I think this is normal. Everyday some people sign up so the graph goes up, but the tendency of the network is that people get less active.
The people who sign up daily could still be very active, but if more people leave (that signed up, maybe some time ago) the graph still will go down. Right?
@randahl but if for example in one month 100k people newly sign up, but 150K get less active this would explain why one graph is rising and one falling
@randahl#Mastodon is more a space for subcultures and less for mainstream users. I know people who prefer bluesky, since they don't fell wellcome over here. The #Fediverse is too exclusive and special interest for many people I know.
@randahl
I am posting this here from beeing on Mastodon since many years, and having collected about 1600 people that i follow. Or at least that i wish i could still follow.
About 50 of them are still active. The other people have slowly abandoned their accounts. Espically the recent year, everyone ran to Bluesky.
About 10 of the people that i follow are pretty active and make now more than 80% of my timeline.
About 3 or 4 remain who share my primary interests.
@randahl coming back to this, it's easy to misunderstand the dynamic at first glance considering the vastly differing scales on both axes of the graphs.
if we limit both to jan 2023 onward, you'll see growth of 3 million (rather than 6 million), and if you pay attention to the y axis you'll notice monthly users are down about 400k in the same period
i'm betting most of those users under "growth" are actually account migrations, and perhaps declines in user activities are exacerbated by servers going down or fediblocks somehow (the latter being a reporting bug)
@randahl Part of it is this: Mastodon can be put aside relatively easily. Twitter provoked people periodically (literally) by using algorithmic engagement. Eg showing you something you'd previously reacted to. The less I used Twitter the more obvious it was. Once properly recognised it was easier to ignore.
@randahl
You ask about Mastodon but post fedi charts. First of all, all the numbers are self-reported and unverified, second, software other than Mastodon represents 27% in the fedidb totals, and many of those software do not report their MAU.
Example: https://fedidb.org/software/mobilizon
@randahl critical mass build-up. People don’t just start using a new platform daily. They make a decision they need to check this new thing out, wait a couple of months to a couple of years, sign up, see their friends are not there, get out, get a reminder through other media, come back to see their friends are here and active and then start using it daily. I signed up for Twitter in 2010. I didn’t start using it until 2013, and daily - until 2015.
@randahl That might be the wrong chart. Instead of labeling as “growth”, that just total signups. Make a chart for “churn”. How many leave, and how long did they stay? If folks churn out in short period of time, that’s diff than long term churn. Then you’d understand was active users high because short-term churners signed up before but not now? Or are long term churners the cause of the declines? The former isn’t as big a deal as the latter.
@randahl Most of the people I follow don't post much anymore. Whether we like it or not, the absence of an engagement algorithm of any kind does kind of work against, well, engagement.
@randahl
I'm guessing bot signups. When I used to run a discussion forum board, it happened a lot to all discussion forums and that's what launched #ProjectHoneypot back in the day.
@randahl Some signups might just be bots and spam. The active count looks to me like „some people stick more than others“ - and that makes sense to me, given the Fedi‘s features and culture.
@Aethelstan@randahl it certainly does take more time and patience to find the better accounts on here (though they’re better than the best ones on the other apps once you do)
@Aethelstan more and more of them can be followed here as well. Especially now that threads is a Fediverse thing. Just this week, I noticed @gtconway3 for instance.