HackerRank open sourced its ATS. My resume scored 90/100. Oh wait 74. No – 88

https://substackcdn.com/image/fetch/$s_!7pmg!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4bfee137-15fc-406e-b2db-91db9cdbb9f7_1980x900.png
An AI resume screening tool is flawed due to non-deterministic outputs from its LLM, leading to inconsistent scores. This tool fails to differentiate between qualified and unqualified candidates, essentially filtering based on luck rather than quality.

Age verification is just a precursor to automated attribution of speech

US states and countries are introducing age verification regulations under the guise of protecting children, but it's actually a precursor to attributing digital identities to physical ones, making it easier for law enforcement to identify and harass individuals. This could lead to automated identity attribution and increased surveillance, allowing governments to target inconvenient people.

Pollen tried to remove my article and Google is assisting with it

https://storage.ghost.io/c/39/f8/39f85cc7-8637-40fc-a57c-f45754453717/content/images/2026/06/Screenshot-2026-06-27-at-17.33.43.png
Google removed an article about Pollen's collapse from its search results due to a bogus copyright claim filed by a fake profile from an uninhabited island. The article's author suspects that Pollen or its former CEO hired reputation firms to file the claim, and is now investigating further lawsuits against the company.

Qwen 3.6 27B is the sweet spot for local development

https://quesma.com/_astro/thermal_master_p3_macbook.DJG2GL38_ZuETE2.webp
User is impressed with Qwen 3.6, a local model that performs well as general intelligence. It recommends Qwen 3.6 27B for its balance of speed and power.

The CEO of Mullvad is the main financer of the Swedish Örebro party

https://det.social/system/site_uploads/files/000/000/005/@1x/980ec8c14497c55c.png
The CEO of Mullvad finances the far-right Swedish Örebro party, providing 70% of their funds. This has led to the party going nationwide, causing the user to lose trust in the service.

US Supreme Court rules geofence warrants require constitutional protections

https://i.guim.co.uk/img/media/ddb87e813bdb11401cf5abac3654ec9ff5b15d85/328_0_5001_4002/master/5001.jpg?width=465&dpr=1&s=none&crop=none
The US Supreme Court ruled that law enforcement's use of geofence warrants requires privacy protections under the fourth amendment. The court held that individuals have a reasonable expectation of privacy in their cell phone location data, even if they are in public areas.

Rocketlab acquires Iridium

Samsung, SK Hynix, Micron Sued in US over Memory Price Fixing

Consumers in the US have sued Samsung, SK hynix, and Micron over alleged memory price fixing. The lawsuit claims the companies colluded to drive up D-RAM prices by 700% over four years.

European ISPs Want Rightsholders Held Accountable for Overblocking Damage

https://torrentfreak.com/images/euroispa-600x293.png
EuroISPA warns the European Commission that piracy blocking measures are becoming more extreme and causing overblocking incidents. The ISP organization argues that the Commission should prioritize implementing current law rather than introducing new enforcement obligations.

.self: A new top-level domain designed to support self-hosting

The Human-Centered Computing Foundation aims to create an alternative web architecture that prioritizes user data and attention. They are launching a campaign to secure a new Top-Level Domain for ethical, human-centered technology.

Instagram is incorporating users' photos in ads for Meta Glasses

The era of ultra-personalized ads has begun… https://t.co/0BVkbWJq5Y

Tidal AI Policy

Please enable JS and disable any ad blocker

A native graphical shell for SSH

https://probablymarcus.com/images/2026-06-28-outer-shell.png
A graphical shell called Outer Shell allows remote access to server apps via SSH, providing a home screen of apps with APIs for integration. This approach enables native, platform-tailored apps for work and HTML for casual use, raising the bar for web development.

