A post from 2 days ago presented a graph that showed an important variation in the active userbase: https://discuss.tchncs.de/post/52565659
Using the daily rather than monthly view on https://lemmy.fediverse.observer/dailystats&days=120 shows a much stable line (especially if you take into account Piefed's growth: https://piefed.fediverse.observer/dailystats&days=120 )
Going through the comments in the other posts, a few recommendations that can help with the overall experience
-
use different feeds: either using different Lemmy/Mbin accounts (one account per type of content), or Piefed personal feeds, but being able to browse different feeds such as "Good news", "Hobbies", "Art", "Life advice" help to see more content than politics and tech
-
discover communities: subscribe to !communitypromo@lemmy.ca, !fedigrow@lemmy.zip and !newcommunities@lemmy.world to add active communities to your feeds
-
go to general communities rather than specific ones: the current user base only allows so much specialization. Your favorite city builder community may not exist, but !citybuilders@sh.itjust.works does. !stationery@lemmy.world and !pen_and_paper@lemmy.world may be inactive, but !journaling@sh.itjust.works is not.
-
use a client that allows for comments consolidation: I don't remember which mobile apps does it (Sync, I think?), Piefed has that feature built-in too. It allows to see all comments on a cross-post in the same view: https://piefed.zip/c/privacy/p/928874/worst-in-show-ces-products-include-ai-refrigerators-ai-companions-and-ai-doorbells#post_replies
-
report toxic users and avoid communities that do not handle your reports: quite a few comments mentioned that issue in the other thread. Mods can't see everything, reporting helps to keep the atmosphere of a community enjoyable.
-
use a client that implements keyword filters: quite a few mobile apps and alternative Lemmy front-ends do, Piefed has it built in. It can really help avoid the "doom and gloom" overwhelming your feed.
Finally, a few communities recommendations for lighthearted communities
A statistician explain to me why these graphs seem correlated beyond general trend? They both seem to have localized events on the same day, but given their different timescales that doesn't seem like it should be possible.
I raised the same concern on the other post too, but idk enough about statistics to for sure say something seems fishy.
(not a statistician) The events that WOULD be synchronized are: instances going down, users being deleted, and new user signups.
Maybe this is enough to make them look like this. Those are the events that cause jumps, whereas users hitting the 6 months of inactive time would just be a steady, stable, gentle slope.
Im not sure how these stats are collected, I assume that they query each server for its to make the chart, rather than query every server every day and copy the results.
If they're really copying the results, then you're absolutely correct that temporary instances outrages would cause those correlated downward blips, but I'm surprised to hear you wouldn't just be able to query servers to get this data on demand.
But then again if a server went permanently offline you'd lose that data forever. Hmmm