HN user

dnw

748 karma
Posts44
Comments83
View on HN
tomjohnell.com 6d ago

OpenInterstate – Canonicalized U.S. Interstate Highway Data

dnw
2pts0
driveamble.com 24d ago

Amble One

dnw
3pts0
blog.calif.io 1mo ago

(Reverse Engineering) Apple Internals: Swift in the Kernel

dnw
4pts0
www.nature.com 1mo ago

General purpose LLMs outperform specialized clinical AI on medical benchmarks

dnw
4pts0
www.fidelity.com 1mo ago

SpaceX IPO available to Fidelity customers with as little as $2k

dnw
2pts2
shopify.engineering 1mo ago

Under the River

dnw
2pts0
openai.com 1mo ago

Building self-improving tax agents with Codex

dnw
2pts0
www.msuiche.com 1mo ago

Supply Chain Attacks Cluster: 230K Advisories, Five Patterns

dnw
3pts0
medium.com 1mo ago

Shannon Got AI This Far. Kolmogorov Shows Where It Stops

dnw
5pts0
www.amacad.org 2mo ago

Future of AI-Facilitated Medicine

dnw
3pts0
jxnl.co 2mo ago

Codex-maxxing

dnw
99pts81
til.simonwillison.net 2mo ago

Using LLM in the shebang line of a script

dnw
4pts0
wix-ux.com 2mo ago

When life gives you lemons, write better error messages

dnw
4pts1
gist.github.com 2mo ago

Implementation Details of Codex /Goal

dnw
2pts0
www.provos.org 2mo ago

Finding Zero Days with any model?

dnw
2pts0
twitter.com 3mo ago

Polymarket weather bet manipulated with a hairdryer

dnw
61pts18
raw.works 3mo ago

RLMs are the new reasoning models

dnw
2pts0
hermes-agent.nousresearch.com 3mo ago

Hermes Agent by Nous Research

dnw
3pts0
ourworldindata.org 3mo ago

Our World in Data

dnw
3pts0
www.anthropic.com 3mo ago

Scaling Managed Agents: Decoupling the brain from the hands

dnw
2pts0
www.anthropic.com 3mo ago

Emotion concepts and their function in a large language model

dnw
191pts204
tomjohnell.com 3mo ago

Who should you subject to your vibe-coding?

dnw
2pts0
www.scbench.ai 3mo ago

Benchmark for measuring code erosion under iterative specification refinement

dnw
1pts0
old.reddit.com 3mo ago

Claude connects the dots on sleep apnea diagnosis

dnw
3pts3
paulgraham.com 4mo ago

Taste for Makers [February 2002]

dnw
2pts0
tomjohnell.com 4mo ago

Pike: To Exit or Not to Exit

dnw
67pts32
www.provos.org 4mo ago

IronCurtain: A Personal AI Assistant Built Secure from the Ground

dnw
3pts0
arxiv.org 4mo ago

Semi-formal reasoning helps agents reason about code without executing the code

dnw
1pts0
blog.katanaquant.com 4mo ago

LLMs work best when the user defines their acceptance criteria first

dnw
461pts422
www.cnbc.com 4mo ago

OpenAI resets spending expectations. Compute target is around $600B by 2030

dnw
6pts0

I would say it is the responsibility of the CTO (whoever is responsible) to talk some sense into the CEOs (whoever in the exec team is having AI psychosis). Yes, yes—what if they themselves are having psychosis but that’s the point.

I had some questions in my mind about DCs using closed loop systems and read the article a little. In case anyone else also has similar thoughts: —-

The company said its water consumption was so high last year because of temporary construction-related activities, such as concrete work, dust control and site preparation. Once operational, the company said the data centers only will use water for domestic needs, such as bathrooms and kitchens. That will total the equivalent of what four U.S. households use per month, the spokesperson said. That may not happen for another few years, however. The company is still actively building and expanding its Fayetteville data center campus. It aims to finish in three to five years.

——

There is that (you can still use coding agents on the other 80% to polish, by the way) but to me this situation underscores the value of a good "editor"--someone who says this is good to ship vs. not this not now.

Claude Code Team: Please fix the core experience instead of branching out into all these tertiary features. I know it is fun and profit to release new features but you need to go deeper into features not broader into there be dragons territory.

Honestly it is unclear to me still. Last month wrote a whole coding principles AGENTS/CLAUDE MD file but not clear if the produced code has changed _because of_ that or other reasons. Couple of things give me confidence that the models actually do read them and follow them.

1) I have a sentence that says: "I am your pairing buddy Bob. Occasionally address me by my name." And you will see them address you by the name, occasionally. 2) I do have per project "Project Tenets." Each tenets is a catchy couple of words, like Design for Agent First, Tolerant Interfaces over Robust Interfaces. If you notice the coding assistant referring to this you know they are aware, and they do sometimes refer to it. I also have a common list of principles but I think they are less useful because those are generic for software engineering. 3) Occasionally, I will do the following prompt: i) Who am I? ii) What is our design philosophy and tenets? iii) Go through the repo and ensure the architecture and codebase adheres to our design philosophy and tenets.

Kind of works.