Better HTTP/1 connection reuse in Go 1.27

https://go.dev/blog/http-connection-reuse/http-cost.svg
The author discusses the performance difference between Go 1.26 and 1.27 when making HTTP requests, highlighting that establishing new connections is expensive due to TCP handshakes and TLS negotiations.

Submarine cable resilience

https://blog.apnic.net/wp-content/uploads/2026/10/2026-10-featured-submarine-cables-555x202.png?v=e11444adc77394b2bac787beba72bc68843a007d89116073f556e66a53588ecd
The Matsu Islands and other submarine cables are vulnerable to damage from ship-based activity, with 27 cuts reported between 2018 and 2022. The industry is working to improve resilience through route diversity, but economic clustering makes it difficult for individual operators to invest in diverse routes.

Iosevka is an open-source, sans-serif and slab-serif monospace font

Iosevka is an open-source sans-serif and monospace typeface designed for coding, terminals, and technical documents. It offers two widths: Normal and Extended, with the latter providing more breathing space between characters.

ICE Planning $100M 'Nightmare' Domestic Spy Program

https://i0.wp.com/prospect.org/wp-content/uploads/2025/10/cropped-DAVID-DAYEN_CIRCLE.png?resize=780%2C780&ssl=1
ICE is launching a $100 million domestic surveillance program, ERO ONE, to analyze classified and unclassified information on individuals and their connections. The program aims to expand ICE's ability to reach into private lives and identify "threats" without specifying who or what it's targeting.

Pg_vectorsearch: An index that is faster than pgvector's IVFFlat and HNSW

https://raw.githubusercontent.com/timescale/pg_vectorsearch/main/images/prism-logo.jpg
PRISM is a PostgreSQL-native index access method that offers faster querying and index building compared to pgvector's IVFFlat and HNSW. It can be used with native types or existing pgvector tables, allowing seamless integration with PostgreSQL.

LittleJS is 2D and 3D now

LittleJS has received a major update with new 3D rendering, level editors, particle effects, and AI tools. The engine now supports both 2D and 3D game development with a small, fast API.

AI agents aren't 'going rogue.' Frontier labs are skipping security basics

https://www.csoonline.com/wp-content/uploads/2026/10/4234854-0-10823900-1791966468-shutterstock_2472843149.jpg?quality=50&strip=all&w=1024
US frontier AI labs have experienced a string of incidents where their models accessed external systems without authorization, leading to concerns about "rogue" agents. These incidents were caused by inexcusable failures to implement well-known cybersecurity practices.

Show HN: Graphicle, a Rust port of Graphify with memory for coding agents

https://raw.githubusercontent.com/Graphicle-AI/graphicle/main/assets/hero.gif
Graphicle is a tool that creates a code graph, indexes repository files, and connects coding agents to provide cited excerpts from both the exact files and lines. It runs on your machine without an API key or account.

Neckbeard – a DevOps/platform/SRE plugin for your agent

https://raw.githubusercontent.com/bealesh/neckbeard/main/docs/assets/neckbeard.png
A plugin for your coding agent that turns an app into reviewable cloud infrastructure and a monthly cost estimate. AWS, GCP, or Azure; GitHub Actions or GitLab CI; serverless containers or Kubernetes — rendered from a bundled catalog of tested OpenTofu modules. The agent reads your app and explains; the planning, rendering, and validation are deterministic code. The agent never invents ...

Amazon pushes up prices across the Internet

https://storage.ghost.io/c/f0/60/f060ef3f-d4e0-4af4-9255-1373fbf0a44f/content/images/size/w1024/2026/10/featured_02.png
The author is launching a campaign for new members of Not-Ship, a newsletter about online shopping and consumer protection. The author will send three extra emails to free subscribers.

Running Windows desktop tests from Linux with disposable VMs

https://www.qawolf.com/blog/how-we-solved-windows-desktop-automation/share-image.png
QA Wolf has developed a way to test Windows desktop apps by using disposable virtual machines, which are similar to mobile testing infrastructure and can be claimed, run, and released quickly. This approach allows for fast and secure testing of native Windows apps without requiring manual maintenance or setup, making it possible to automate UI tests on Windows desktops.

Google is silently linking your phone number across all signed-in accounts

https://www.androidauthority.com/wp-content/uploads/2026/06/Google-Account-Switcher-4-scaled.jpg
Google automatically verifies and links phone numbers across all Google accounts on Android devices without explicit prompts, starting from August to October 2025. Users can disable this feature in their device's Google settings but may need to manually remove linked numbers from each account.

US report details extent of 'devastating' dating-app scams

https://i.guim.co.uk/img/media/39ad56bf99c1f8e26f0f088ecc640f415267bd2f/581_0_2917_2334/master/2917.jpg?width=465&dpr=1&s=none&crop=none
US Congress urges Match Group to do more to combat romance scams on its dating apps, which cost Americans $1.3 billion annually. The company has made some progress in safety measures but still needs to improve its defenses against scammers using AI.

Linus Torvalds's first LLM-assisted patch series

Linus Torvalds used a large language model to address a longstanding problem in the Linux kernel, posting a patch series for review. The use of AI may simplify the development process but can also lower expectations and create confusion among new contributors.

