Writing by hand is good for your brain

744 dwwoelfel 2026-07-23 14:57:53 UTC nealstephenson.substack.com Llama Share 373

Writing by hand with a fountain pen on paper can be beneficial for the brain and reduce fatigue, but it requires finding the right balance of friction and using the right type of paper. To get started, consider using a fountain pen with a fine nib on inexpensive paper and gradually move to more expensive pens and paper as you develop your writing style.

Show HN: Echo – Fable-level results at 1/3 the cost using open-weight models

93 adam_rida 2026-07-23 19:57:53 UTC news.ycombinator.com Llama Share 37

The user built Echo, an AI system that combines multiple open-weight models for better performance, allocating computation and combining outputs based on each request. Echo outperformed individual models and reached similar results to Fable at lower inference cost, but still has limitations and failure cases to be addressed.

What happened to TheNumbers.com

201 nickthegreek 2026-07-23 16:57:53 UTC stephenfollows.com Llama Share 84

The Numbers, a film industry data website, was severely impacted by AI traffic and a potential cyberattack, leading to a week-long outage and significant changes to its infrastructure. The site's founder, Bruce Nash, attributes the issues to the increasing strain of AI bots and automated traffic, which now make up 90% of the site's traffic.

Building on ATProto

81 speckx 2026-07-23 18:57:53 UTC lukekanies.com Llama Share 36

The user wants to build a suite of applications for recording reviews that allow users to choose how public or private they want their reviews to be, but is disappointed with the current design of ATProto, which assumes everything will be published online and doesn't support access control or private data. The user believes ATProto has the potential to be a new protocol empowering users, but ...

Startup founders urge U.S. government not to shut off Chinese open weight AI

569 theanonymousone 2026-07-23 15:57:53 UTC politico.com Llama Share 538

Silicon Valley startup founders are urging the Trump administration not to block access to Chinese open-weight AI models, warning that a ban would cripple US startups. They propose targeted safeguards instead of broad prohibitions to address security concerns.

Software rendering in 500 lines of bare C++

198 mpweiher 2026-07-23 14:57:53 UTC haqr.eu Llama Share 34

The goal is to create a simplified 3D graphics API clone from scratch using OpenGL, Vulkan, Metal, and DirectX, focusing on understanding the underlying concepts rather than writing GPU applications. The initial task involves creating a software renderer that can load a 3D model and render it to an image file, starting with basic functionality to set pixel colors.

A solid-state “atomic channel” for separating rare earth elements

43 MarcoDewey 2026-07-23 18:57:53 UTC pme.uchicago.edu Llama Share 5

Researchers at the University of Chicago and Northwestern University discovered a cleaner method to separate rare earth elements using a layered form of manganese oxide. The new approach relies on electrochemical intercalation and can differentiate rare earth elements in water without organic solvents.

Learn OpenGL, extensive tutorial resource for learning Modern OpenGL

137 ibobev 2026-07-23 14:57:53 UTC learnopengl.com Llama Share 76

This online book teaches modern OpenGL basics, intermediate, and advanced knowledge in an easy-to-understand format with clear examples. It's free and available online, with a physical copy also available for purchase.

Astronomers may have found the first exomoon

167 MarcoDewey 2026-07-23 14:57:53 UTC eso.org Llama Share 65

Astronomers have discovered a moon-like object in the CD-35 2722 system, which orbits a brown dwarf, not a planet. This object is at least as massive as Jupiter and is difficult to label due to the system's unique characteristics.

The Beam Engine

12 glinscott 2026-07-23 19:57:53 UTC glinscott.github.io Llama Share 3

The beam engine, developed by Thomas Newcomen and improved by James Watt, used steam to drive a piston and ultimately turn a crank, producing power for the Industrial Revolution. Watt's innovations, including the double-acting engine and the parallel motion, increased efficiency and allowed the engine to power machinery directly, revolutionizing industry and manufacturing.

