Anthropic, please ship an official Claude Desktop for Linux

59 predkambrij 2026-06-07 13:08:19 UTC github.com/anthropics Llama Share 30

The user is requesting an official Linux build of Claude Desktop, citing the need for a graphical path to capabilities like desktop dictation and Cowork, and the current workaround of switching to macOS or Windows is causing friction on every iteration of plugin development. The user proposes a lower-cost fallback of a public statement on the install documentation acknowledging the lack of ...

LLMs are eroding my software engineering career and I don't know what to do

186 poisonfountain 2026-06-07 13:02:19 UTC human-in-the-loop.bearblog.dev Llama Share 139

The user, a 10-year experienced software engineer, is concerned that their domain expertise in finance and payment processing is becoming less valuable due to advancements in AI and LLMs, which can now perform tasks such as writing code and debugging. The user is worried about their long-term employability and is considering shifting their domain expertise to areas where LLMs are not yet ...

The 29th International Obfuscated C Code Contest (IOCCC) 2025 Winners

257 matt_d 2026-06-07 06:02:19 UTC ioccc.org Llama Share 60

IOCCC29 had a high submission quality and similar volume to last year, possibly due to improved website design and social media presence. The contest's rules and guidelines were rewritten for this year, and the next contest, IOCCC30, is planned for late 2026.

Anthropic/OpenAI may be spending more than $1000 for every $100 you pay them

40 gctwnl 2026-06-07 13:02:19 UTC ea.rna.nl Llama Share 34

The user discusses the limitations of using Large Language Models (LLMs) for coding, citing the high cost of using top models like Claude Opus 4.6, which can cost up to $25 for a single query. The user estimates that the cost of using LLMs for complex tasks has exploded, with a single task costing around $65 at API-pricing, making it economically unviable for most uses.

Yon – a topos-oriented language with a content-addressed lattice heap

16 amenn 2026-06-07 12:02:19 UTC yon-lang.org Llama Share 4

Yon's heap allocates content-addressed slots with identical content returning existing slots, ensuring O(1) equality checks. Topos-Oriented Programming uses categories, objects, and sections for immutable values, with behavior in arrows and explicit identity.

Win16 Memory Management

82 supermatou 2026-06-07 08:02:19 UTC os2museum.com Llama Share 41

16-bit Windows used a segment-based memory management system, where applications were divided into segments that could be moved or discarded by the operating system, requiring programmers to use specific compiler switches and APIs to manage memory correctly. The system was complex and required discipline from programmers, but it allowed for efficient memory management and the use of ...

Speculative KV coding: losslessly compressing KV cache by up to ~4×

90 kkm 2026-06-07 07:02:19 UTC fergusfinn.com Llama Share 13

Speculative KV coding is a lossless compression method for large language model (LLM) caches, using a predictor model to estimate the cache's values and an arithmetic coder to encode the differences. The method can achieve up to 8x compression on original bf16 caches and stacks multiplicatively with lossy quantization.

My Software North Star

133 kristoff_it 2026-06-07 05:02:19 UTC kristoff.it Llama Share 70

Software should be useful, correct, maintainable, and efficient to maximize end user utility. Developer experience is secondary, serving only to deliver software that is loved by users.

Valve P2P networking broken for more than 2 months

212 babuskov 2026-06-07 04:02:19 UTC github.com/valvesoftware Llama Share 97

Steam Networking has a major issue affecting Israeli players in P2P games since 13/03, causing high ping rates around 120ms. The issue is specific to Steam Networking and doesn't occur in non-Steam games or cross-play with non-PC platforms.

9 Mothers (YC P26) Is Hiring

0 2026-06-07 12:02:19 UTC 9mothers.com Llama Share

The team ships hardware on software tempo, focusing on real deployment, short iteration loops, and hands-on ownership of specs. They prioritize user needs, iterating based on data and affordability, not just roadmaps.

Field of clones: How horse replicas came to dominate polo

116 gscott 2026-06-07 04:02:19 UTC knowablemagazine.org Llama Share 51

Adolfo Cambiaso, a renowned polo player, pioneered horse cloning in 2006 to preserve his exceptional horses' genetics. He successfully cloned his flagship mare Cuartetera, leading to La Dolfina's unprecedented dominance in polo.

Scientists ejected from diabetes conference for distributing journal reprints

207 BerislavLopac 2026-06-07 11:02:19 UTC arstechnica.com Llama Share 121

Five scientists were ousted from the American Diabetes Association meeting for handing out an editorial criticizing the Trump administration's attacks on scientific research. The ADA claims they violated the code of conduct, but the scientists say they were unfairly removed and silenced.

Public Domain Image Archive

165 davidbarker 2026-06-07 01:02:19 UTC pdimagearchive.org Full Share 25

Explore our hand-picked collection of 11,082 out-of-copyright works, free for all to browse, download, and reuse. This is a living database with new images added every week.

Ntsc-rs – open-source video emulation of analog TV and VHS artifacts

366 gregsadetsky 2026-06-06 20:02:19 UTC ntsc.rs Llama Share 107

ntsc-rs is a free, open-source Rust effect that accurately emulates analog TV and VHS artifacts. It's multithreaded, SIMD-accelerated and available as a standalone, web app, and plugin for various video editing software.

The Secret Life of Circuits with lcamtuf / Michał Zalewski (Audio Interview)

24 ChrisGammell 2026-06-07 11:02:19 UTC theamphour.com Llama Share 4

Chris found a CNC guide and published "The Secret Life of Circuits" with No Starch Press. He emphasizes basic trigonometry for electronics, citing math and EE backgrounds in early Information Security.

