@nolan Same, except I just use the federated timeline as my "home" timeline, and i only follow people whose toots i don't want to miss

Having accounts on different instances is also a great way to tame the chronological timeline. My mastodon.social and freedom.horse timelines are pure silliness, which I occasionally need as a respite from the dry tech stuff I follow on toot.cafe.

Evans' point about how "normal people" don't want to curate their feeds is probably right, though. Algorithmic timelines offer the tantalizing possibility that you can have your cake and eat it too. "Follow as many people as you want! We'll just show you the good stuff."

When I was on Twitter I totally fell into that trap, following 1000+ people and just hoping the algorithm would take care of the noise. Unfortunately once you've reached that point, it's almost impossible to switch to curation.

@nolan iirc Twitter did have lists but they really made it as awkward as possible to use

@nolan This was, in my opinion, one of the only things that Google+ got right. It was easy to curate your following and likewise where to share things, since it was included as a first-class part of every part of the platform.

@nolan So why not allow people to choose their own algorithms?

@wohali Ooooh, that's an interesting option. Yeah, it'd be neat to have multiple timelines with different algorithms for each one, and you could choose between them.

@nolan

Maybe someday Mastodon could have an algorithmic timeline. At least unlike Twitter/Facebook, it'd be open-source and auditable. But I kind of suspect most Mastodonians would rebel against that, since they're explicitly joining Mastodon because they want *more* control over their online feeds.

I guess the question is whether curation/filing really is for the geeks and the power users, and "normal" people will always want an algorithm to do the work for them.

@nolan I'd like some algorithms, but I want to know what they are and how they work, and to be abkle to choose between them freely. :/

@nolan Boy when I see the reactions that happened after the "who to follow" PR almost a year ago I dont want to imagine the uproar that would trigger an algorithmic timeline :'D

@nolan (and that's too bad because the feature in itself was nice imo, w.t had it to try and it was nicely done, but the gut-reaction from ppl condemned the feature even before it could prove its own point :( )

@nolan I don't think the problem with Mastodon will be ever so pronounced as with Twitter, because Twitter is massively spamming and pushing you to follow, follow, follow. Without this constant push I doubt most people would end up with thousands of followings.

I believe humans are pretty good at judging when they have enough friends. We learn this through our whole life, after all.

@nolan
Well, those algorithms could be opt in. And even have many of them. Like, “show me only toots with media” show me toots from people I favourite the most”, and such.

@nolan Why not collections a la G+ so we get a better granularity and baked in curation?

@nolan @kemonine I never did get much out of G+ collections, but maybe that's because the are bound to persons too. Their minor use was that you could filter for just the aspects of the person that are interesting to you by following one of their collections instead of the user.
I very much would like to be able to subscribe to topics instead of persons, and have discussions on the topic federate. I doubt that's feasible on the Mastodon platform though.

@nolan The point you made regarding algorithmic use - it can be definitely be made as an opt-in kind of feature. I can see the case for it being made.

@nolan

One feature I'd like to see is an ability to follow hashtags or certain keywords.

I contend geeks and everyday folks are overlapping segments with different needs @nolan. #Mastodon could very well serve both.

Alternative cultures already thrive on #Mastodon. Plenty can be done, I as OSS they should naturally be served.

Everyday users are difficult. "Pick an instance" filters most out. Signing up, owning data, and locating others are absolutely crucial.

@nolan I wouldn't be opposed to an algorithm as long as I have a choice. It'd be especially cool if we had the opportunity to tune how suggestions are weighted.

Like on Reddit, I rarely spend more time than checking my front page. Sorted for what's hot is fine but I'm always late to the party. Analogously, if the post has a high probability of being interesting, it'd be nice to see it. But I'm not going to trudge through countless posts hoping because this platform lends itself to real-time.

@nolan Opt-in algorithm, with user-tunable parameters, might be interesting?

Maybe.

@nolan Nowadays when AI is getting stronger it's hard to imagine we will need to keep organizing our shit like we did before.
An open-source algorithm to control my timeline doesn't mean losing control over my feeds. It all depends on how I can control the algorithm.

