Mats retweeted
🚨NEW: The Online Safety Act has been a failure, according to a new report from Ofcom.
Ofcom have found "no significant change" in children encountering harmful content since the law came into effect last July. This is despite the signifigant threats to free speech and privacy.
Literally the only "reckoning" anyone in social media faced over the moderation controversies of the late 2010s/early 2020s was us ex-Twitter employees + a few axed teams elsewhere (eg Meta fact checkers).
And it was because of over-moderation, not the other way around.
Tech's approach to customer support (or lack thereof) is essentially a civilizational regression to feudal norms:
Your only real chance of relief is that a lord, addressed under a regnal name ("Elon", "Sundar", "Zuck"), personally hears your petition and mercifully grants it.
Last year, I covered that Facebook allowed mass abuse/doxing rings ("are we dating the same guy?") and that Apple Trust & Safety failed to enforce their own rules against the Tea app.
Basically the same logic applies here.
Replying to @Mat_IRL
At face value, Tea is in violation of Apple's App Review Guidelines.
Furthermore, combining gender-based exclusion with the collection of sensitive data seems both dubious policy-wise and, frankly, it's a recipe for disaster.
Apple T&S should enforce their rules.
Not only is the supposed vilain entirely right, but one can only wish that the tech platforms had retained such a principled approach throughout the COVID/J6 era.
The reality was underwhelming and disappointing in comparison.
Also, I'm disappointed Aaron Sorkin didn't take me up on my offer to model of his "vilains" after me 😔
Replying to @Mat_IRL
If Aaron Sorkin needs someone to consult for inspiration on the archetypal White-male-Tech-Bro holdout who somehow inadvertently caused J6 by not wanting to deplatform half the US population (one assumes this is going to be the core thesis of the movie), I'm available.
The other prediction (very common across industry circles at the time), which also turned out to be wrong, is that Twitter would find its way back to the old content moderation policies:
Mats retweeted
AI Safety has killed more people than AI
This stuff is totally ridiculous. Kill "all" humans. How about 1 million humans? 10 million humans? 100 million humans? what are the odds of that? how about a single human? this is the most morally deformed and confused, consequentialist industry in human history, which speaks about a sincere belief in extinction level events in a flippant, exclamation-marked quote tweet. there should be zero tolerance for this.
Mats retweeted
Censorship can be self-reinforcing, because criticism of censorship itself is censored. An example comes from today's premier liberal democracy, the United Kingdom, where reading criticism of the Online Safety Act requires giving up your ID.
Once more, there is no reason for X to tolerate this content. The implications are clear enough for a moderator to act on.
Elon, who sometimes complains of the consequences of radical rhetoric on his personal safety, can solve this at any time with a single message to his guys.
If you ever ask yourself "is this new law really about children safeguarding?":
It is worth considering they are coming partly from those who --having failed with their earlier 'anti-misinformation' rhetoric a few years ago-- have pivoted to a new rationale for online regulation
Mats retweeted
Recovering crack addict Hunter Biden says that he has no memory of dropping off his laptop at John Paul Mac Isaac's repair store.*
Therefore, Biden concludes, the hard drive must have come from Rudy Giuliani, who sourced it in Ukraine in 2018 through a Russian GRU agent.
Tucker Carlson: "I believe that."
--
* The FBI matched that laptop to him by serial number in open court.
Mats retweeted
The mass censorship campaign chronicled in this essay was personal for some of us in right-of-center media. For me, it meant getting locked out of my X (then Twitter account) for more than a year.
Why? Because in March 2022 I posted that Rachel Levine, the second-highest ranking official in Biden's HHS, who dresses as a woman and calls himself a "trans woman," is in fact a man, and always will be.
At the time, Twitter labeled that "hateful conduct" and locked me out of my account. I really only use the account in a professional capacity, as a senior editor at @FDRLST. So getting locked out wasn't some small thing.
But in order to unlock my account, Twitter wanted me to delete the post. According to Twitter, doing so would be an acknowledgment that I had engaged in hateful conduct, and I could get my account back.
I refused. More than a year went by, and I created a anon account just to be able to monitor the news cycle and easily open links on Signal chats.
Eventually, Elon Musk bought Twitter and someone I know who had access to him personally intervened on my behalf. I never got any notice, but one day I tried to open my account and it was unlocked.
There was never any acknowledgement from Twitter (now X) that I my account had even been locked or that any action had been taken.
This is how the censorship regime works. You often don't know you're being censored, there is no record of it, and no acknowledgement that it ever even happened. But I know, as do my @FDRLST colleagues who have had our worked censored, throttled, shadow-banned, and de-monetized for YEARS under the censorship regime of the Biden years.
And make no mistake, it isn't gone. As soon as a Democrat occupies the White House again, the censorship will come roaring back, worse than before. Count on it.
New essay on the closure of the Internet, the mass censorship campaign that transformed every platform and the Internet as a whole from free and open to a closely monitored playpen 2015-2022. Longer than usual, around 27,000 words, so maybe best read in multiple sittings. Link in next post.
Before 2015, the Western Internet was almost entirely free and open, except for actual criminal activity. Internet censorship was opposed by the major relevant institutions and seen as something distinctive to authoritarian states like Russia and China. Then within just a few years, every major platform (I cover YouTube, Reddit, Facebook, Twitter, Amazon, Wikipedia, and Google Search, as well as more general deplatforming of websites, forums and open source projects, and comments sections) created a colossal censorship apparatus with tens of thousands of human moderators, opaque and ever-shifting rules, and an entirely new field of machine learning spun up ex-nihilo.
























