Instagram and Threads moderation is out of control
theverge.com
theverge.com
The core issue with it is the algorithm. There's nothing inherently "incorrect" about 99% of the stuff I see on there but it clearly influences my mind over time regardless, it's just a bad information diet.
Messages will evolve according to the conditions of the medium that carries them; similarly to how biological organisms either adapt and thrive or fail according to the constraints of their environment, there exist selective pressures that determine, ultimately, what content is promoted and thrives according to the whims of the social media algorithm. "The medium is the message," and the predominant message of social media is short-form brainrot. The need to publish "content" on a regular, daily, or even live-streamed 24/7 basis incentivizes the production of quick takes, short videos, and soundbytes instead of slow, deliberative, measured thoughts that may take months, years, or longer to assemble into a literate medium. The latter form of content is actively selected against in a social media environment; thus, the proliferation of groupthink and stupidity.
I find myself thinking about nutrition a lot after reading this a number of years ago, and trying to be careful about my intake. HN is pretty much my only (vice) news source left at this point.
> All beings subsist on nutriment: edible food, sense-impressions, volitions, and consciousness. [src](https://www.accesstoinsight.org/lib/authors/nyanaponika/whee...)
I am low-key thinking about that problem all the time. The more I think the less I believe that it is even surmountable.
From that vantage point it sure looks like Meta does an incredible, excellent top-notch job on that. That's the failed 0.001% that are notoriously and hilariously wrong. While it sucks and hits the pain points, we can all agree it's more or less inevitable. No system can be 100% perfect. And yes, I have been on the receiving end of that lance multiple times and yes it sucks.
As people move on to other apps, scams and fake content on "classic" fb seems to still be reaching boomers and people in other countries. As usual, I do not get the impression anyone at meta cares.
The standard techie response for how to solve this problem is to get someone internal at meta to escalate. Otherwise known as not a solution to the problem. My experience doing this is that internal people handling such tickets are actually not very bright and your internal contact needs to repeatedly push back at people who want to close tickets with no action.
For context, I’m based on the West Coast, which might influence my experience. Just wanted to share my perspective!
My experience on Instagram: I press not interested. It keeps showing exact same thing over and over.
At the scale of Instagram you're basically expecting them to act as a censor for billions of people. It's like imagining an agent on every street corner preventing people from discussing what they like in the pub. You might be able to influence the direction, and you might be able to use models to blanket ban certain things, but real moderation? You'd need to employ hundreds of millions.
Finally unblocked after complaining on twitter about them
Especially with so many well-funded, state sponsored actors using LLMs to sow division as we've seen so often on Twitter recently.
The current system, where you can vote in primaries, and where electors reflect their states’ vote, is just a few decades old.
Penalties for “faithless electors” are somewhat recent and only exist in states which passed laws to give themselves that power.
the blatant, sneering disdain for lowly commoners that supposed proponents of "democracy" are continually unable to contain for even a few paragraphs at a time will somehow never fail to shock me
Democracy cannot work without freedom of speech, and I would argue anonymous freedom of speech. Block bots, not people.
As we're hearing more and more from the large tech companies, this is a hard line to walk. How do you identify what's a bot as they become more indistinguishable from humans? Apparently identity verification and/or remote attestation, both of erode the idea of anonymous freedom of speech.
Whether one believes them is another question.
Now that I think of it, maybe this can be done with TLS certs, although I don't know any CA that can take anonymous payment.
Not to mention that this would do nothing to stop the problem of bots, as bad actors will simply do what they do now and buy/hack verified accounts.
You could make a system like this based on IRL contacts but that would potentially compromise your privacy even more. It's also easier to exploit in some ways than a simple payment.
No it doesn't, unless you're going to somehow limit the # of "tokens" per human, which seems to me to be impossible if the goal here is anonynimity.
> then still one worries about state actors and deep pockets.
You and OP seem to assume that driving the cost up will scale the cost of an attack at the same rate - it won't. These actors will just commit more crimes in order to acquire more "tokens" - either phishing/hacking them, or stealing from distributors, or...
At the same time, you'll be making it difficult/impossible for poor people to access the Internet freely, further widening income and class inequality both on the Internet and likely in reality too.
"Buy an NFT to access the Internet anonymously" is a terrible idea in every conceivable way.
The intent is not to limit the number of tokens per human. The intent is to establish confidence that an account is real by allowing tokens to be associated with it. Additionally, the system would do well to have a database of all accounts associated with a single token. If someone wants to have 10 accounts with 10 separate tokens, that is fine. At least you know it's probably not a bot farm running thousands of fake accounts, because nobody could afford it.
>You and OP seem to assume that driving the cost up will scale the cost of an attack at the same rate - it won't. These actors will just commit more crimes in order to acquire more "tokens" - either phishing/hacking them, or stealing from distributors, or...
Increasing cost does limit the scale of attacks. You might as well be saying that the high price of drugs is not a limit to the scale of drug abuse because people will steal them lol. Committing more crimes is not free of cost. It requires time, opportunity, and expertise. I think a system like I have described can be used to track the tokens and prevent abuse, while allowing people to obtain them anonymously. Every crime committed adds a trail of evidence that can be used to uncover abuse.
>At the same time, you'll be making it difficult/impossible for poor people to access the Internet freely, further widening income and class inequality both on the Internet and likely in reality too.
No, that is all BS. All this means is that they may have to pay more to be anonymous. People of all social classes already have to pay more for guaranteed anonymity, whether it be through buying additional devices, VPN subscriptions, or travel costs as they go to Internet cafes. If you are poor, you still have the option of being less anonymous. We're talking about being provably anonymous here and instilling confidence in one being a real person here, which is a distinctly different problem from being online in general.
...I can't even respond to this.
> system like I have described can be used to track the tokens and prevent abuse, while allowing people to obtain them anonymously.
How do you track the tokens to prevent abuse while ensuring anonymity?
OK it's not a great example but what you wrote is an example of arguing that exceptional cases dominate the whole situation. Like "It's stupid to lock your door because a criminal can just break your door/wall to get in."
>How do you track the tokens to prevent abuse while ensuring anonymity?
The same company that sells the tokens can cooperate with any client that uses them to validate accounts to maintain a database of where the tokens are used and how many times they are used. That would allow you to disconnect accounts from each other as much as you like, but you could not register 100 accounts with a single token without raising some eyebrows. It would be on you to not connect the token to your real identity, if you don't want to be identified by that token.
Even if it was 99.99% accurate there would still be thousands of false positives positives/negatives like you describe.