We need to differentiate the recommendation algorithm parts of these products from the communication parts. If Facebook is recommending something, that's on them, and if they want to not recommend my content to anyone, so be it; but I should be able to post something seen only by the people who opted in to following me without them getting in the way.
I would take this a step further and say that all social media recommendation algorithms should be publicly reviewable.
The claim is that this algorithms are neutral. But we know many contain ways for the owner to artificially boost preferred content. And algorithms tend to have the biases of their creators in them.
Controlling what people see is an important power and one that should be regulated.
I don't see how this will help. FB et. al. work by shoehorning complex nuances into intentionally crude metrics like "engagement.". If they show the engagement-focused algorithm, that shows they're "just trying to engage users."
What would the algorithm show that isn't already apparent?
Most people that want to see the algorithm want that knowledge to use it for their own ends. They are forgetting about the second-order effects where literally everyone will also be doing that, and then they are gaming each other instead of "The Algorithm".
Yes. The entire point of government is to delegate important functions that cannot be handled correctly by a "free market" to elected officials, which would include appointing regulators. So that would be the system working as intended.
I don't really think there's enough reason for recommendation algorithms to be publicly reviewable, in many cases the algorithm itself is the most important asset of the product, and in a competitive business environment, you can't really force them to disclose it.
What would be ideal is to think of an incentive for these companies to give users an option to disable the use of their algorithm, aka natural flow. That's unfortunately not the trend things are moving, and even services that still have something like that employ dark-patterns to throw user back into their "controlled" timeline (Twitter for example reverts you back "Home" after a few days with a tiny message that is barely noticeable.)
I'm even willing to pay for a feature that "turns off" the algorithm, unfortunately that would never happen because it'd entail these companies admitting that they design their algos not for the benefit of the user but for stickiness and the dreaded "engagement", which shouldn't be a problem in itself, but then it'd quickly become obvious when they're acting sanctimonious.
I agree that the algorithms should be publicly reviewable. But I'm not sure that solves the central issue which is that there's a tension between what's financially good for social networks (algorithms that increase engagement - which disproportionately favors echo chambers, controversial and shallow content, etc.) and what's good for the general public.
Social networks aren't in the business of showing you what's "good for you." They're in the business of showing you what's good for them (e.g. things that will increase your engagement).
There are two (or more) feeds of information on facebook:
- The main one, which only shows posts sent to the "recommend this to people queue" based on facebooks recommendation algorithms.
- Messenger, which directly sends messages to people in a FIFO fashion, which I assume facebook doesn't censor as heavily (though I can't say I've tested).
The fact that the recommendation based stream is more popular is a reflection that facebooks curation is actually useful, but it doesn't mean messenger doesn't exist.
Nor do I see that messenger must exist, facebook isn't morally obligated to provide a "non recommend feed" to you if that isn't the product they want to sell you. Perfectly good alternatives even exist, in the form of "blogs" (say, substack).
I assert that content I post to my timeline and which you see on your feed as my friend shouldn't be subject to Facebook's commentary or opinions. I take strong issue with the claim that to obtain that I have to use Messenger--which is an entirely different content paradigm--and I think it is outright dishonest to try to claim that people posting to timelines means that "the recommendation based stream is more popular".
Facebook wants to recommend content to me from random people: that's their issue, and they should take responsibility for it--whether legally or simply morally--and thereby should have to be careful with what they show people (maybe to the point where they simply can't scale it to the scale they are at automatically! I will not cry over their loss). Facebook also shows me content from my "friends"... I opted in to the content from my friends: if I don't like my friends, I can and should unfriend them, as they were my responsibility, not Facebook's (in the same sense that if I call someone on the phone I don't whine to AT&T about how much they suck). I should not get labels from Facebook telling me my friends' content--content I explicitly wanted to see--is scary or wrong, and they certainly should not be banned from showing it to me.
I think the most egregious category error you are making here is to equate messaging with communication... the entire mechanism of social media to followers is primarily direct communication! We can see how ridiculous this has become by looking at how, even if you have a private account--one random people aren't even allowed to see!--Facebook, Twitter, and Google currently consider it their business what we post there... an account you post things to and which other people follow is identical to a chat channel that happens to maintain a buffer.
(FWIW, to steel man your argument as best I can, "yes": Facebook also applies recommendation algorithms to friend feeds. Twitter I believe still doesn't. Instagram didn't, but then started... AFAIK people don't like it that much, but it might be more profitable for Facebook? Either way: it is helping me reorder content I already curated, and so I don't feel a need to claim that this is something they published. There is simply still a fundamental difference from platforms surfacing "trending" content or suggesting new people or channels I might want to follow / posts I might want to see, and acting as an intermediary for content from people I was directly linked to from people I am communicating with. The former might as well be illegal as far as I am concerned, while the latter needs to be sacrosanct.)
