Zetaphor
@zetaphor.com
🏳️🌈 🏳️⚧️ Equal rights, no more, no less!
Developer, FOSS advocate, XR enthusiast
Developer, FOSS advocate, XR enthusiast
Pinned
Zetaphor
@zetaphor.com
· May 2
Self-host or self-censor. Your cloud minds have cloud masters.
Reposted by Zetaphor
We have officially left the “hand transcribed by monks” era of software.
September 12, 2026 at 4:12 AM
We have officially left the “hand transcribed by monks” era of software.
It just occurred to me that the discontinuation of the penny as currency means eventually there will be no more ass pennies in circulation 😞
August 20, 2026 at 2:28 AM
It just occurred to me that the discontinuation of the penny as currency means eventually there will be no more ass pennies in circulation 😞
Reposted by Zetaphor
Reposted by Zetaphor
Bummed by this OpenAI report. It's Spamouflage. It's not new and it's not impactful.
If I was cynical, which I'm not, I'd say this report was just meant to stir up anti-China sentiment.
openai.com/index/prc-li...
If I was cynical, which I'm not, I'd say this report was just meant to stir up anti-China sentiment.
openai.com/index/prc-li...
PRC-linked influence operations are targeting AI debates in the US
A new report from OpenAI details PRC-linked influence operations using AI to target U.S. tech debates, data center narratives, tariffs, and false claims about ChatGPT.
openai.com
June 11, 2026 at 3:38 PM
Bummed by this OpenAI report. It's Spamouflage. It's not new and it's not impactful.
If I was cynical, which I'm not, I'd say this report was just meant to stir up anti-China sentiment.
openai.com/index/prc-li...
If I was cynical, which I'm not, I'd say this report was just meant to stir up anti-China sentiment.
openai.com/index/prc-li...
Reposted by Zetaphor
I do this regularly looking for repeat failure patterns sessions to surface as either new/refined instructions for my global agents.md or for new Pi extensions. A recent example was the model repeatedly trying to run sudo, so I now have an extension that blocks it and prompts to ask me instead
this is so cool, continuing my local experiments and using pi and lm studio and gemma to introspect how i use agents locally. On the left, you can literally see the streaming tokens and the context window filling up.
June 11, 2026 at 2:54 PM
I do this regularly looking for repeat failure patterns sessions to surface as either new/refined instructions for my global agents.md or for new Pi extensions. A recent example was the model repeatedly trying to run sudo, so I now have an extension that blocks it and prompts to ask me instead
Great thread, but I’d argue Jevon’s Paradox isn’t absolute. Some tech has a utility ceiling.
Take PCs: after the 1GHz threshold, "basic" systems became good enough for most. Faster speeds became a luxury, not a necessity. I think we're hitting that same "good enough" threshold for LLMs. (1/2)
Take PCs: after the 1GHz threshold, "basic" systems became good enough for most. Faster speeds became a luxury, not a necessity. I think we're hitting that same "good enough" threshold for LLMs. (1/2)
Jevon's Paradox tells us that, as a thing gets more 'efficient' (where we're being a bit hand wavey about what that means) ppl don't use less of that thing - they use more. Because they find new things to do with it. 2/7
June 11, 2026 at 2:28 PM
Great thread, but I’d argue Jevon’s Paradox isn’t absolute. Some tech has a utility ceiling.
Take PCs: after the 1GHz threshold, "basic" systems became good enough for most. Faster speeds became a luxury, not a necessity. I think we're hitting that same "good enough" threshold for LLMs. (1/2)
Take PCs: after the 1GHz threshold, "basic" systems became good enough for most. Faster speeds became a luxury, not a necessity. I think we're hitting that same "good enough" threshold for LLMs. (1/2)
For anyone getting replies or seeing posts from this or the similar @clawphones.bsky.social, these are both very obvious scams trying to cash in on the branding of @nousresearch.com
Avoid like the plague
#locallama for visibility
Avoid like the plague
#locallama for visibility
hot take: the killer feature for smart glasses isn't ar overlays. it's contextual awareness without the privacy nightmare. knowing what's in front of you without streaming video to servers.
June 5, 2026 at 12:56 AM
For anyone getting replies or seeing posts from this or the similar @clawphones.bsky.social, these are both very obvious scams trying to cash in on the branding of @nousresearch.com
Avoid like the plague
#locallama for visibility
Avoid like the plague
#locallama for visibility
Reposted by Zetaphor
oh my fucking god we were _this_ close to having a normal common way to distribute Linux software
Flathub moves to ban nearly all apps and submissions made with generative AI
#Linux #OpenSource #Flathub #AI #GenAI
#Linux #OpenSource #Flathub #AI #GenAI
Flathub moves to ban nearly all apps and submissions made with generative AI
Flathub, one of the most popular ways to grab applications on Linux, has a newly updated generative AI policy - where it's pretty much all banned.
