#BlueSky and #Fediverse developers could learn a lot from each other: right now it feels like they could learn a lot from here on moderation…but seeing their new “anyone can create and use and share their own algorithm” - that’s great and Fedi should immediately learn from, implement and improve on that idea…
@mike@tchambers Even in the client, folks still object to indexing & processing of posts without explicit consent.
Personally, I’m coming to an opinion that if you stick a message in a bottle and toss it into a sea designed to replicate bottled messages, you’ve consented to what folks do with bottles that wash up on their shore. But, there is angry disagreement with this notion.
@lmorchard@tchambers@mike I think reasonable efforts should be made to find compromises acceptable to all, but a society as large as the fediverse is too large to function with absolute consensus. It's literally impossible.
@lmorchard@tchambers I've used the anology asking for an invisibility cloak in the town square. I simply don't understand pretending to have ownership over something you've voluntarily entered into the public domain.
@mike@lmorchard@tchambers I don’t think it’s visibility, necessarily. It’s not knowing what would be done with it, for how long and to what end.
I am don’t mind my posts being read by everyone on the fedi, where I have settings to manage reach & retention. I do mind the idea of someone taking my posts from this setting or using them as a resource to drive their own objectives.
@mike@lmorchard@tchambers I’m not tossing bottles into the sea. I’m floating self-destructing messages into a small pond with blocks and parameters I manage.
@bomkatt@lmorchard@tchambers Like being quoted on the record or off the record. Somehow I don't see this working on a broadcast medium like social networking though.
it may be less of a technical problem than a governance-and-approach one? We’ve built so many opt-out models, like those GDPR banners, and can’t-avoid models, like Facebook’s shadow accounts.
I’d also question if social media is really broadcast. It can and is used that way, but other people tend to use it more like a forum. Is it more like a blog or messaging app?
@bomkatt@mike@tchambers FWIW, my own objectives in mind are personal. That is, I want to tame the firehose a bit for myself and see & find things sent my way.
I’m uninterested in republishing things outside boosts or letting other folks use my personal feed or search an index of what landed in my client.
Personality I think that should be okay, but there is loud disagreement. Which, fine, but also I wouldn’t want to bring a fight to my instance’s door over it
@lmorchard@mike@tchambers I have zero qualms about something like that, even if it was used by lots of people.
I apparently hear index and thing big tech, not indie web and individual people building things for themselves. “How can we make money from this?” instead of “you know, I could make this better if…”
Yeah, last thing I want to do is “make money” from something like this in a big tech sense - except maybe conceivably if I could make a decent livelihood for my family building personal-scale tools. But I’m nowhere near that either heh
@mike@lmorchard@tchambers does mastodon (for example) have terms of use which license sharing as YouTube and Twitter have? I’m not sure if fair use/fair dealing covers copyright here. Q for lawyer obviously
@lmorchard@mike@tchambers At some level you’re probably never going to make the people who demand to have a public conversation with the utmost privacy happy
@lmorchard@mike@tchambers@JamesGleick the concepts of public and private are long established. If I’m in public people can look at me, photograph me, listen to me - all without my ‘consent’ because that’s what public means. Conversely I have legal protections against those things in private because that’s what private means. I don’t see why we should go into contortions trying to reinvent the meanings of public and private because it’s digital.
@lmorchard@mike@tchambers This discussion made me think of the following discussion between @MartinEscardo, @johncarlosbaez, and others about how mastodon sometimes seems to fail to surface content to those interested in seeing it (in this case specifically on category theory).
@johncarlosbaez@lmorchard@mike@tchambers@MartinEscardo with feeds being primarily driven by chronology, I think it's just too easy to miss things if you follow more than a few people. There are potentially ways to cope with that somewhat, using lists and hashtags, but I don't think you'll ever get enough people to do that consistently to promote conversation the way one might like.
Based on past experience on social media, people are rightly apprehensive about algorithms, but perhaps they have perhaps over-corrected. It seems plausible that transparent algorithms under the control of users themselves could be a good thing. Otherwise we are effectively stuck using a very poor algorithm to decide what we see.
@lmorchard@mike@tchambers Consider the thoughtful concerns of reasonable people and try to address them.
Unreasonable people can be ignored.
You are not obligated to acquiesce to every single person’s niche complaint. Indeed, it’s impossible to do so, and to the extent it has already been tried has held back the usability of this network.
@lmorchard@mike@tchambers I do find it odd that folks are okay with "reverse chronological of folks I choose” as an algo, and "boosts of folks I follow," but draw the line when those folks effectively make clear what their rules for bossing are and stick to them. FWIW, over on #LawFedi we've had luck with a roll-our-own algo via @icymi_law It's an account that acts as a community algo. See https://icymilaw.org/about/#bot
I’m pretty confident we could find a way to give people control over this.
To follow the analogy: to make some bottles opaque, some only translucent, some crystal clear; and further, some that can be copied infinitely and some only a few times.
@DavidM_yeg@tchambers@lmorchard You don't need opt in or opt out. If you implement an algorithmic curated thread on the client. It's no different that a filter or a list. It simply applies criteria you select to posts already coming to you.
That said, I’ve already been playing with code that ingests a few days of toots from the steaming API on a mastodon instance into a sqlite db.
From there I can get my own trending links, do some searches, more heavily weight folks from whom I don’t hear as often - all kinds of interesting stuff
There will come a time, possibly sooner than later, when someone will do their own API that scrapes the federated stream for hashtags and keywords ~ ostensibly for academic purposes.
