freamon

Yeah, I know what you mean. That note is misleading, and kinda redundant too - you can physically de-select Undetermined in the UI, but the change won’t actually take if you press ‘Save’.

Most likely reason is that you unticked ‘English’ as a language you understand when you were playing around.

A new one appeared at pretty much the exact moment you commented, ha ha.

They’re not automatic atm, 'cos of site upgrades throwing things off, so it’s when I remember to run it manually. Should be back to automated soon, hopefully.

Edited 'cos feddit.uk’s upgrade to 0.19.2 skewed the stats. They’ve stabilised now, apparently. (That poor “we’re all quackers here!” bloke is never going to get a break)

When/If lemmy.world ‘upgrades’ to 0.19, I’ll increase the minimum threshold for active users. I’ll also start excluding communities where the growth is mostly coming from negative engagement.

Shitty Life Pro Tip - just the one?

lemmy.ml fell over yesterday, btw, so their communities are missing today.

Testing comment notification 1

The new post in cremposting is titled ‘Trying to revive this community’, which’ll be extra challenging considering it doesn’t look like the post federated anywhere.

Edited to remove some erroneous results, but it’s perhaps by-the-by, since federation seems noticeably worse since 0.19. ‘Resolve_Object’ doesn’t seem to make any effort to retrieve a remote community that the instance hasn’t already heard of. e.g.:

curl --request GET --url 'https://feddit.nl/api/v3/resolve_object?q=https://reddthat.com/c/dirtroads' --header 'accept: application/json'
gets
{"error":"couldnt_find_object"}
every time

You can use the web-ui to force through a blank version, but a community without any posts isn’t much use.

Posts from the past couple of days were MIA due to federation issues from feddit.nl (related to the 0.19 upgrade, I think).

Re: notheonion@lemmy.ml - they’ll be the odd crazy stat like this for a while, 'cos a dead community will only have 1 or 2 active users irrespective of how you’re measuring them. The first new post in a such a community will spike the Active Users level, which I can’t accommodate for because the ‘first new post’ can happen at literally any time. In future though, levels will be averaged out by similar levels of negative growth and will have to compete with more communities measuring things in the new way.

Re: fbstolencontent@lemmy.ca - from the instance I’m currently on (lemmy.world), there only appears to be 19 posts from 3 months ago (same as on other big instances I’ve tried). Visiting the site directly, there’s loads more, but they haven’t come through because there’s apparently no subscribers from lemmy.world (which is the way it’s supposed to work, I guess)

Some of the strangeness is probably because, at the minute, the ‘active users’ list is a mix of instances that count them as people who post, comment or vote (e.g. lemmy.ml), vs. those that count them as people who just post or comment (e.g. lemmy.world).

The rest is probably because, for those that have changed how they count, there’s a bit of a ripple effect which’ll take a while to dissipate. For example, lemmy.ca upgraded to 0.19.0 overnight, so results from any of it’s communities aren’t included today, because it would be trying to calculate a change from one measuring system to another (e.g. nostupidquestions went from ‘103’ to ‘791’ active users, but it didn’t really, obviously).

However, whilst I’m trying to mitigate the changes from measuring systems, I can’t do much about the fact that now including voters has an exponential effect on how many active users a community has. When lemmy.ml upgraded, diy@lemmy.ml’s active users went from ‘6’ to ‘23’, and I didn’t include that change, but there’s a been a popular post since then, and as we’re now including the voters, the users are at ‘138’. This is a much bigger change (from ‘23’), but as far as the new measuring system is concerned, it’s a legitimate one.

So, until all the instances are on the same version, and until the change in metric has had a chance to bed in, there’ll be some odd-looking results, that I don’t think I can do much about.

All I can say is that any linked community does have at least one recent post, so they shouldn’t be completely dead. If they appear to be so it’s probably for some other mundane reason (blocked users, defeds or poor federation, different languages, etc)

I re-ran it manually today, excluding lemmy.ml and lemy.nl, 'cos the percentage growth from one way of measuring to another way of measuring is pretty meaningless (I realise Ategon warned me, but it’s tricky to catch the changes before the bot does)