Can we create and use social networking algorithms on the Fediverse that serve our own personal goals, such as more and deeper personal relationships, career advancement, and more participative citizenship?
-
@evan Yes we can create silos!
@gam3 every account is a silo of one!
-
@evan is this polling a policy/cultural angle or a technical one?
@Spoofer3 Could you unpack that question? I'm not sure what you're asking. Are you asking if *I'm* asking whether this is technically possible, ethically acceptable, or politically feasible, or some combination of the three? Or are you asking something else?
-
@evan The “buts” for me is they have to be completely optional and, ideally, controllable at the client end. It would be interesting to set up a side feed filtering for specific content and such however I’d want that to be something I configured.
-
@Spoofer3 Could you unpack that question? I'm not sure what you're asking. Are you asking if *I'm* asking whether this is technically possible, ethically acceptable, or politically feasible, or some combination of the three? Or are you asking something else?
@Spoofer3 If you are asking that, the question is all three, plus any other angle you can think of.
-
@evan My answer is a "No" without a but. Any form of classification or algorithmic unpacking reflects intent. Intent with bias. That cannot be seen or changed. No matter how much we try. That's at least my lesson based on many years of experience. And yes, I am fully aware that my personal timeline also reflects a lot of bias based on the people I follow. But that is ultimately a good thing in a world of disinformation and clickbait.
-
@evan My answer is a "No" without a but. Any form of classification or algorithmic unpacking reflects intent. Intent with bias. That cannot be seen or changed. No matter how much we try. That's at least my lesson based on many years of experience. And yes, I am fully aware that my personal timeline also reflects a lot of bias based on the people I follow. But that is ultimately a good thing in a world of disinformation and clickbait.
@jwildeboer What if the bias is on the users' part?
For example, if I create a list in Mastodon with the accounts of my close friends, that is an attempt on my part to stay better connected to those people and not lose their updates in the flow of the hundreds of other accounts I follow.
This is biased -- it assumes that a clearer view of my friends' posts will improve our relationships, and it also assumes that I know who my real friends are. But that's up to me to decide.
-
@jwildeboer What if the bias is on the users' part?
For example, if I create a list in Mastodon with the accounts of my close friends, that is an attempt on my part to stay better connected to those people and not lose their updates in the flow of the hundreds of other accounts I follow.
This is biased -- it assumes that a clearer view of my friends' posts will improve our relationships, and it also assumes that I know who my real friends are. But that's up to me to decide.
@evan It's the classical dilemma between "knowing it all, overwhelm me" and "what do my peers think". I don't have a simple solution. But I do think that a conscious decision to look at the firehose and a "filtered" personal timeline helps. Which should be account based, not algorithmic. And no, a simple group approach might not be siffcicent. Le sigh.
-
@evan yes, but... who is "we"?
If I want to write a client that reshuffles 24h of posts from my feed for some purpose I desire, I can't see why that isn't OK. We the individual users.
If my instance imposes nontransparent, complicated, algorithms not under user control..... well that's unacceptable. We the software designers and operators.
-
@evan It's the classical dilemma between "knowing it all, overwhelm me" and "what do my peers think". I don't have a simple solution. But I do think that a conscious decision to look at the firehose and a "filtered" personal timeline helps. Which should be account based, not algorithmic. And no, a simple group approach might not be siffcicent. Le sigh.
@jwildeboer that's an algorithm!
-
@evan yes, but... who is "we"?
If I want to write a client that reshuffles 24h of posts from my feed for some purpose I desire, I can't see why that isn't OK. We the individual users.
If my instance imposes nontransparent, complicated, algorithms not under user control..... well that's unacceptable. We the software designers and operators.
@johnefrancis such an interesting point. For much on the Fediverse I expect a lot of alignment of interests between users and developers, though.
. -
@evan an algorithm is basically “arrange my timeline according to this heuristic” and I see no reason why, if that heuristic were well-defined by the user, it couldn’t be specific to their client filtering. What calculates the result of that heuristic could be the server, so if you want ultimate control you run your own server, and rely on community-run calculations if you use a larger network, but never a centralized algorithm or scoring calculator. That is how I would distribute it.
-
@jwildeboer that's an algorithm!
@evan Is it? When you see a simple decision based on "it's everyone" and "It's not everyone" as an algorithm, I guess it is. I still call it a filter and don't see an algorithm.
-
@evan My answer is a "No" without a but. Any form of classification or algorithmic unpacking reflects intent. Intent with bias. That cannot be seen or changed. No matter how much we try. That's at least my lesson based on many years of experience. And yes, I am fully aware that my personal timeline also reflects a lot of bias based on the people I follow. But that is ultimately a good thing in a world of disinformation and clickbait.
And what's wrong with having transparent bias and users' choices about which one to use ?
-
And what's wrong with having transparent bias and users' choices about which one to use ?
-
A full documentation of how the algorithm operates, and open discussion about the biases it entails.
-
@evan@cosocial.ca yes, but we must keep being vigilant that it's still dangerous in terms of what we can uninentionally inflict on ourselves. An echo chamber or rage maximizing algorithm is worse when controlled by an external entity with specific financial goals and complete disregard for human wellbeing. It is however still bad (even if a bit less) when self imposed by a person. We still think of social interactions the way we did pre-internet and we certainly have the emotional reactions to social interactions that we elvoled to have for (tens of?) thousands of years before the internet.
I'm assuming what someone already mentioned here, that said algorithms will be somehow entirely user controlled. Anything else is bound to end up as what we have in corporate social media. -
@evan I said Yes instead of "Yes, but..." because the but is essentially "you should probably understand what the algorithm does and you need to be responsible for any effects of using it, including on others"...
But to me that's like "Should you be alive?" with an answer of "Yes, but you should obey the golden rule" which is really "Yes".
-
@evan I want algorithm like old school art platform features+its algorithm (DeviantArt, Pixiv, etc).
Sort by newest or popularity are never a viable solution for art community.
I remember seeing my DeviantArt feed, and discovering sketch, photography, cosplay, calligraphy, origami, in one place. Some recommendations are literally 8 years old post.
Without that system, art community will turn into trend chasing, maximalizing for engagement instead of personal uniqueness. This already happens on Twitter, Bluesky, and Instagram.
Even if Pixelfed had the algorithm, it is not a solution, as it was never meant to accomodate variation of post like proper art platform.