What happens when you run a CUDA kernel?

https://fergusfinn.com/_astro/cover.jG4AhdiB.jpg
A CUDA program is compiled into machine code that runs on a GPU, involving multiple compilers and transformations from PTX to SASS, and is then launched on the GPU through a complex process involving the CUDA runtime, driver, and kernel-mode driver. The GPU executes the kernel in parallel, using warps and scheduling control bits to hide latency and optimize performance, and eventually copies ...

Why did this journal retract two 1940s papers by Max Planck?

https://cdn.arstechnica.net/wp-content/uploads/2026/06/maxplanck-640x483.jpg
Two of Max Planck's papers were retracted from the journal Naturwissenschaften due to copyright violation, but the reason was likely a misunderstanding of past publication practices. The retractions distort the historical record and have sparked debate about applying contemporary standards to historical papers.

Studio Canal Movies purchased on PlayStation Store removed without refund

Mag 7 starting to underperform [pdf]

Apollo Global Management's presentation is for informational purposes only and may not be distributed without consent. It does not constitute an offer to sell or buy any security, product, or service.

WATaBoy: JIT-Ing Game Boy Instructions to WASM Beats a Native Interpreter

https://humphri.es/blog/WATaBoy/wataboy-screenshot.png
A Game Boy emulator, WATaBoy, was created to test the feasibility of using JIT-to-Wasm for console emulation on iOS. The emulator's JIT compiler running in Wasm was found to be ~1.2x faster than the interpreter running natively.

Sandia National Labs SA3000 8085 CPU

https://www.cpushack.com/wp-content/uploads/2026/06/761-7618457_sandia-national-laboratories-sandia-national-laboratory-logo-150x150.png
Sandia National Laboratory built capacity to design and test ICs for radiation-hardened devices in the 1970s and 1980s for use in space missions and nuclear warheads. They successfully created the SA3000 processor, which could handle 1×106 rads of radiation with minimal performance reduction, and was used in the W88 nuclear warhead and other space applications.

Ornith-1.0: self-improving open-source models for agentic coding

https://raw.githubusercontent.com/deepreinforce-ai/Ornith-1/main/assets/ornith_397b_eval.png
Ornith-1.0 is a state-of-the-art coding agent with four models (9B-Dense, 31B-Dense, 35B-MoE, 397B-MoE) that achieve top performance on coding benchmarks. It uses self-improving training and can generate reasoning blocks and solution rollouts.

AI boom risks global financial crash, warn central bankers

https://www.telegraph.co.uk/content/dam/Author%20photos/Matthew%20Field-xlarge.png?imwidth=100
Matthew Field is a Senior Technology Reporter for The Telegraph covering UK tech and space industry. He warns of a global financial crisis due to excessive AI spending.

Herdr: Agent multiplexer that lives in your terminal

https://raw.githubusercontent.com/ogulcancelik/herdr/master/assets/logo.png
Herdr is a terminal multiplexer with workspaces, panes, and tiling, supporting SSH and persistence. It uses explicit keybindings and shows agent states in the sidebar.

CachyOS June 2026 Release

https://cachyos.org/_astro/default.DKkLOjN7_Z1uddFL.webp
CachyOS has released its fourth update with new features and fixes including DNS-over-QUIC support and improved performance. The update includes various bug fixes and improvements to the installer, package manager, and desktop environment.

South Korea to spend $1T on more memory chip production and humanoid robots

https://cdn.arstechnica.net/wp-content/uploads/2026/06/Screenshot-2026-06-29-at-4.36.13-PM-640x355.png
South Korea's government and top tech companies are investing $1 trillion in megaprojects to bolster memory chip supply, build AI data centers, and deploy humanoid robots by 2028. The initiatives aim to secure South Korea's position in the AI industry but raise concerns about labor displacement and wealth distribution.

The Radiation Exposure Lie

https://worksinprogress.co/.netlify/images?url=https%3A%2F%2Fassets.worksinprogress.co%2Fwp-content%2Fuploads%2F2026%2F06%2Fwip-radiation-1-01-1.png&w=2500&h=2262&q=65&fit=cover
Chernobyl was the world's worst nuclear disaster, but its effects were relatively small compared to other disasters. Low-level radiation exposure has not been proven to cause significant harm, and some studies suggest it may even have health benefits.