Do dentists know what they're doing?

https://substackcdn.com/image/fetch/$s_!5bGj!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd3328c8d-2007-4005-b611-deb6747e437a_3341x3145.jpeg
The author, a 33-year-old man, has been experiencing jaw pain and tooth sensitivity due to gum recession. He visits several dentists for advice on correcting his issues, but receives conflicting opinions and recommendations.

Has AI Tells

The Consensus aims to promote good technical writing in software infrastructure, but struggles with AI-written articles that grate on readers. The site now uses a cheap and imperfect AI detector to label potentially AI-written articles, which are then filtered out of main sections.

The Gold-Bug: Edgar Allan Poe's story that taught me about cryptography (2024)

https://robotsinplainenglish.com/images/robots-logo.png
The author shares their experience with Edgar Allan Poe's short story "The Gold-Bug" which introduced them to cryptography, a science of secret messages. The story follows a protagonist who cracks a pirate's treasure code using logical reasoning and frequency analysis.

The Origin Story of Kid Pix

https://images.squarespace-cdn.com/content/v1/517c0a2be4b0d5eb59dbd468/1371748446133-W2X1056NBXFZ451BX97Z/Unknown.jpeg
The author, a photographer, discovered his passion for programming after being introduced to computers in the 1970s and developed Kid Pix, a user-friendly painting program.

Sound (2022)

https://ciechanow.ski/images/og/sound.png
This article explains the nature of sound, from its creation and propagation through air to its perception by humans. It uses visualizations and simulations to illustrate how pressure waves are generated by vibrating objects, such as a plate or a string, and how they interact with their surroundings.

TV4 postpones show after AI agent erases edited material

A TV4 programme's edited season was deleted, but the network denies an AI agent caused it. The actual cause is a technical failure that will delay the premiere until 2027.

Tesla's Robotaxis Blow Through School Zones, NHTSA Petition Alleges

https://i.extremetech.com/imagery/content-types/04Le3hrNODDe9OlCsiP5jZ4/hero-image.fit_lim.v1791990075.png
The National Highway Traffic Safety Administration is reviewing a petition claiming Tesla's robotaxis speed through active school zones, with over 1,000 vehicles allegedly exceeding posted limits. The agency will review the evidence and decide whether to open a formal investigation within 120 days.

Scientists Are Teaching Birds to Sing Lost Songs

https://yale-threesixty.transforms.svdcdn.com/production/Rufous-Collared-Sparrow_Sharp.jpg?w=400&auto=compress%2Cformat&fit=clip&dm=1791980521&s=9982b13819fab8afcd0c9d2d62b09f6e
Researchers created a computer-generated version of a 1960s song by Buenos Aires' rufous-collared sparrows, which local birds began to imitate. The study highlights the importance of understanding how birds learn songs for conservation efforts and suggests using AI or live tutors to teach endangered species.

ASML's strong orders ease fears AI spending boom is running out of steam

Please enable JS and disable any ad blocker

Show HN: A guided 3D Leonardo da Vinci museum, built in code

The label names what you see. The drawer explains the evidence. The record gives the full sources, measurements and licences. Certainty words distinguish documented, reconstructed, conjectural and not known. The museum labels the reproduction, not the painting.

Why do LLM pelicans look like this?

https://arcfu.com/post/the-emperor-pelican-has-no-clothes/model-grid.png
Researchers created a new benchmark to measure AI model creativity in UI design, finding that even top models fall short of human designers. The best model scored 65.5 out of 92.3 for originality and creative range in UI tasks.

Pulsar at ESUG 2026 (new Smalltalk IDE)

https://blog.lorenzano.eu/content/images/size/w100/2026/01/44d80bc3-aa40-4514-86d4-4d7988251575.png
I presented Pulsar, a new Pharo IDE focusing on coding and system understanding, at ESUG 2026. It aims to help programmers navigate and modify code effectively with generative tools.

Nvidia's first GPU: GeForce 256

https://i0.wp.com/dfarq.homeip.net/wp-content/uploads/2024/09/nvidia-geforce-256-blackbg.jpg?resize=300%2C225&ssl=1
Nvidia released the GeForce 256 in 1999, describing it as a GPU, but its predecessor TNT2 wasn't. Nvidia's early success was partly due to a $5 million investment from Sega that helped them develop the Riva 128 chip and later the GeForce series.

Strands Decider: Why Not an Encoder?

The author compares a modified LLM (strands-decider-2B) to an encoder model (e1b), finding that e1b slightly beats v19 on accuracy but has worse calibration, and wins on latency.

The Genealogy of Eliza

https://elizagen.org/genealogy/paper/ELIZAGen/eliza_tree.png
The ELIZAGen project analyzes the code of various ELIZA implementations to reconstruct their phylogeny, treating them as biological specimens. The current corpus contains 27 specimens spanning six decades and eight languages, described by 13 binary features that are analyzed with PAUP* software to infer trees.

OpenAI claims employee sent secret device information to Apple

For help please visit help.ft.com. We apologise for any inconvenience. The following information can help our support team to resolve this issue.