you are viewing a single comment's thread.

view the rest of the comments →

[–]JasonCarswell 2 insightful - 2 fun2 insightful - 1 fun3 insightful - 2 fun -  (4 children)

Yes, Lemmy is a problem. Lenny is a kind of solution. It's not perfect, but it's better than having SaidIt be a single target.

Discussions.app seems like it's for phones. Is there a PC GUI?

 

I have an idea for a solution - of course this is all easier said than done, but IMO, it seems like it could logically be done. Build a database that 1) collects metadata and statistics, 2) compares the stats, 3) weighs the odds, 4) makes predictions, 5) learns from the results. Obviously you'd need a large enough group, and newer or original content would be less predictable, but you would eventually even be able to weed out the legit users from shills, etc - simply by their results. This could of course create bubbles too, but I would rather think of them as organic communities. To make sure you had an authentic bubble rather than a shill bubble you'd need to mix it up now and then, intentionally. I like putting control in the user's hands so they would have a say in their settings, to a limit, because part of this would require throwing some crap at users now and then in order to verify their tastes.

You get a feed with 10 things: 2 you're sure to like, 2 you'll likely like, 2 you may like, 2 you won't like, and 2 wild cards. Imagine something like this, customizable to a point yet also with mandatory participation at the bare minimum. Obviously this is just an example but the fundamental idea I think is good.

[–]hellosaidit[S] 2 insightful - 2 fun2 insightful - 1 fun3 insightful - 2 fun -  (3 children)

The discussions.app ui should change based on scale so there is both pc and mobile, neither is perfect as the dev is not great at frontends. Also as they just did a big change things are still in flux with bugs and features.

I have never seen algorithms work well. Reddit just introduced an ai doing what you describe and its a nightmare.

Pocketnet.app is big on algos and they dont work there.

Bubbles are unavoidable but by allowing the user to control it and not little dictating mods there is more hope that bubbles will pop or defuse. Its why I am so gunho about discussions.app and their moderation or content management. In theory it should solve the problems of reddit.

[–]JasonCarswell 2 insightful - 2 fun2 insightful - 1 fun3 insightful - 2 fun -  (2 children)

the dev is not great at frontends.

Flow is critical. It might be worth finding some expert help in GUIs. We recently tried RetroShare which is IMO an ambitious GUI mess with lots of features but not organized well - as if they patched it all together from spare parts without intentional purpose.

My dream GUI and user preferences would have 16 modes: day/night + mobile/pc + 1) default-reset-plug'n'play, 2) basic beginner, 3) advanced, 4) expert.

I have never seen algorithms work well. Reddit just introduced an ai doing what you describe and its a nightmare.

I suspect that's because Reddit is trying to control things. I doubt the algorithm, the data, the analysis, and what they do with it is open source. If they just let it do it's thing then the results would speak for themselves and the "communities of commonalities" would reveal themselves like and insanely complex Venn diagram and users could write their own custom expressions and/or have GUI preferences to set up how they want to interface with their interests.

IMO the algorithm would be best kept as simple as possible and let the complexity unfold in the opinion votes, topical metatags, and other metadata - and how they are classified/interrelated.

Bubbles are unavoidable but by allowing the user to control it and not little dictating mods there is more hope that bubbles will pop or defuse.

Exactly!!!

Its why I am so gunho about discussions.app and their moderation or content management. In theory it should solve the problems of reddit.

Unfortunately none of this is apparent at first glance. It might be worth drafting up an outline and stuff on a wiki somewhere for lay-people, your target demographic. I'm guessing there's a GitHub thing but that's all Greek to me.

[–]hellosaidit[S] 2 insightful - 2 fun2 insightful - 1 fun3 insightful - 2 fun -  (1 child)

[–]JasonCarswell 2 insightful - 2 fun2 insightful - 1 fun3 insightful - 2 fun -  (0 children)

Thanks!

Also, you may find some stuff in /s/IdeasForSaidIt interesting regarding SaidIt, including all the SaidIt-related subs in the sidebox.