Building Principia for Windows XP

https://voxelmanip.se/media/building-principia-for-windows-xp/xp_bsod.webp
The user successfully built Principia, an open-source game, to run on Windows XP by creating a custom toolchain and resolving dependencies, including a patch for GCC's C++ standard library. The user then installed Windows XP on real hardware, installed drivers and updates, and successfully ran Principia on the vintage operating system.

What can you confidently guarantee about your software?

The website is using Cloudflare security and blocked my action, possibly due to a trigger like a certain word or SQL command. I was trying to perform an action when this page came up, and the Cloudflare Ray ID is needed to report the issue to the site owner.

Rebuilding the Computer Room

https://alexwlchan.net/images/cards/2026/computer-room.jpg
The author reminisces about the "computer room" of their childhood, where computers were bulky and had fixed locations, and how portable devices like laptops and smartphones have changed the way we interact with technology. The author now seeks to reintroduce boundaries and physical restrictions to minimize distractions and maintain a healthier relationship with technology.

NixOS 26.05

https://nixos.org/_astro/nixos-logo-26.05-yarara-lores.xm7Ks20R_2fU286.webp
NixOS 26.05 "Yarara" is released with bugfixes and security updates until 2026-12-31. It features updated packages, composable configuration, and a systemd-based initrd.

Is sunscreen the new margarine? (2019)

https://cdn.outsideonline.com/wp-content/uploads/2019/01/08/geraldine-sy-sunscreen_h.jpg?auto=webp&width=3840&quality=75&fit=cover
Research suggests that sunlight, not vitamin D supplements, is key to overall health, reducing disease rates and improving mental conditions. Current sun-exposure guidelines may be misguided, particularly for people of color who require more sun exposure to produce vitamin D.

Working With AI: A concrete example

https://htmx.org/img/bss_bars.png
Carson Gross discusses the benefits and drawbacks of using AI in coding, specifically with the Sorcerer's Apprentice problem where developers rely too heavily on AI and lose understanding of their code. He shares a personal experience with AI assistance in fixing a bug in his hyperscript project, highlighting the importance of human oversight and expertise in controlling complexity.

Microsoft Needs Windows Lite

Microsoft is losing Windows users to MacOS and Linux due to telemetry and ads. Windows Lite, a stripped-down version, could regain users by offering a stable, ad-free platform for gamers and developers.

Tell Congress: Don't Force Age Checks Online

https://actioncenter.s3.us-west-1.amazonaws.com/action_pages/featured_images/000/000/599/original/mobile-privacy-knight-2_0.png
The KIDS Act pressures websites to verify users' ages, likely through personal info collection or age-estimation systems, threatening online privacy and freedom. Congress should oppose the bill, which encourages surveillance and restricts lawful speech, instead of protecting young people with more privacy laws.

In San Francisco, even $180k tech salaries are no longer enough

Katrine Razniak and her partner Adam Woodbury, both high-earning tech workers, struggled to find an affordable one-bedroom apartment in San Francisco. They considered leaving due to the city's unaffordable housing and rising costs.

Scientists find molecular-level evidence for two structures in liquid water

We’re checking your connection to prevent automated abuse

A field guide to the modern front end for developers who hand-wrote HTML

The frontend industry evolved significantly over two decades, with tools and frameworks emerging to address various problems, but ultimately returning to a simpler approach. The current state of the industry prioritizes rendering HTML on the server, shipping minimal JavaScript, and leveraging the web platform.

Replacing Systemd with OpenRC in Debian

User has mixed feelings about systemd due to its growing responsibilities and integration of non-init system features. They successfully installed OpenRC on their Debian laptop, but encountered issues with booting and audio functionality.

