https://www.nytimes.com/interactive/2025/04/23/business/elon...
or from webarchive
https://web.archive.org/web/20250423093911/https://www.nytim...
I think you might want to check the article again. The interviews were not just based on fired engineers. EM did fire one engineer after he told Musk that interest in Musk was declining.
> What people see feels organic. In reality, they’re engaging with what’s already been filtered, ranked, and surfaced. Naturally, I— and I think many humans have this too- often perceive comments/content that I see as a backscatter of organic content reflecting some sort of consensus. Thousands of people replying the same thing surely gives me the impression of consensus. Obviously, this is not necessarily the truth (and it may be far from it even). However, it remains interesting, because since more people may perceive it as such, it may become consensus after all regardless.
Ultimately, I think it’s good to be reminded about the fact that it’s still algorithmic knobs at play here. Even if that’s something that is not news.
there were three main pieces of actual evidence i leaned on:
1. the nytimes story on account silencing, https://www.nytimes.com/interactive/2025/04/23/business/elon...
2. the visible boost of low-value tweets (though i should have connected to the x api hose and quantify the data),
3. this paper — https://github.com/timothyjgraham/AlgorithmicBiasX/blob/3f4c...
which actually didn’t make it to the the post because the essay was basically finished by the time i remembered i had it. I should have included it. it shows an algo change that boosts elon’s reach specifically.
every other source was speculative:
superbowl fiasko - https://www.theguardian.com/technology/2023/feb/15/elon-musk... - little evidence
elon forced 100 engineers to boost his tweets ? - hearsay
there are also supposed whistle blowers - https://substack.com/@theconcernedbird/p-154577954 - but again. no evidence.
And the algorithm is still closed source.
i’m sorry the writing beyond the graph didn’t land for you.
mr gruez - I'll do better next time.
-- the author