I'm not sure anyone could stop it, let alone should they stop it. A lot of researchers' modeling of society involves what is happening on social media
@tchambers@DavidM_yeg@lmorchard My sense is that even if it picks from your federated feed it's still content that had the chance of reaching your eyes. All the curation script is doing is catching it and saving it for later consumption.
@tchambers@mike it is one of the big reasons it seems ppl don’t like Mastodon. I like it but I do still find that twitter- before it went all wrong- worked better for me professionally.
@Htaggert@tchambers If you want to promote something, even a benign something like your instance, you have to boost it several times a day to make sure it gets in front of everyone. That's something an algorithmic feed might solve.
@mike@Htaggert@tchambers so what. Shows you care and put the work in. I want a human on the other side of the line. Now with generative AI in our lives, precisely this is ever more important
@mike@tchambers All reasonably modern versions of #Mastodon have an "explore" tab where you can see the posts that are the most popular with that instance's users. Download the content for something like 100 instances, de-dup the list, and then figure out the topic (either by analyzing hashtags or by analyzing the nouns within the toot). Once you get that far, you can analyze the users most recent toots by noun or hashtag and you have a recommendation list. I wouldn't call it easy but 🤷♂️.
@mike@tchambers What if you had something on the recommended toot akin to "I am not interested in cheese graters" (if the identified noun or hashtag was "cheese grater") and something within the settings where you can explicitly say that you are interested in cheese graters? If you are always angry tooting about them but you definitively don't want to see any new toots about them, that should do it?
@mike@tchambers ... I wouldn't want to see an instance implement this, but it seems pretty implementable if a mastodon client wanted to add something like that.
as long as we control it, otherwise it functions like advertising. I don't want a fedi reddit alternative, I want a view for my regular feed that lets me surface posts and replies however I choose.
sorry, I view this medium as being about attention, that we collectively finally get to decide what art, information and ideas deserve to go viral in the culture. not the owners of big media or advertisers. that's revolutionary and I don't want to lose it before it's really begun.
@wjmaggos@tchambers Currated by an algorithm where you set the parameters, to take content out of your feeds and present it to you as per your conditions. That's why it needs to be done in the client, so no central authority can control the algorithm.
that would be fine. I guess what I assume and fear would happen, is that somehow one becomes the default for new users, cause people say this place is too boring without it etc. then the creator is somehow able to roll out modifications automatically and they get a lot of control over what most people see. we still have the option not to use it, but most people don't care. the algo creator says it's "safer" etc. it becomes like Google's pagerank.
@SymbolicCity if it were only relating to public posts, and only to those federating and at least potentially available to you already, via hashtag search, etc, then what fedi principles are at issue?
@SymbolicCity I am equally unclear on those items - and I’m a concerned that if they are doing what you suggest it simply may cause other problems there….
@tchambers I suspect their original goal was to avoid moderation by having users “label” themselves (and maybe each other?) and tools to ignore labels they didn’t want to see.
That approach doesn’t seem to be holding up so well.
@ramsey@tchambers It feels shockingly naïve to think anyone would ever think that would work. It there is one thing we've learned, it's that 0.01% of the population is willing to do nearly anything to grief, and their impact goes well beyond their representation.
"Self moderation" is meaningless, to do it properly requires serous "oh no you don't" tools, preferably ones which can grow cooperatively through the community.
I’ve felt their approach has been very naïve since I first heard about it. It’s as if they’ve collectively decided not to learn from any social media successes and failures in the past and have decided to start from scratch.
@ramsey@tchambers But back to Tim's original point (which I agree with!) which is that both sides have done some good work, we need to cross pollinate and get something better for everyone (e.g. that super-list concept from #Bluesky)
@scottjenson@tchambers I’m not saying they’re making all-around bad decisions. They’re very smart people, and they’re extremely approachable and transparent.
@tchambers how do users make them? *key software have antennas where you can include/exclude words and users so they are more than just following hashtags but I'm clueless about what you have to do on BS to create one of those lists.
@sparkit@tchambers Users don’t make them. Techbros make then and the users subscribe to them.
A "custom" feed is simply a server built by a third party with an open port that answer to a certain bluesky API that gives you a feed of post IDs in a certain order. Your bluesky client uses those IDs to retrieve the message from the big bluesky graph server (BGS).
So basically your just trusting an unknown software running on an unknown server to give you good posts and you have no way to check what they are doing.
Some of the current and most popular feeds on bluesky are already close source and "will-opensource later". People are subscribing to those feeds and only the creators know how they work. Even if they open source it, you have no way to tell that the code running is effectively the one they published on github.
So shadowbanning is possible, corruption to the feed creator is possible, secret advertising or boosting for money is possible.
It seems there are many things they are supposed to do "later". Even if there is really something worth doing with AT I wonder how much time it will take to make it really available to the public with federation and all. Years?
@tchambers Yeah, with emphasis on clients rather than servers. I still consider https://news.feedseer.com to be an excellent approach and would love for eg. @decius and @ivory to work on embedding it in the native experience
@tchambers@decius Also: #BlueSky, #Fediverseand the #IndieWeb can learn from one another. The IndieWeb’s focus on plurality and composition of independent standards is worse from a PR perspective but better from a resiliency and innovation perspective
Add comment