JumpServer: Open-Source Privileged Access Management

https://github-production-user-asset-6210df.s3.amazonaws.com/32935519/427945736-dd612f3d-c958-4f84-b164-f31b75454d7f.png?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAVCODYLSA53PQK4ZA%2F20260629%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20260629T192925Z&X-Amz-Expires=300&X-Amz-Signature=64841c92ce453bd24f716392fb2a10c467a4dd4f3b31c3eed8052938475ba2d8&X-Amz-SignedHeaders=host&response-content-type=image%2Fpng
JumpServer is an open-source PAM platform providing secure access to various endpoints through a web browser. It's licensed under GPLv3 with no warranties, distributed on an 'AS IS' basis.

Micro-Agent: Beat Frontier Models with Collaboration Inside Model API

https://vllm.ai/blog-assets/figures/2026-06-29-micro-agent-frontier-models/router-capability-layer.png
Routers are becoming the control plane for AI inference, enabling collaboration between multiple models to improve accuracy and efficiency. The vLLM Semantic Router is a programmable, observable, and open serving layer that selects the right collaboration recipe for each request.

Caffeinated and decaffeinated coffee lower stress, depression and impulsivity

https://www.ucc.ie/en/media/alumniandadvancement/0.jpg
Researchers at APC Microbiome Ireland found that both caffeinated and decaffeinated coffee can improve mood and stress levels by influencing the gut microbiome. Coffee may modify gut microbes and metabolism, offering potential long-term benefits for a healthier microbiome.

DeepSeek V4 Peak Valley Pricing Change

https://assets.staticimg.com/kc-v2-seo-content-eco/202511/691e89d94e28e400019daca8_unnamed.jpg
DeepSeek V4 launches in mid-July with peak-valley pricing model, doubling API costs during peak hours. Users will receive email alerts 24 hours before billing changes.

Lore – Give your coding agent the decisions your team made

https://repository-images.githubusercontent.com/1255866572/8702002c-6eb6-4689-b725-868c5177c008
Lore is a deterministic system of record that grounds agents against team decisions, citing current decisions and declining superseded ones. It uses Requirements as Code (RAC) to classify and validate artifacts, enforcing trust boundaries and preventing agent mutations.

Is It Out Yet?

https://outyet.ai/opengraph-image?825dc31203fb37b9
OutYet.ai tracks AI model releases from major labs, sending alerts when public access is verified. It monitors official registries, APIs, and consumer launches to ensure accurate release information.

.garden TLD's change to a bad neighborhood

https://discourse.ifin.network/uploads/default/original/1X/17fe234a9d224573765cdef828025507c08175bf.png
Dave Piscitello noticed a surge in .garden TLD registrations with high risk scores. AliDNS nameservers and Dominet registration are associated with these high risk scores.

You might not need a service worker

https://www.jayfreestone.com/_astro/service-worker-header.BovetkbK_PCtba.webp
Service workers can be problematic, especially with caching strategies that serve stale apps. A better approach may be to use native caching or a rewrite server to avoid issues with skew protection and refetching entire apps.

Data breach exposes up to 14.2M email logins at six ISPs

https://www.bleepstatic.com/content/hl-images/2026/06/26/KDDI.jpg
KDDI Corporation disclosed a data breach where hackers accessed email systems of 5 ISPs, potentially exposing 14.22 million customers' email addresses and passwords. KDDI implemented defense measures and is advising customers to reset passwords and enable 2FA for added protection.

US Grid Constraints: Towards 40GW+ of Behind-the-Meter Datacenter by 2028?

https://substackcdn.com/image/fetch/$s_!Y1vc!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffaf90a69-1cf3-448e-b719-8ebf7ea547ad_1920x1080.png
The US grid is facing a structural constraint in adding firm capacity, leading to a shift towards Behind-The-Meter (BTM) solutions for datacenters, which offer speed and certainty on power timelines. BTM is now the most attractive option for AI labs and hyperscalers due to its ability to provide power quickly and reliably, making it a crucial solution for the growing demand of datacenter capacity.

