Fediverse.observer (one of the longest running fediverse instance and software trackers, as it started for Diaspora) shut down. I think it would be really good to figure out how we can at least get a copy of the historical data and get that back up somewhere.

learned this thru cake.kobel.fyi/c/fediverse@lem

cake.kobel.fyiFediverse Observer is offline and no longer being developedThe domain https://fediverse.observer redirects to sedo now. On October 11th [these two commits](https://gitlab.com/diasporg/poduptime/-/commi…
😮1

Replying to @⁨liaizon@wake.st⁩

@liaizon I'm for backups of the data but also it should definitely be contextualized with, people (of those who have been indexed) have been lying in the API responses (as they should) since the start to make the graphs pretty&funny. Like it was an interesting collective-project but not an uncomplicatedly objective one.
So hopefully the backups of the data somehow keep that context in or adjacent to them.

Replying to @⁨liaizon@wake.st⁩

@liaizon That there is no objective "bounds"/"border" to the fediverse that you can point at or ask about everything "inside", that summing self-reported numbers of posts/accounts don't add up to a coherent total "activity" in any useful sense -- nobody is experiencing or has the ability to experience that, it has never been assembled or queryable, and the different posts on it aren't the same things -- an instance with 8 people talking to each other a few times a day which might never even have turned up in the aggregator's awareness isn't ten-or-more times less "active" than an instance with 80 RSS/twitter mirrors updating on the hour that hooked up to a relay, in any useful, more than a few steps away from immediately-circular, sense.

Replying to @⁨gaditb@icosahedron.website⁩

@gaditb I agree with this in general but being able to navigate this data even in its very rough form has been super useful regardless of its inaccurate and messiness. if its not aggregated somewhere its even harder to get a sense of the legitimacy of various servers. how long have they been filled with spam for instance. sometimes the fact that an instance is reporting 1000 active users even though there is only 1 real human makes it so you can feel confident that its unmaintained and blockable

Replying to @⁨liaizon@wake.st⁩

@liaizon @gaditb I use it for finding unmaintained servers so that I can autoblock them (and automatically revoke the blocks once a server is updated) - this is keeping a lot of spam away, because spam bots love unmaintained servers with open signups.

Bulk querying from fediverse.observer is a lot more efficient than querying each node individually. Fortunately, I was already using FediDB as a second data source.

Replying to @⁨gunchleoc@mastodon.scot⁩

@gunchleoc @liaizon So this sort of usage I can honestly respect, but it's notably not what gets presented top-line by any of the Observers/Stats Aggregators there.

Comparing an instance's self-reported stats to themselves over time, or to a qualitative look at the instance itself, are reasonable comparisons that keep those numbers in operations sharing approximately the same unit ("number of posts /in this space/", "number of [active] users /in this community/").

Those are not adding columns of different numbers that are not meaningfully the same unit, and putting it into a graph. But that's exactly the stats -- Accounts and Monthly Active Accounts, Servers, Posts, globally summed or summed by self-reported server software -- that FediDB (and unless I'm forgetting Observer) surfaces, solely (unless you dig and know to dig), as What Its Data Is.

And those stats are just absolutely meaningless.