@nolan oh god please bless fedeverse without any algorithms

@EdwardTorvalds @nolan Oh yes, Mastodon must keep the chronological timeline! 🙏 I'd only like to see the 'algorithms' if they're user-defined and user-controlled. i.e. In the form of optional tools such as extra filters. (Could think of it like "Power Lists") 🤔...

@EdwardTorvalds @nolan Re. Desire to curate, I don't think too many people here on Mastodon care if they're "normal" people 😉 But seriously I see it as a big benefit of Mastodon myself. And it's not like it should be a big issue for newbies because when you start off you'll only be following a few people. And possibly on a small instance too.

@peter @EdwardTorvalds Hmmmm yeah that's an interesting point, that the fediverse is already sorta pre-filtered for newbies thanks to the Local Timeline. That does save some effort with curation.

Re: "normal" people, point taken. 😉 But then again, I'd like the fediverse to grow beyond the tech-savvy and enthusiasts, and I wonder if that means less focus on curation tools and more focus on automation/suggestions.

@peter @nolan @EdwardTorvalds I recently heard the idea of an 'algorithm marketplace' where users can opt for a different model if 'likes & reshares' isn't their thing. The idea of per-user algorithmic systems appeals...

I'd definitely like to see an inversion of the current standard model - if I like something, show me less of it. Challenge my world view; help me fix my cognitive bias ;)

All of this has to be under users control, of course!

@gwmngilfen @EdwardTorvalds @nolan @peter now that'd be cool

@gwmngilfen @EdwardTorvalds @nolan @peter A bit like https://gobo.social which offers tunable algorithms.

@22 @peter @nolan @EdwardTorvalds ooh that's interesting. Seems to rely on other social media accounts though, which I don't have ;)

I've also always liked the idea of an RSS bot (or similar) with Bayesian filtering or machine learning, so I can subscribe to way more than I'll read, and tune my own feed... Maybe I should investigate that a bit more...

@gwmngilfen @EdwardTorvalds @nolan @22 Reminds me that I used to use Zite a lot before it shut down, but I always wanted more "power user" features, like the ability to say "less of this" / "more of that" - but with the insight to see what values that was actually changing under the hood (e.g. ontology weightings) and change them directly if I wanted too. (Been on my "ideas" list to try making something like that for years but never got round to it 😄)

@peter @22 @nolan @EdwardTorvalds if we're wishlisting stuff, can we also get a local AI for labelling email? I'm sure it'd be better than constructing 284746 filter rules :)

@peter @EdwardTorvalds @nolan
cool System for user-defined "alogrith" would probably be like the taskwarrior priority system:
https://taskwarrior.org/docs/priority.html

@nolan Something related I've been considering is the idea of a 'toot proxy' for more powerful filtering. I like to keep things light on social media, avoiding political or politicised subjects; I make good use of manual feed curation and it's also why I'm working on getting regular expression filtering added to Tusky, but my regular expression is already growing into an unmanageable beast. I guess it could even use ML; feed it with some seed terms / topics and let it get to work.

@nolan It could also be opt-in

@nolan I think rather than algorithmic timelines, perhaps better tools to help manage curation.

@nolan how do you see a user oriented algorithmic timeline work?

@nolan the other road is like Friendica has - you can clearly opt for options liked time posted, time commented, etc.

@nolan I think it would probably be OK if it was optional. A certain portion of users would probably rebel, but if it was an option that an individual could turn on or off, that might be cool.

That's all many of the "make Instagram chronological again" people are asking for. They want the option, not necessarily the elimination of the algorithm.

@nolan That kind of works - I'm trying to balance multiple instances myself - but it also feels like I'm managing more social networks, not different facets of one.

@nolan Having accounts on multiple instances is the one ting that really confused me at first (I still only have one account). Is it _almost_ necessary to have multiple? I'm shooting for tech conversation or just daily gripes and such.

@kzisme It's not required, but I kind of like it. I don't always keep up with each account equally well though.

@nolan I'm big on multiple accounts as well. I just wish there was a client that would allow me to specify my "home" instance, and regardless of which one I'm reading, any follows or replies happen from my home account.

@nolan I agree. List/groups are important to me for Twitter and my RSS reader.