Meta said it based the $1.4 trillion figure, which is nearly as large as the company’s entire market cap, on how the attorneys general of California, Colorado, Kentucky and New Jersey have argued p…
There are always algorithms. You literally can’t make software without them. As far as algorithms for choosing which content is displayed, Lemmy still has to have one of those to weight votes vs. age. Is a post that’s +50/-0 from five minutes ago the best to show you, or the one that’s +400/-100 from an hour ago, or the one that’s +1000/-100 from last year?
I’m going to hazard a guess that most of us here know what an algorithm literally is. Words can take on new meanings, so the term has become a common shorthand for big-tech’s approach to user experience. Not just to curation of feeds, either, but to the myriad contrived, fucked up methods they’ve concocted for the sole purpose of leading their users into vast bland-scapes full of personalized echo chambers, teeming with bots – which just so happen to be perfect environments in which to effectively milk users of their attention by cycling their emotions and getting them to engage in frequent, lengthy scroll sessions. Just sat there looking at mountains of slop and occasionally clicking on something, all so that people who already have too much can make a few more bucks.
Your example of which vote-sorting algorithm to use is a logistical challenge of deciding which content to deliver, without regard to what that content is. The question is simply: how do we most efficiently deliver and rank the content the community is telling us it values? Basically, how do we democratise the decision of what to boost?
That is in stark fundamental opposition to “The Algorithm” of Facebook, Reddit, et al. who seek to suppress certain content, boost other content, and group users into perfect compartments, where they’ll be endlessly manipulated into chasing dopamine, guaranteed to consume an endless firehose of lowest common denominator content, and if we’re lucky they’ll click on ads.
Machines that perform the conversion of human attention into cash are what their actual customers are paying them to produce. These are the unholy apparatuses consisting of dark patterns paired with machine learning, and now, slathered on top of what was already a massive shitshow, what is referred to as “AI” has supercharged these efforts.
Fueled by sweet sweet ad revenue and all that juicy personal data and telemetry “insights”, their purpose is to sort people. The Fediverse has no such mandate - the same pressures just obviously don’t exist here. I assume the vast majority of us are here because we’ve rejected the (big “A”) Algorithm. The Fediverse is simply a bunch of (mostly) chill people who decided to get together and create a (mostly) pleasant and self-governing little oasis for each other to land on, and maybe even make some genuine human connections. The algorithms making this run are downright quotidian, utilitarian necessities which prioritize organic growth, and center the human by getting out of the way. It’s unrecognizable compared to what’s under the hood of meta and friends.
And how do you quantitatively differentiate between “sorting posts” and “sorting posts, but evil”? Moving a post up the list is by definition prioritizing it. Moving another down the list is suppressing that one. Which posts get which treatment is governed at least in part by dopamine, since people are going to make pull requests based on how they want to use the site.
There are a couple of fundamental properties of the Fediverse that give it decently robust immunity against the “evil” shit that either has been, or will be, the downfall of all the traditional closed-source social media entities.
Because it’s FOSS, there’s transparency into the way internal things are supposed to work, and everything that happens on here, from mod actions down to the level of individual votes can be tracked and audited.
AFAIK, the actual sorting/ranking of your feed happens on the client side (on the device you’re browsing from). And is strictly for user quality of life. Because those are the views the bulk of users like to see. It may even differ depending on your client, but they seem to operate more or less the same.
The server-side has no reason to give different weights to individual posts/comms/users, nor any mechanism with which to do so. The incentive for someone to unilaterally decide what users see on this platform simply doesn’t exist… I’m sure it’s targeted by individual bad actors all the time, but the very nature of this platform makes it next to impossible for any one person or entity to gain enough leverage to rise above nuisance-level. There’s really nowhere on here to hide malicious intent long enough to accomplish a meaningful level of funny business before the network cuts you out or grows around.
The instance database of course stores the communities, the actual content of each post and comment, associates everything with the relevant users, keeps track of timestamps and the individual vote counts. Most everything beyond that gets layered on by whatever you’re using as a client.
Given the utter simplicity of the ActivityPub protocol and the ability of any interested party to gain granular visibility on everything down to the level of who voted which way on what comment? It seems to follow that anything falling outside the pattern of those parameters would raise immediate suspicion and get dealt with pretty quickly.
Advertisers and corporate types are just not particularly attracted here. The prevailing incentives just aren’t compatible with their aims.
I suppose you could say “well, what if the devs have a change of heart and go in a different direction, go rogue or start walling things off?” Whatever project that happened to would get forked the instant a whiff of impropriety was detected. The fork would probably end up better than whatever it replaced!
