sauce https://fediverse.observer/stats

https://fediverse.observer/dailystats

Whats that rapid up and down in July?

edit: Hey now OP, that ain't right to edit the graphic after the fact and then NOT EVEN MENTIONING IT!

Do you mean December-January?

Someone went on vacation.

No, OP changed out the Graphic and on the datapoints representing mid July first went high up, then dramatically went down and then up again continuing at that point trending slightly up.

this one:

“Are you winning, fediverse?”

Sorry I just embarassingly realized the graph in the background was only over the scale of a couple of days and thus wasn't impressive or noteworthy at all lol.

I included a link to the daily graph pertaining to your question from the original meme I made in the post.

My evil scheme has worked perfectly....

do you mean the period between 06-06 and 06-12? i.e. early june or do you mean the end of june 06-27?

If the question is the former I would imagine these kinds of changes are likely structural changes in the fediverse between instances or changes in how the numbers are calculated for this graph that are artifacts of the algorithm being updated or changed, though it may be a "real" signal I doubt it is indicative.

What is indicative is the steady, consistent increase of conversations and posts over the scale of years!

To put it another way, what I consider a likely real signal that we should feel positive about (a steadily increasing number of posts and comments) is overlayed with an offset event that is probably an artifact and not something to draw broad conclusions about the fediverse from... though who knows I could be wrong.

yeah that was a brainfart, I meant June 6th to June 12th.

How did they go down? Is it like, a major instance going dark?

Think of it this way, in a federated system of multiple interacting instances running multiple different mostly overlapping softwares the question of "how many total posts are there?" and "how many total comments are there?" can only be answered with contextualized statistical observations that attempt to estimate a value that at a philosophical level may not even be establishable as a precise question... which isn't even getting into whether you include really creepy/toxic insular communities on the fediverse or not which is an ethical, cultural and practical question with many grey areas and details.

Also, remember this is all at least to a certain extent opt-in statistics (as it should be) so you can't expect you are sampling the entire signal... which means that stepwise offsets like this may just indicate differences in how much of the signal that can be observed that arise from changes in how instances federate, report information to bot crawlers and quantify posts and comments themselves.

nice!