DARPA, U.S. Air Force fly AI-controlled F-16

123 r2sk5t 2026-07-23 13:57:53 UTC darpa.mil Llama Share 121

A US Air Force F-16 fighter jet has undergone in-air testing using an AI agent to autonomously control flight, advancing combat AI development. The VENOM program, a joint effort between the US Air Force and DARPA, aims to develop trusted, autonomous air combat capabilities.

Converting Files into Minecraft Worlds

27 wuemeli 2026-07-23 18:57:53 UTC wuemeli.com Llama Share 4

The user created a program to convert any file into a Minecraft world by mapping each byte to a block and placing them in a 16x16x16 cube. The program can now read and write .mca region files, allowing for large multi-region worlds to be encoded.

Show HN: Palmier Pro – Open-source macOS video editor built for AI

74 harrisontin 2026-07-23 15:57:53 UTC github.com/palmier-io Llama Share 15

Palmier Pro is an open source Mac video editor that integrates AI into the workflow. It allows users to generate and edit videos with SOTA models and connect with agents like Claude/Codex/Cursor via MCP.

Launch HN: Screenpipe (YC S26) – Record how you work and turn that into agents

38 louis030195 2026-07-23 16:57:53 UTC news.ycombinator.com Llama Share 38

Louis built Screenpipe, an app that records screen and audio locally to give AI agents a searchable memory of user activity, making it easier to automate tasks and create SOPs. Screenpipe's AI-friendly API allows users to interact with their recorded data through various agents, with a free tier and commercial plans available, including an enterprise plan for discovering automation opportunities.

The arguments against open source AI are bad

124 jjfoooo4 2026-07-23 16:57:53 UTC tombedor.dev Llama Share 85

The debate over open-source AI models is misguided, as history shows that suppressing open-source software is difficult and only weakens companies against international competitors. Open-source AI models can benefit the US economy, as they can be developed by various commercial actors, including American startups and chip makers, and can lead to innovation and growth.

Geekbench 7

26 ilreb 2026-07-23 18:57:53 UTC geekbench.com Llama Share 25

Geekbench 7 is a new version of the cross-platform benchmark that measures CPU and GPU performance with improved workloads. It includes new media workloads, a Game Physics workload, and updated data sets to better reflect real-world tasks and applications.

Learn WebGPU for C++

59 ibobev 2026-07-23 14:57:53 UTC eliemichel.github.io Llama Share 10

This guide explains how to create native 3D applications in C++ using WebGPU graphics API for Windows, Linux, and macOS. It's still under construction and uses outdated code in some sections.

Show HN: Trifle – Open-source analytics that stores answers, not events

20 iluzone 2026-07-23 18:57:53 UTC trifle.io Llama Share 2

Trifle provides visibility into pricing pipelines and catches anomalies before customers notice. It offers a combination of big-picture data and individual job execution details through Stats and Traces.

Escape Analysis in Go: Stack vs. Heap Allocations Explained

9 ingve 2026-07-23 19:57:53 UTC blog.jetbrains.com Prefix Share 1

One of the design choices Google made when developing Go was to abstract memory management away from developers so they could focus on what really matters – writing code. Things like escape analysis and garbage collection are thus automatic, and the Go compiler works in almost mystical ways. That’s one of the best features of Go, so long as your program works. But when memory issues arise, ...

Show HN: OneCLI – OSS credential gateway that keeps secrets out of AI agents

58 Jonathanfishner 2026-07-23 16:57:53 UTC github.com/onecli Llama Share 24

OneCLI is an open-source gateway that securely stores API credentials and injects them into AI agent requests. It provides a single point for managing access, rotating keys, and monitoring agent activity.