If a rogue instance started being a little too weird it would get dealt with and defederated just as swiftly by a large mass of federated instances, isolating it.
There are always algorithms. You literally can’t make software without them. As far as algorithms for choosing which content is displayed, Lemmy still has to have one of those to weight votes vs. age. Is a post that’s +50/-0 from five minutes ago the best to show you, or the one that’s +400/-100 from an hour ago, or the one that’s +1000/-100 from last year?
I’m going to hazard a guess that most of us here know what an algorithm literally is. Words can take on new meanings, so the term has become a common shorthand for big-tech’s approach to user experience. Not just to curation of feeds, either, but to the myriad contrived, fucked up methods they’ve concocted for the sole purpose of leading their users into vast bland-scapes full of personalized echo chambers, teeming with bots – which just so happen to be perfect environments in which to effectively milk users of their attention by cycling their emotions and getting them to engage in frequent, lengthy scroll sessions. Just sat there looking at mountains of slop and occasionally clicking on something, all so that people who already have too much can make a few more bucks.
Your example of which vote-sorting algorithm to use is a logistical challenge of deciding which content to deliver, without regard to what that content is. The question is simply: how do we most efficiently deliver and rank the content the community is telling us it values? Basically, how do we democratise the decision of what to boost?
That is in stark fundamental opposition to “The Algorithm” of Facebook, Reddit, et al. who seek to suppress certain content, boost other content, and group users into perfect compartments, where they’ll be endlessly manipulated into chasing dopamine, guaranteed to consume an endless firehose of lowest common denominator content, and if we’re lucky they’ll click on ads.
Machines that perform the conversion of human attention into cash are what their actual customers are paying them to produce. These are the unholy apparatuses consisting of dark patterns paired with machine learning, and now, slathered on top of what was already a massive shitshow, what is referred to as “AI” has supercharged these efforts.
Fueled by sweet sweet ad revenue and all that juicy personal data and telemetry “insights”, their purpose is to sort people. The Fediverse has no such mandate - the same pressures just obviously don’t exist here. I assume the vast majority of us are here because we’ve rejected the (big “A”) Algorithm. The Fediverse is simply a bunch of (mostly) chill people who decided to get together and create a (mostly) pleasant and self-governing little oasis for each other to land on, and maybe even make some genuine human connections. The algorithms making this run are downright quotidian, utilitarian necessities which prioritize organic growth, and center the human by getting out of the way. It’s unrecognizable compared to what’s under the hood of meta and friends.
And how do you quantitatively differentiate between “sorting posts” and “sorting posts, but evil”? Moving a post up the list is by definition prioritizing it. Moving another down the list is suppressing that one. Which posts get which treatment is governed at least in part by dopamine, since people are going to make pull requests based on how they want to use the site.
There are a couple of fundamental properties of the Fediverse that give it decently robust immunity against the “evil” shit that either has been, or will be, the downfall of all the traditional closed-source social media entities.
Because it’s FOSS, there’s transparency into the way internal things are supposed to work, and everything that happens on here, from mod actions down to the level of individual votes can be tracked and audited.
AFAIK, the actual sorting/ranking of your feed happens on the client side (on the device you’re browsing from). And is strictly for user quality of life. Because those are the views the bulk of users like to see. It may even differ depending on your client, but they seem to operate more or less the same.
The server-side has no reason to give different weights to individual posts/comms/users, nor any mechanism with which to do so. The incentive for someone to unilaterally decide what users see on this platform simply doesn’t exist… I’m sure it’s targeted by individual bad actors all the time, but the very nature of this platform makes it next to impossible for any one person or entity to gain enough leverage to rise above nuisance-level. There’s really nowhere on here to hide malicious intent long enough to accomplish a meaningful level of funny business before the network cuts you out or grows around.
The instance database of course stores the communities, the actual content of each post and comment, associates everything with the relevant users, keeps track of timestamps and the individual vote counts. Most everything beyond that gets layered on by whatever you’re using as a client.
Given the utter simplicity of the ActivityPub protocol and the ability of any interested party to gain granular visibility on everything down to the level of who voted which way on what comment? It seems to follow that anything falling outside the pattern of those parameters would raise immediate suspicion and get dealt with pretty quickly.
Advertisers and corporate types are just not particularly attracted here. The prevailing incentives just aren’t compatible with their aims.
I suppose you could say “well, what if the devs have a change of heart and go in a different direction, go rogue or start walling things off?” Whatever project that happened to would get forked the instant a whiff of impropriety was detected. The fork would probably end up better than whatever it replaced!
If a rogue instance started being a little too weird it would get dealt with and defederated just as swiftly by a large mass of federated instances, isolating it.