Point of fact first, twitter does filter tweets via a recommendation algorithm [1]. If you troll through HN you should be able to find threads with people complaining about that too... I assume instagram does too but don't use it and haven't verified [edit: It does https://later.com/blog/how-instagram-algorithm-works/].
> Facebook also shows me content from my "friends"... I opted in to the content from my friends: if I don't like my friends, I can and should unfriend them
This is not the service Facebook offers, nor has it ever been. Facebook has forever been about friending everyone you are actually friends with (and then some), and you (or at least the vast vast majority of people) do not want to swallow the entire firehose of content that their facebook friends post. Facebook doesn't want people unfriending people to avoid seeing their content, that's not how the model works - doing so would break all of other facebooks services like acting as a birthday reminder, event planner, and so on and so forth. Even the language is indicative of this, "unfriend" (I don't like you anymore) not "stop showing me these posts" (I don't like what you post).
> the entire mechanism of social media to followers is primarily direct communication!
On the contrary, the entire mechanism of the majority of successful social media companies is curated communication. As it turns out 99% of everything (e.g. posts) is crap, and people go to the sites where less than 99% of what they see is crap.
Types of curation vary, but the existence of it is very consistence. To create some categories: algorithms (facebook, tiktok, youtube, twitter), community upvotes + community moderation (reddit, HN, slashdot), just straight up heavy moderation (many forums, forum.nasaspaceflight.com comes to mind as one that is still going strong). Also worth pointing out that most of these platforms actually mix it up, e.g. youtube, twitter, and facebook all have forms of upvotes, and facebook has forms of community moderation.
But if you want direct communication, that exists. I.e. the previously mentioned facebook messenger (which admittedly doesn't do followers) and substack (which is a competitor of sorts that does).
Direct communication is not what facebook is offering as their main product (but is what they are offering with messenger), it's never been, and I don't see that you have any right to demand that it becomes what their main product does.
That said, SMS carriers filter and block TONS of messages and have been doing it for years. Our startup works in that space so we have to deal with this problem all the time and I can tell you last November if you tried to send a text message with the word "election" in it in an automated fashion, it was most likely getting blocked.
Yes, and now the U.S. mobile telcos are applying a completely subjective "reputation" "brand trust score", which sounds a lot like the Chinese social credit score. Oh, and they're charging a ton more for the privilege of sending a small number of texts, too.
I guess when you engage in price-fixing in the open and call it "setting a standard", then anti-trust rules don't apply..
I promise you there is tons of actual keyword based content filtering going on. We had to build an AI just to detect which keywords different carriers block. Anything socially contentious, like "BLM", "election", "covid" increases the chances your message will get a 30008 error aka carrier-level content blocking. More disturbing is that plenty of carriers will also block messages but report them back to Twilio as delivered.
This starts to get to the issue here. FB is not a carrier (like a telephone company). Their business model isn't built around carrying information or broadcasting it.
FB is a content platform who is about making ad revenue. They are crafted around that and how to make that the most profitable possible.
Another nuance is that due to Facebook post privacy settings, an individual post can be restricted to a limited friend list, and not viewable by the general public. However the "fact checking" appears to apply to both private and public posts.
The fact checking aspect of it is new twist, but Facebook's servers already checks the contents of messages and will disappear messages if the content is deemed bad. It started off as protection against spreading viruses/malware which are unequivocally bad and has evolved from there to include urls to sites with extreme political views and other controversial content.
If everyone involved in the conversation wants to be involved, and no laws are broken, and the platform interferes, then it's censorship.
If there's some kind of algorithm putting the content in front of people who didn't ask for it (e.g. not followers/friends/subscribers/whatever), then you have a point.
Technology changes - magnifies, accelerates, projects, distorts - many social effects. One of those is removing the previously invisible, small cost of spreading an idea.
If two people want to have a conversation or exchange a letter, or one person wants to speak to everyone within earshot or post a sign for people walking by to read, that's one thing. Maybe you say something dumb, and it gets shot down, or maybe it gets propagated, but it needs to have an R0 greater than 1 to endure. If you need some level of capital and agreement to have a publisher run a thousand pamphlets with your idea, that's another, you can leverage previous efforts to amplify weak ideas, but more people are involved and able to provide some sort of sanity filter. On the Internet, the cost to promote an idea is near zero, and algorithms can amplify something dumb but catchy to millions of people in a heartbeat.
It used to require a few seconds of talking per person you wanted to reach, now it costs a few seconds to post a message that might be seen by millions. The difference is minuscule in absolute value (perhaps a monetary value of a few cents) but huge in relative terms (how many percent less than a few cents is zero cents?).
