla #(streams) #Forte

(continues)

[feralthoughts@union.place]

@ricci

Also: if FEP-ef61

https://codeberg.org/fediverse/fep/src/branch/main/fep/ef61/fep-ef61.md

becomes widely adopted in future, then other ActivityPub-based software may incorporate nomadic identity, and you may have to worry about all of the above at that point.

#ActivityPub #ATProto #Fediverse #Bluesky #Threads #Hubzilla #(streams) #Forte

[ricci@discuss.systems]

@feralthoughts

I think Eugen is wrong.

Some great question, answers below.

[ricci@discuss.systems]

@feralthoughts

My data currently comes from https://fedilist.com/ . They do not describe much about their crawling process, unfortunately (but they do provide the data in a way I can access it in a complete form, unlike others like fedidb, which is why I'm using them). However, based on other sites that do describe their crawling process in more detail, I assume that what they are doing is starting from well-known seeds, asking each its peers, then querying nodeinfo on each of those peers. This process does not, as far as I can tell, turn up much, if any, stuff that is not ActivityPub based.

You can have a look at the data itself here: https://codeberg.org/ricci/are-we-decentralized-yet/src/branch/main/fedilist-fromhtml.csv

[ricci@discuss.systems]

@feralthoughts As mentioned above, I'm using someone else's dataset that gets user information from nodeinfo, so as far as I know, this is not an issue (for better or worse) with the data.

[ricci@discuss.systems]

@feralthoughts The dataset I have (link above) currently includes 4 instances that identify themselves as Forte. All report zero users. I assume this is a deliberate decision, not that they actually have no users - other software similarly misreports user counts in nodeinfo. I can only deal with the data they give me...

[ricci@discuss.systems]

@feralthoughts Yeah I don't know what the practical effect of this on nodeinfo will be.