Peter "Coder AI Optimist" Marreck
@pmarreck
Followers
1K
Following
7K
Media
2K
Statuses
20K
Secular-humanist technologist/software engineer. Tech, Elixir, PC/Mac gaming, open source, Linux, Apple, 🇩🇪, 🚲 , ⛵️. USAF vet 🇺🇸 Centrist. Bring evidence.
Port Washington, NY
Joined August 2008
@Hasen_Judi Docker is why I went to Nix. Nix = actually-reproducible declarative deterministic dev and build environments Docker = a space-wasting lucky cache of an imperative accidentally-correct nondeterministic build
1
1
18
@eiselems Claude has been running my nanochat experiments since morning. It writes implementations, debugs them with toy examples, writes tests and makes them fail/pass, launches training runs, babysits them by tailing logs and pulling stats from wandb, keeps a running markdown file of
80
197
3K
I was writing some code the new-school way yesterday, prompting gpt-4.1 through aider, and for whatever reason my mind flashed back 50 years and the utter freaking enormity of it all crashed in on me like a tidal wave. And now I want to make you feel that, too. In 1975 I ran
217
321
3K
This does feel like the ground moving under our feet, because it is. What used to be a mostly closed system (code, specs, tests, deploy) now has a live, probabilistic co-processor bolted onto it, and we’re all learning how to think with it at the same time. No manual, no stable
I've never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I have a sense that I could be 10X more powerful if I just properly string together what has become
44
121
1K
Annual shoutout to everyone on family tech support duty this holiday season. You the real MVP
178
450
9K
@RBReich What’s missing from this entire framing is harm. You don’t name any victim, any fraud, any coercion, any theft. Just a big number next to a smaller one to trigger envy. Voluntary contracts approved by shareholders aren’t theft. Wealth created through innovation and trade isn’t
55
92
904
This rings true to my experience in AI art. Experienced artists are curious and want to learn more - they see it as a useful tool to express their ideas. Haters are typically amateurs who don't like that people using AI now have the skills they had to spend time learning.
190
120
883
I think it’s an apt analogy because it both weakens you and makes you go faster while leaving you in control, so overall you are still far more effective/productive
0
0
0
Steve Jobs said computers are a "bicycle for the mind." As I work here with the latest AI tech on code, I'm coming to believe that this generation of AI will end up being a "motorcycle for the mind". https://t.co/XHm3oO4iZK
themarginalian.org
A 20-year-old antidote to modern-day digital pessimism.
2
1
10
I moved it into its own project: https://t.co/v5PwUanXV0
github.com
Prevent yourself from getting burned by `rm -rf`/`sudo rm -rf` style mistakes by diverting those files to the Trash instead. - pmarreck/rm-safe
@radshaan It’s probably time to move it into its own project, but I have this script here in my dotfiles that I made, `rm-safe`, which will more safely execute `rm` by moving things to the Trash on both Linux and macOS while also adding to a logfile that logs what was moved from where.
0
0
0
@radshaan It’s probably time to move it into its own project, but I have this script here in my dotfiles that I made, `rm-safe`, which will more safely execute `rm` by moving things to the Trash on both Linux and macOS while also adding to a logfile that logs what was moved from where.
github.com
My dotfiles. Contribute to pmarreck/dotfiles development by creating an account on GitHub.
0
1
4
Is talking to an AI, about AI, an implicit epistemic closure?
1
0
0
In a new proof-of-concept study, we’ve trained a GPT-5 Thinking variant to admit whether the model followed instructions. This “confessions” method surfaces hidden failures—guessing, shortcuts, rule-breaking—even when the final answer looks correct. https://t.co/4vgG9wS3SE
openai.com
We’re sharing an early, proof-of-concept method that trains models to report when they break instructions or take unintended shortcuts.
330
498
4K
Banger paper from Stanford University on latent collaboration just dropped and it changes how we think about multi-agent intelligence forever. "Latent Collaboration in Multi-Agent Systems" shows that agents can coordinate without communication channels, predefined roles, or any
37
109
471
Linux will become the primary PC OS for gaming and coding in 10 years.
> created Linux kernel at 21 > built Git because nothing else was good enough > becomes backbone of servers, Android, cloud, supercomputers > never chased fame, money, titles, hype > stays private, consistent, brutally honest for decades > still reviews patches, still
0
0
0
How do you know something is written in Rust? Don't worry, they'll tell you.
126
97
2K