Maintaining policy on censorship while ignoring this massive change in the landscape is shortsighted. Yes, a physical public venue ought not censor someone who wants to talk to other people there, but a digital platform with an audience of millions should think carefully about the effects of messages that their technology amplifies.
You only have two choices, let all information be available regardless of whatever downside there may be, given that at least you still have some control over how to deal with that, or live in a truly Orwellian society in which you have no control over what you know.
The world isn't a black and white place and there are a variety of more nuanced stances between those two extremes. Any amount of censorship doesn't necessarily and immediately devolve into 1984.
Free speech issues are never simple - but their solutions is never found in the extremes. We've already, as a society, compromised free speech to make exceptions for discriminatory speech and violent destabilizing speech - we're already compromising how available all information is and it's necessary to have a functioning society.
Right now the US, specifically you guys - most of the rest of the world is handling this better - has a huge issue with false information around vaccine efficacy and dangers. This issue must be resolved if you want to be included in an open world once again - some level of censorship is going to be required.
Can you substantiate this? It seems pretty loaded, and I'm reminded of that fallacy where you state there are only two extremes possible as outcomes, you doing that on purpose or what?
How to propose there is a middle ground? What everyone imagines is that if we only censor what is reasonable to censor it will be fine.
The fallacy is that we always view this from our personal perspective, yet we will not be the ones who make those choices. We give up that role to someone else.
Who has this role over society has immense power. Some would argue greater than governments themselves. It is only a matter of time before that vector will be exploited. It is in principle the same idea as regulatory capture, yet the incentives for capturing speech are far greater than a typical regulatory body.
Unless I put her account on snooze, I regularly see posts that my (literally) abolish-money/anarcho-communist ex shares, even those from groups which I have repeatedly marked as “block all content from this group”.
So just unfollow her. It sounds like you have some major issues with her personally anyway so how is this a failure by the platform and not just you subjecting yourself to a negative situation?
The stuff she writes directly (rather than liking memes from groups she’s in) is as interesting as any other friend’s posts. Reason I mention her politics is because they’re about as extreme as you can get.
The problem is that Facebook is convinced that I, a British citizen living in Berlin, want to see “Bernie Sanders Dank Memes” (or whatever it was, there are many) even though I’ve clicked on the button labeled “don’t show me ‘Bernie Sanders Dank Memes’“.
The “even though I’ve clicked on the button” is especially egregious, in this context.
Yes, I hate how Facebook (and especially Twitter!) have now chosen to not just show me my friends' posts, but posts they like and respond to. A like has basically become a retweet.
Facebook is a private company that allows you to sign up to use its wholly-owned platform, and it's allowed to censor for whatever clever/asinine reasons it comes up with. Your only recourse is to disengage.
The reason the private-company argument is really tired is that it's simply not something that we take to its logical end in society; the returns of company freedoms are diminished and even counter-productive when a company reaches ultimate freedom to do whatever it wants. A diverse society isn't sustainable if people from different backgrounds don't have the same opportunities to participate in society. This is exactly why, no, you can't only allow whites into your business and you can't ignore the needs of the disabled, among any number of things. It would work excellent under a feudalist system, however.
Likewise, if companies like Facebook get so large and influential that they (and a small number of other NGOs) provide the only meaningful channels of communication between groups, we are dooming freedom of expression if Facebook, Google, or whomever are free to silence you in order to pander to politics and advertisers. Especially not when Facebook works for the federal government and gets tax breaks and subsidies. Your individual rights mean more than the right of a giant corporation to make lots of money and have undue amounts of power.
> In the United States the statement is often heard, that “corporations are private businesses, so they can do whatever they want.” This assertion is particularly false when referring to entities like Facebook, Amazon, Exxon, and Pfizer, because…
> - the phrase goes directly against a basic knowledge of the history of incorporation — corporations were originally designed and granted special legal privileges by government, only because they were expected to serve a pubic good.
> - the phrase goes directly against the dictionary definition, and investment industry terminology — a public company is defined as a company whose shares are traded freely on a stock exchange, hence the term IPO (initial public offering).
> - the phrase ignores real-world government involvement—many large corporations are state and federally sponsored (e.g. subsidized, bailed out, and given perks), by money which ultimately comes from the tax-paying public.
> In reality, all big corporations are some combination of state-chartered, publicly-traded, and government-sponsored. By definition many are public companies, while others have complicated hybrid characteristics.
Not being allowed to do things is called being regulated. So Facebook shouldn't be regulated in the same way that ATT is because they currently aren't being regulated in the same way that ATT is?