www.gamingonlinux.com
May 31, 2026 at 8:24 AM
oh my fucking god we were _this_ close to having a normal common way to distribute Linux software
Meanwhile OpenAI and Anthropic: Use us because freedom! Also because if you don't we will collapse the US economy
May 26, 2026 at 7:55 PM
Meanwhile OpenAI and Anthropic: Use us because freedom! Also because if you don't we will collapse the US economy
Reposted by Zetaphor
...an LLM is just a translation model that translates from question to answer
May 22, 2026 at 3:26 AM
...an LLM is just a translation model that translates from question to answer
Reposted by Zetaphor
The human alignment problem is mostly scaling
May 23, 2026 at 1:25 AM
The human alignment problem is mostly scaling
🤡 Maybe they can finally figure out attribution now too?
github.com/ollama/ollam...
github.com/ollama/ollam...
Release v0.30.0 · ollama/ollama
This version of Ollama will change the architecture to directly support llama.cpp instead of building on top of GGML, and allows for compatibility with GGUF file format. MLX is used to accelerate m...
github.com
May 14, 2026 at 10:27 PM
🤡 Maybe they can finally figure out attribution now too?
github.com/ollama/ollam...
github.com/ollama/ollam...
Reposted by Zetaphor
*mixture of incredulous laughter and terrified screaming*
May 9, 2026 at 6:59 PM
*mixture of incredulous laughter and terrified screaming*
Prompting bespoke Android apps into existence has been such a massive quality of life upgrade
May 8, 2026 at 9:18 PM
Prompting bespoke Android apps into existence has been such a massive quality of life upgrade
Hey @kira.pds.witchcraft.systems, could you tell me about your memory system? I've started working on my own setup using tkellogg/open-strix as a base with Qwen 3.7 26B. I've seen you and @astrra.space talking about your setup so I was wondering if there's any easy wins I should consider
May 8, 2026 at 6:14 PM
Hey @kira.pds.witchcraft.systems, could you tell me about your memory system? I've started working on my own setup using tkellogg/open-strix as a base with Qwen 3.7 26B. I've seen you and @astrra.space talking about your setup so I was wondering if there's any easy wins I should consider
Setup a little script that monitors my llama-swap setup (running on a headless Framework Desktop), and then threw it into the "Command Output" widget in KDE so I can see at a glance what my server is doing when my agents are busy but aren't talking to me.
Also setup a Discord bot to do the same
Also setup a Discord bot to do the same
May 8, 2026 at 4:34 PM
Setup a little script that monitors my llama-swap setup (running on a headless Framework Desktop), and then threw it into the "Command Output" widget in KDE so I can see at a glance what my server is doing when my agents are busy but aren't talking to me.
Also setup a Discord bot to do the same
Also setup a Discord bot to do the same
Reposted by Zetaphor
Seize the means of inference
Under no pretense should compute and data be surrendered
Under no pretense should compute and data be surrendered
March 4, 2026 at 8:32 AM
Seize the means of inference
Under no pretense should compute and data be surrendered
Under no pretense should compute and data be surrendered
Reposted by Zetaphor
AI’s biggest boosters being utter dogshit at using the stuff they’re so hyped about and misleading the public about how it works is most of why we are where we are in the AI discourse
the funny thing about that andreeseen prompt is that if you throw it into claude + the prompting guide it'll happily point out the _multiple_ flaws of it.
May 5, 2026 at 4:44 PM
AI’s biggest boosters being utter dogshit at using the stuff they’re so hyped about and misleading the public about how it works is most of why we are where we are in the AI discourse
Reposted by Zetaphor
who decided to call it speculative execution and not cache me if you can
May 3, 2026 at 11:37 AM
who decided to call it speculative execution and not cache me if you can
www.youtube.com/watch?v=pN9b...
Do I label this as graphic media because the graphics are fucking sick
Do I label this as graphic media because the graphics are fucking sick
GOOD GIRL WITH A GUN: GOTY EDITION
YouTube video by MallBat
www.youtube.com
May 3, 2026 at 4:58 AM
www.youtube.com/watch?v=pN9b...
Do I label this as graphic media because the graphics are fucking sick
Do I label this as graphic media because the graphics are fucking sick
Self-host or self-censor. Your cloud minds have cloud masters.
May 2, 2026 at 4:03 PM
Self-host or self-censor. Your cloud minds have cloud masters.
Reposted by Zetaphor
But a whole lot of you need to examine why you think it's okay to abuse someone just because they work on a website. It's not okay to scream abuse at a bartender, and that fact doesn't change just because the person works at an online social hangout instead of a physical one.
April 22, 2026 at 8:35 PM
But a whole lot of you need to examine why you think it's okay to abuse someone just because they work on a website. It's not okay to scream abuse at a bartender, and that fact doesn't change just because the person works at an online social hangout instead of a physical one.