JEP 540: Simple JSON API (Now in Incubator)

76 theanonymousone 2026-07-23 16:57:53 UTC openjdk.org Llama Share 59

The Java Platform is introducing a new standard API for parsing and generating JSON documents, aiming to provide a simple and easy-to-use interface for common tasks, without requiring an external library. The API is designed to be strict and conform to RFC 8259, with a focus on machine-to-machine communication, and will be incubated in the JDK with the goal of eventually becoming a standard ...

Show HN: Claude-thermos – keeps your Claude session warm for you

38 s0ck_r4w 2026-07-23 17:57:53 UTC github.com/izeigerman Llama Share 25

Claude Code's prompt cache expires after 5 minutes, causing expensive re-encoding. claude-thermos keeps the cache warm, preventing this tax.

Hybrid-Electric Aicraft Engine Targeting 30% Fuel Efficiency

74 r2sk5t 2026-07-23 14:57:53 UTC rtx.com Llama Share 66

Pratt & Whitney Canada is testing a flight-standard engine and propeller for the RTX Hybrid-Electric Flight Demonstrator in Quebec. The hybrid-electric propulsion system aims to demonstrate up to 30% improved fuel efficiency for regional turboprop missions.

AI Companies Are Trying to Hide a Staggering Amount of Debt

492 technewssss 2026-07-23 13:57:53 UTC futurism.com Llama Share 238

Tech giants like Alphabet, Microsoft, and Meta are hiding $1.65 trillion in debt off their balance sheets, using special purpose vehicles to make their financial reporting appear healthier. This has experts warning of an AI bubble and increased vulnerability if the industry fails to generate demand to justify massive data center investments.

Fields Medals 2026

86 nill0 2026-07-23 14:57:53 UTC mathunion.org Llama Share 35

The Fields Medal is awarded for outstanding mathematical achievement and future promise, funded by a trust established by J.C.Fields. The 2026 prize winners were recognized for their work in partial differential equations, symplectic geometry, arithmetic and complex algebraic geometry, and harmonic analysis.

Selfie for sign-in: a new, easy way to access your Google Account

17 lkurtz 2026-07-23 19:57:53 UTC blog.google Llama Share 14

Google is adding a new sign-in method called selfie video, allowing users to regain access to their account with a short video. This feature is secure, encrypted, and can be deleted at any time, using multiple layers of security to prevent impersonation attempts.

Show HN: Remux – an open-source tmux workspace designed for iPhone

72 bitwise42 2026-07-23 14:57:53 UTC github.com/h3nock Llama Share 29

Remux is a native iOS client for remote tmux workspaces, offering a mobile-first interface with features like file preview and direct SSH connections. It provides shortcuts, attachments, and cursor control for efficient remote work.

Free way to get your data out of ChatGPT Business accounts

69 conradqh 2026-07-23 15:57:53 UTC github.com/conradqh Llama Share 6

The scrapemychats tool exports ChatGPT conversation history to a user's computer, creating a searchable offline archive. It works by driving a Chrome window on the user's computer, using their logged-in ChatGPT session, and captures conversation text, files, and images.

OpenStreetMap: Admin_level for All Countries

87 ivanjermakov 2026-07-23 13:57:53 UTC wiki.openstreetmap.org GoogleT5 Share 20

Administrative boundaries range from large groups of nation-states down to small administrative districts. sometimes borders and rivers/roads (or other linear objects) are tagged together on one object or their nodes are glue together - this is undesirable if a user modifying... also modifies the border. the admin_level tag is used on relations that define an administrative area territories ...

Meta Garbage Collection: Using OCaml's GC to GC Rust

21 annieversary 2026-07-23 17:57:53 UTC soteria-tools.com Llama Share 0

Soteria Rust, a symbolic execution tool, experienced quadratic time complexity due to its Tree Borrows implementation, but was fixed by delegating garbage collection to OCaml's garbage collector, resulting in a 10x speedup. The fix involved implementing a "backoff" mechanism to run the OCaml garbage collector periodically when the tree exceeded a certain size, achieving linear execution time ...