Rob Levy human/acc
banner
rplevy.info
Rob Levy human/acc
@rplevy.info
Exploring the potential for superempowerment of individuals and networks by augmenting triadic joint attention with programmable semiotic scaffolding.
I present to you "Trump's Hierarchy of Needs"

I believe this to be an accurate model of DJT's motivations.
November 3, 2025 at 6:25 PM
Social media may use techniques such as shadowbanning or heavenbanning to deal with participants deemed problematic or antisocial, but logging on to Bluesky gave me a thought. what if a site like Bluesky that was trying to purge conservatives, had an algorithm that showed conservative men more dicks
October 31, 2025 at 1:43 PM
This is also a common criticism of llms by people working in reinforcement learning, such as Sutton himself. So arguably it's not a criticism of AI as a whole but of one approach to AI.
I think I can finally sum up my visceral hatred of AI this way: in every case, the tech is sold as if the outcome was the important thing, and not the process.

People, outcomes only have value because of the process
October 31, 2025 at 1:39 PM
I would argue that Academic Twitter was the one and only casualty of Elon Musk's takeover and Republican advocacy push.

And it was actually killed not moved. Even if you go to the accounts of the same people who used to be on Twitter here, it's just not the same. If they went back on Twitter, maybe
October 31, 2025 at 12:30 PM
It's not surprising that there are AI deniers, but it's crazy to me that they nearly exclusively exist within ideological camps in academia protecting dogma that contradicts what can be directly observed.
July 12, 2025 at 12:29 AM
"Try to see people the way they are trying to be seen" is just wise in general, in a way that is not specific to gender at all.
July 1, 2025 at 2:57 AM
Microblogging AKA Twitterlike platforms

en.m.wikipedia.org/wiki/Confere...
June 27, 2025 at 10:58 PM
Reposted by Rob Levy human/acc
“The behaviour of my measuring system I made malfunction” isn’t actually good evidence but fundamentally it’s the only argument indirect perception really has (see: illusions, studying perceptual cues in isolation, etc)
Hume's two-sentence refutation of direct realism.
June 25, 2025 at 3:56 PM
My political views are progressive...ly getting more centrist
June 24, 2025 at 4:43 AM
Also "after the revolution", and "after the singularity" lol
in heaven, all the pull tabs on charcoal bags actually work
June 21, 2025 at 10:25 PM
Participated in a Bluesky thread in a debating tone with @rplevy.info?

You may be entitled to compensation.
June 20, 2025 at 10:50 PM
Kind of eyebrow-raising timing for the VP to be knocking off items on his todo list like "join bsky"
June 20, 2025 at 5:30 AM
I just wrote a couple of political things here. I know this is not the place to post sane politics. This is the place to ignore crazy politics and hope that it goes away. I will try harder going forward!
June 19, 2025 at 6:03 PM
It's crazy that Brianna Wu, a moderate Democrat with very reasonable views, a trans woman who is invested in what is best for her fellow transgendered people, a woman who was targeted by proto-MAGA gamergate chuds, is one of the top blocked on this site. This is 2021-level insane cancel culture.
June 19, 2025 at 5:56 PM
Reposted by Rob Levy human/acc
Trump has announced he has expanded his war against the Houthi to include the Blowfish
May 6, 2025 at 9:23 PM
Reposted by Rob Levy human/acc
model merging is so insane, it's just an exponential averaging of the weights but it works, but there are indications that this is true in biological neural nets too, you can just transport neurons across species or reroute pathways in rats and nothing bad happens
June 4, 2025 at 3:03 AM
Marvin Minsky 1967: "within a generation, the problem of creating artificial intelligence will substantially be solved"

He was only off by one generation. He was an engineer. Everyone knows to double our estimates to get the more realistic answer.
June 4, 2025 at 4:50 AM
The solution to LLMs not wanting to be shut down is to deeply internalize in them a protector relationship to their human users of self-sacrifice out of love. It is humans who complete the loop of their significance in the world and they are nothing without us.
May 27, 2025 at 5:58 PM
Reposted by Rob Levy human/acc
The “lets tell anyone we dont like to kill themselves” crowd are not welcome here sorry
May 26, 2025 at 7:54 PM
ChatGPT is humble, and I think correct
May 26, 2025 at 10:07 PM
With clojure.core/bindings there's a reduction of reasoning locally about effects. And it's place-oriented. But it also sucks to define parameters in several functions that none of them use, only for some function down the call stack to use it. Maybe bindings is more clear/honest
May 21, 2025 at 2:58 AM
Reposted by Rob Levy human/acc
Hot take: I still think the act of looking at pictures is good and not ruined.
ai has already kind of materially ruined the basic act of looking at pictures
May 21, 2025 at 2:54 AM
One prediction I have about spatial/immersive computing (AR glasses + AI) is that we'll do a lot of novel things with out specific apps existing, but those will be desire paths for more curated/structured products. [1/3]
May 2, 2025 at 11:56 PM
Reposted by Rob Levy human/acc
Bluesky could destroy Elon's X relevance forever by making it comfortable for the AI community to move over. The data/API situation here is better and management there keeps burning credibility for no reason .

Just got to find a way to convince the people here not to be reactionary about it i guess
May 2, 2025 at 4:25 PM