I only see a handful of communities posts, but miss a lot from smaller communities that aren’t as active but do get daily posts. Reddit blended in smaller communities really well and it’s something I’m struggling with here.
Have you tried playing with different sort types?
“Hot” is the best sort type imo
I actually like sorting by 'New" on fedi - it promotes engagement in new content and it’s generally decent content here as opposed to what you see in “new” on reddit…
I’m sorry for showing up so much there
Please continue, us commenters need things to comment on. And the lurkers need something to read.
We need people like you!
Thanks
Makes me wonder if we’ll all just end up good friends or fade into obscurity if Lemmy doesn’t take off. There’s a good amount of discussion on here but it’s still small enough that you kind of know folks.
I guess I should add that I’ve done hot, top for the day, 12 hr, new, active, etc. I have a handful of communities I’ve never seen on my feed. I have to go to them manually.
That’s because there is no sorting algo based on… well anything. Tracking your interests and how much time you’ve spent on a comm as well.
If you choose popular/active, it gives you the most active posts, if you choose new, you get the newest ones. It doesn’t prioritize in any way. The idea is NOT to track what users do.
Though I miss comm/sub tracking as well, it gave good suggestions and kept the feed filled with interesting posts.
It could probably be implemented client wise (in app), but this thing barely works as it is now (so do the apps), so that will take time.
I’ve been feeling the same pain, of small, less active communities just not showing up in the feed. While in principle I am very much against “the algorithm”, this will lead to a feedback loop where small communities remain small because they will be much harder to find and revisit, even more so as large ones grown even larger.
We need a kind of sort that is able to get posts of smaller/less active communities interspersed with the rest. This does not/should not be a user profiling algorithm, etc. Just a blind “show the latest post from every community unless it is over x days old, and only then show the second newest,etc” or similar would help.
As things stand now, I’ve considered unsubscribing from some communities so that they do not overwhelm the feed, but it feels like a bad solution.
I have about 30 communities I’m subscribed to and 90% of it is the memes and shitpost communities. So 2 communities are basically all I see. They’re probably the most active but it’s still extremely disproportionate. I’m like you I want to unsubscribe but it doesn’t seem like the right way. Maybe I’ll do it to test if smaller communities content comes through.
I think I found a good solution: I created 3 accounts in 3 different instances. In one instance, I only subscribe to serious stuff like worldnews and technology and things like that. In another account, I subscribe only to memes. Both these accounts have NSFW disabled. The third account is the NFSW one.
At least in the Liftoff app, it’s trivial to change between accounts - so I basically have 3 separate feeds, with the added benefit that if any one of the instances is down I can still use Lemmy and see what’s going on in the “All” tab in any of the instances.
This is a good idea to try out. I’ll try it thanks!
I’ve considered this, but fragmenting my identity seems a bit of a hassle as well. Ultimately, it will alleviate the problem but not solve it, anyway.
In any case, there is an open issue on github (someone beat me to it). Let’s hope it gains traction!
You don’t need user tracking to try to show smaller communities in between larger ones, you just need some sorting method that incorporates that. Hot is already scored with some algorithm, that algorithm could take into account size as well, or just a new sort with that.
The equivalent of multireddits could help, but I don’t think it’s implemented yet.
Could probably be done even via app.
Yeah, you’re probably right… something could be done in that regard I guess. But the algo needs to be more complicated if it doesn’t actually track user data… or less complicated, but less effective.
I’d have to imagine that this is something that could be implemented per-instance, as well. For example, some instances could work toward a more aggressive recommendation algorithm, while others can aim for a more neutral feed. I can see the appeal to both versions, and honestly would probably keep an active account on both of those such instances, myself.
That would have to be done in the BE, so even if it custom implemented, you’d have to reimplement it after every update. That is a PITA to maintain trust me. Why do you think so many companies contribute to Linux code, cuz it’s a PITA to maintain your custom patches to the code, you maintain a very small subset of those and let the rest be maintained by everyone else.
I know specifically for kbin that the topic of what shows up in what section is a point of discussion with the other devs. Getting these things to work “right” is going to be a challenge.
I like the idea of more “metrics” being used to help “weight” the search results. E.g. if you’re viewed heaps of content from “magazine A”, then content should be surfaced more from that magazine.
I think search / these collections will take a while to get into a really polished state (and it’ll change as the site grows)
I like the idea of more “metrics” being used to help “weight” the search results. E.g. if you’re viewed heaps of content from “magazine A”, then content should be surfaced more from that magazine.
I think that is the way ‘best’ works on reddit and it is the way I preferred to use that site. I agree with you that using metrics would be awesome with the added filter ‘best’.
