That graph is so misleading. Makes it look like almost all the users disappeared but the Y axis only covers a small range at the top.
The full range is about 5.5%. So while it is misleading, a 5% drop in a graph that consistent isn’t nothing. Something substantial absolutly changed
Maybe whole Instance that went offline.
That’s my guess to
Cool then it should be 5.5% of the visual space for it to not be misleading. But it’s represented much larger. And OP is (edit to sound less mean) not updating the post sooo…
I call shade
It’s like 70k users.
Exactly. Not the over a million that it looks like at a glance.
The user count isn’t helpful anyway, active users is a much better measure.
true that
I captured the graph with the number after the decrease at the bottom right to try to show the number of lost users, but I see where you come from
If you saw why people would criticise it then you’d… edit the post, recapture the graph with an accurate visual representation including the zero on the Y axis?
Well…
- that’s how the website displayed the graph when I took that screenshot, I’m pretty sure you can’t modify the axis
- I’m not on my computer at the moment, editing is the most I can do right now
- if you can capture the graph with 0 on the Y axis, please post it below and I’ll update the picture
It’s the only way to demonstrate the fall, of you did it at scale it would’ve even be noticeable.
5% is noticeable unless the graph is under 20 pixels tall. Even then, dithering or antialiasing techniques could make it visible.
There are people here who didn’t even notice the axis and are confused. How do you expect everyone to notice a 5% drop lol.
Well if they didn’t even notice an axis on a graph, they might be too stupid for a graph…
My point was only to say that 5% is very mich visible on a 0-100% graph.
My point is that it is hardly visible and doesn’t show the exact numbers properly. I prefer the current style of graph for showing sudden changes if the previous data points isn’t that important in the context. But everyone has their preference I suppose.
Lies, damned lies and statistics.
no it’s not?
you can see the axes and op even mentions that it’s a 5% drop
the graph is clearly just fitted to the data
I edited the title after their comment, it wasn’t that clear at the beginning
In my classes on analytics, we were taught to prefer using normalised axes starting at 0 to more accurately put changes into perspective.
the graph is clearly just fitted to the data
That’s the problem. It’s heavily skewed when compared to the greater overall engagement statistics.
would this not happen if a whole instance got shut down?
Either that or blocked the crawler on their Instance.
I have a feeling it was caused by something like that, likely a failure of some kind. The drop is way too steep and abruptly for it to be natural.
An active instance would have an impact on active users
Not immediately though right? since the active users are a month or half-year. Or does it automatically update that too?
Active users are daily too (you have day, week and month counters in the sidebar of all communities)
I was talking about on the fediverse observer. It wouldn’t show up immediately there.
The Fediobserver has daily and monthly stats for active users. I assume daily indeed follow daily variations. You also saw those huge bumps when LW updated and started counting votes as activity
I think the observer shows daily and monthly stats for the active users per month and active users per half year so the active users per month wouldn’t change as fast I think.
Also about it being a botfarm I do think that is a possibility. Actually there is more evidence for it when you see extend the graph to 120 days and see a huge uptick in users and servers at the same time. Edit: 2024-7-29
Edit: wording
I think the observer shows daily and monthly stats for the active users per month and active users per half year so the active users per month wouldn’t change as fast I think.
Indeed, I was remembering it wrong, you might be right!
Also about it being a botfarm I do think that is a possibility. Actually there is much clearer evidence for it when you see extend the graph to 120 days and see a huge uptick in users and servers at the same time. Edit: 2024-7-29
Definitely
The source code is open source so I believe someone can tell us if need be.
Did an instance go down?
Maybe it’s a problem with the crawler.
Probably but which instance has over 70,000 users?
not sure when entire fedi-reddit is like 50k AMU lol
Which one of you forgot to open the app so we all ceased to exist to save on simulation resources?
Go on… lol
At least user a log scale. But start at 0
Feel free to suggest this to the Fediverse Observer team
Ah, I didn’t know.
The monthly graph is pretty interesting in the same way, -400k?
Lots of farmbots going down looking at the servers graph
Interestingly the amount of active users status pretty solid at 200k indeed
Interestingly there seems to have been an uptick in comments.
Indeed, the plot thickens.
Maybe the farm bot owner decided to make all comments by a single bot now.
Curiouser and curiouser.
Probably some mundane explanation but still…
LOL
Was the metric measured before the day the was over which would limit the timeframe of the final day and show fewer users?
I dont see this chart when I click the linkIt should be there, it’s the second graph