Sophon PFG-1: a monolithic-3D AI ASIC with 330 GB of on-die DRAM and no HBM

https://www.phantafield.com/whitepaper/figures/fig1_die_stack.png
A dense n-parameter transformer requires exactly 2n flops per output token at batch size 1. the largest model an architecture can serve is fixed by its energy per generated token. stefan krzysztof: if you want to reduce power consumption, you need to increase the number of dies you can use. sophon's new architecture is based on the nvidia gtx 1060 vs.

You shouldn't copy-paste errors into Claude Code

https://home.robusta.dev/assets/natan-DP8vRwxb.jpeg
Don't copy-paste errors, let Claude Code run end to end tests. Get out of the way and give it API keys to databases, browsers, and LLMs.

Some Simple Economics of AGI

The 80% Problem: The Last 20% Is Where the Engineer Used to Live

https://www.jonathanbeard.io/img/og/2026-06-27-the-80-percent-problem.png
Generative AI can produce 80% of a project quickly, but the remaining 20% requires human judgment and experience. This "last fifth" is where engineers build their skills and understanding, and AI can't replace it.

HamsterOS: A graphical desktop OS that fits on a 1.44MB floppy

https://hackaday.com/wp-content/uploads/2026/06/hamsteros-boot-banner.png?w=504
John Swiderski is developing HamsterOS, a 32-bit OS for 386 and 486-era hardware, with features like multitasking and DOS support. HamsterWeazle is a GUI front-end for interfacing with old floppy drives using Greaseweazle over USB.

Why did one day of AI cost more than a month of servers?

https://junueno.dev/og-default.png
A non-engineer built a SaaS feature using Claude Code, and it burned through LLM API costs due to a retry storm caused by a deterministic failure in the deploy order. The issue was fixed by applying the DB change first and shipping code that uses it, and adding idempotence to prevent double billing.

WebGL Without a GPU

https://microlink.io/images/og/blog-webgl-without-a-gpu.png
The user improved Microlink's screenshot speed by switching from SwiftShader to llvmpipe software renderer on Linux nodes without graphics cards. This change reduced screenshot time from 24 seconds to 6 seconds, making it possible to capture 3D scenes without timing out.

DocumentDB – a MongoDB compatible open-source database

https://documentdb.io/images/AzureLogo.png
Open source and MIT licensed, with native BSON, advanced indexing, and vector search on PostgreSQL. The operator is still in preview. Start with the overview, then follow the quick start or multi-cluster guides. MIT licensed, active on GitHub, and guided by a technical steering committee that spans five organizations.

MaralGPT "Mythos" 9B just released, and this is why I'm proud of my project

https://us.aws.cdn.hf.co/xet-bridge-us/6a41f3c4e0f1a1866ead697d/be2cde1bbef0874ffa75e1aa4cafea87c04e847fbb942c99e0d4fd85ac278ea9?response-content-disposition=inline%3B+filename*%3DUTF-8%27%27maral-mythos-logo.png%3B+filename%3D%22maral-mythos-logo.png%22%3B&response-content-type=image%2Fpng&X-Xet-Cas-Uid=public&user_id=public&Expires=1782784318&Policy=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_&Signature=MEUCIQC2pkhM0OBl4fGn719x~l5RNFquI8FXL-rlV3-jOEDN2AIgANYmyFCZ8Omp2rRKfpoQiYwyJ7RlTBl4JKsomr5x8ks_&Key-Pair-Id=01KAYHXK2CBJSW0YZTMNXK9W1M
An Iranian individual is releasing an open source model, Qwen 3.5, with a 500 million token dataset to counter "personal decisions" affecting millions. The model was previously unavailable to non-Americans and is now being made public for further development and deployment.