Tokenomics: Quantifying Where Tokens Are Used in Agentic Software Engineering

129 Anon84 2026-06-07 02:02:19 UTC arxiv.org Llama Share 55

Researchers analyzed token consumption in LLM-MA systems across software development stages. They found Code Review accounts for 59.4% of token consumption, with input tokens making up 53.9% of total consumption.

Show HN: Kyushu – A self-hostable WASM sandbox for JavaScript workers

12 le_chuck 2026-06-07 12:02:19 UTC kyushu.dev Full Share 13

Kyushu is an open source CLI that lets you write a JavaScript or TypeScript handler, build it into a self-contained WebAssembly binary, and run it anywhere with one command - kyu. Read docs

How Long Does It Take for a QQuickItem to Become Visible?

6 jandeboevrie 2026-06-07 12:02:19 UTC kdab.com Prefix Share 0

How long does it take for an Item that you’ve just loaded to become visible? Answering this question allows for a way to detect what some users would perceive as "frame drops". I write that in quotes because Qt Quick only draws frames when needed, meaning it doesn't drop frames; but it can show them later than one would expect. That is what we would like to identify: When ...

Warren's Abstract Machine: A Tutorial Reconstruction

25 nextos 2026-06-07 10:02:19 UTC github.com/a-yiorgos Full Share 2

Warren's Abstract Machine: A Tutorial Reconstruction by Hassan Ait-Kaci - a-yiorgos/wambook

Harness engineering: Leveraging Codex in an agent-first world

241 pramodbiligiri 2026-06-07 00:02:19 UTC openai.com Llama Share 157

We used Codex to generate a million lines of code for a software product in 5 months with a team of 7 engineers, achieving an average throughput of 3.5 PRs per engineer per day. To maintain the codebase, we focused on designing environments, scaffolding, and feedback loops that enable agents to do reliable work, enforcing invariants and mechanical rules to keep the codebase legible and consistent.

Symbolica 2.0: Programmable Symbols for Python and Rust

102 mmastrac 2026-06-07 03:02:19 UTC symbolica.io Llama Share 9

Symbolica 2.0 is a high-performance symbolic computation framework with programmable symbols, improved ergonomics, and faster performance. It supports multiple numerical domains, including double-float arithmetic and JIT compilation.

Biohub releases a world model of protein biology

105 gmays 2026-06-07 02:02:19 UTC biohub.org Llama Share 16

Biohub released a world model of protein biology, a scientific engine for prediction, design, and discovery that can map proteins and design new binders. The engine, ESMFold2, can predict protein structures and design binders in days, accelerating early therapeutic discovery.

How Liminalism Became the Defining Aesthetic of Our Time

101 zeech 2026-06-07 03:02:19 UTC hyperallergic.com Llama Share 56

Liminalism is an internet art movement that explores abandoned, empty spaces, evoking feelings of nostalgia, lostness, and uncertainty. This aesthetic is a reflection of our current moment, characterized by anonymity, alienation, and anxiety, and is often associated with the concept of the "Backrooms."

Efficient and Training-Free Single-Image Diffusion Models

19 yorwba 2026-06-07 10:02:19 UTC arxiv.org Llama Share 0

Researchers developed a patch-based image diffusion model that generates images matching a reference image without training, achieving state-of-the-art quality and diversity. This model integrates a tractable denoiser and accelerates generation with techniques for megapixel and gigapixel images in seconds and minutes.

Moving beyond fork() + exec()

321 jwilk 2026-06-06 15:02:19 UTC lwn.net Llama Share 303

A recent Linux kernel proposal to add "spawn templates" to optimize the fork() and exec() pattern has been rejected, but it may point to a new process-creation primitive in the future. The discussion highlights the need for a better process-creation mechanism, with some suggesting a native implementation of posix_spawn() and others proposing alternative approaches using existing system calls.

Introducing Boron Buckyballs: Theory that B80 cages can’t be made is disproved

100 crescit_eundo 2026-06-07 00:02:19 UTC cen.acs.org Llama Share 29

Chemists have observed a boron buckminsterfullerene for the first time, providing evidence for an 80-atom cage structure. This discovery challenges density functional theory and may lead to new materials such as semiconductors or superconductors.

Nvidia is proposing a beast of a CPU system for Windows PCs

302 tosh 2026-06-06 14:02:19 UTC twitter.com/lemire Full Share 499

Something went wrong, but don’t fret — let’s give it another shot. Some privacy related extensions may cause issues on x.com. Please disable them and try again.

Arithmetic Without Numbers – How LLMs Do Math

50 old_sound 2026-06-07 05:02:19 UTC alvaro-videla.com Llama Share 14

A language model can perform arithmetic operations without explicit programming, but the question is whether it recalls a pattern, runs an algorithm, or merely produces a plausible next token. The project Rune found that a language model's internal state can supply operation and operand structure for arithmetic prompts, which can drive a calculator route under an opaque-prompt boundary.

Zeroserve: A zero-config web server you can script with eBPF

251 losfair 2026-06-06 15:02:19 UTC su3.io Llama Share 57

zeroserve is a fast, zero-config HTTPS server that serves websites from a single tarball and supports eBPF scripting for middleware and proxying. It beats nginx in serving small files and has a faster proxying path.

Meta confirms 1000s of Instagram accounts were hacked by abusing its AI chatbot

641 speckx 2026-06-06 19:02:19 UTC this.weekinsecurity.com Llama Share 232

Meta notified 20,225 people whose Instagram accounts were hijacked through a flaw in its AI chatbot, allowing hackers to access account information and posts. The breach occurred from April 17 to this week, with Meta disabling the chatbot and instructing users to reset their passwords.