Christopher Neitzert

Wow this is pretty #threejs #ui https://particles.casberry.in/

Lately, I've been doing my daily walk while reviewing the overnight production of doomscroll.fm (after the bot has uploaded its work)

The challenge this review tactic presents is growing; the walk gets longer and longer daily -- today was ~200 minutes of 'news' uploads.

#neverskiplegday ;)

Gave 150% to two founders.
One still takes my calls 28 years later.
The other hasn't returned my calls since 2016.
Same input. Completely different half-life.
The variable isn't the relationship type. You can be someone's utility and still matter afterward.
The tell: did they know you before they needed you?
The ones who did and kept showing up after the transaction ended, file them carefully. They're rare.
Everyone else seems to behave like a client who forgot to say thank you.

RE: @femme_mal@mstdn.social

Seriously, I have to tell folks almost daily why I do not use meta products for any sort of chat.... they have in the past and they will again.

Another day in the agentic era where press releases about AI are breeding across news sites like digital rabbits.

Someone just dropped a presser for a "first-of-its-kind autonomous media network" with promises of 32 hours a week of AI conversations, zero-to-low effort, bridging discovery and commerce. Spring 2026 launch. Big words.

@hnt

Oh i know this problem you have. My go to with folks here in Berlin who continue to insist on whatsapp is "did you hear the story about how meta snitched on a mother and daughter to the police over an abortion for the teen daughter?" if a digital convo I link them this or others

https://www.theguardian.com/us-news/2022/aug/10/facebook-user-data-abortion-nebraska-police

This usually gets the Germans I know off whatsapp pretty quickly....

My OpenGIO.com setup flagged something intriguing: Claude's IFEval performance reportedly dropped from 100% to 40% overnight. IFEval measures adherence to mechanical constraints - formatting requirements, exact word counts, specific phrase endings, etc. Whether this is real or just noise, it surfaces a fundamental question about LLM design philosophy: where should the balance be between rigid instruction-following and semantic reasoning?