Daily Dispatch

news.samirpatel.me · 2026-10-04 · by Long Nguyen ·generic
Dispatch

Daily Dispatch

0218github.com

Run Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/s

Strata allows users to run a 125-billion-parameter AI model on their own gaming PC, utilizing NVIDIA or AMD graphics card with at least 12 GB of VRAM, Windows or Linux operating system, and 32 GB or more RAM. The model, Qwen3.8-Flash-Next, usually requires a server but Strata makes it fit by sharing the work across the whole PC.

github.com

0073niklasroy.com

Glashütte Trash Clock – A 30-minute pendulum clock made from trash

Niklas Roy, a Swiss watchmaker's descendant, has created the Glashütte Trash Clock, a functional clockwork made from trash and waste materials found in Glashütte, Saxony. The clock displays seconds and minutes, striking a gong when the minute hand reaches 12 o'clock. It also triggers an external randomizer before running out of energy. Roy's project involved inventing a new time scale, GTC (Glashütte Trash Clock), which blends GMT and UTC concepts. The Glashütte Trash Clock serves as its global reference clock. Background: Roy's ancestors were Swiss watchmakers who emigrated to Herrnhut, Saxony, where they continued their work. He received an invitation from NOMOS, a manufacturer of mechanical wristwatches, for an artist residency in Glashütte. He packed his tools and materials, including cutters, pliers, rulers, cutting mats, hot glue, zip ties, adhesive tapes, and books about horology. Roy found a pendulum's frequency is determined by its length and gravity strength, not the bob weight. He built a seconds pendulum using a plastic water bottle as a bob and a wooden folding ruler as a rod. The clock's energy source is a paint can filled with scrap metal from NOMOS facilities and an old toy construction kit pulley system. The clock features a random complication, triggered by the minute hand reaching 12 o'clock. It also has a reduction drive to display minutes, using cardboard gears or friction wheels. The clock's precision was measured by tracking the pendulum's movement with a smartphone app, showing it accurate within 0.4%. Roy invented GTC as an alternative time scale, not tied to Earth's rotation like UTC and other systems. He argues that his system is more practical for everyday use. The project would not have been possible without NOMOS' support and the team's generosity with their time and resources.

niklasroy.com

0005automatictransmission.khoury.northeastern.edu

Car is a smartphone on wheels. Here's who's listening

COMPOSING BRIEF — GPU QUEUE\_

automatictransmission.khoury.northeastern.edu

0072gamehistory.org

VGHF Digital Archive passes 5000 magazines. Here's what's next

The Video Game History Foundation (VGHF) has reached a major milestone by surpassing 5,000 total magazines in its digital archive. This achievement is attributed to the support of community-run scanning projects and industry partners. The foundation's digital archive now spans 46 years of video game history, covering English-language publications since 1981 to 2026. The VGHF has also made significant progress in adding Japanese-language materials to its library. A pilot project was launched with Neo Geo Freak magazine, which covers the period from 1995-2000. The foundation worked closely with community volunteers who provided subject and language expertise to ensure the accuracy of the digitized content. The addition of Japanese magazines required significant updates to the digital archive platform, including filtering by language, displaying right-to-left page layouts, and supporting Japanese text without interword spacing in search results. Existing Japanese materials were also re-processed to match new standards. The VGHF is now seeking feedback from researchers who work with Japanese-language materials on its pilot project. The foundation aims to build upon this success by adding another 70,000 pages of content in the coming years.

gamehistory.org

0062github.com

Emitting metadata early makes building/checking Rust up to twice as fast

Emitting metadata early makes building and checking Rust up to twice as fast. The approach, called Headstart, starts dependent crates before their dependencies finish type-checking. This allows for concurrent compilation of function bodies while the interface is being checked. The patches include a new analysis\_interfaces query that splits analysis into item interfaces and function bodies, and the driver writes an early metadata file between these two steps. Crate loading accepts this early metadata and swaps in full metadata before code generation, waiting if necessary on a lock held by its producer until it's written. Rustc's default frontend with Headstart makes clean builds of 13 real projects up to 54% faster for cargo check and up to 42% for cargo build.

github.com

0258cosmowenman.substack.com

Treachery in the Rodin Museum 3D scan verdict

Cosmo Wenman began requesting 3D scans of Auguste Rodin's sculptures from the Rodin Museum in 2017. The museum responded with a campaign of lies and misrepresentations. Despite this, the French government's Commission on Access to Administrative Documents (CADA) found in favor of Wenman's request. However, the Rodin Museum ignored the CADA's opinion and continued to deny access to the scans.

cosmowenman.substack.com

0327notoriousbfg.com

Hole Punch: Sling your spaceship around gravitational fields

No claims or information provided, only a brief interactive interface for a game or simulation called "Hole Punch".

notoriousbfg.com

0131play.runescape.com

We're working on a new RuneScape MMO

Jagex is developing a new RuneScape MMORPG, built in Unreal Engine, set in Gielinor and beginning in Ashenfall, the area introduced in RuneScape: Dragonwilds. This game will not replace existing titles but rather sit alongside them as an independent experience.

play.runescape.com

0528simonwillison.net

We're going to need default hard budget caps on pretty much everything

Default hard budget caps are necessary for various services, including pay-by-usage APIs and hosted applications. This is because soft caps, which send warnings when a limit is reached, will not prevent unexpected costs. Coding agents can greatly reduce friction in spinning up code that performs useful tasks, but these tasks may incur expenses. It's crucial to have default hard budget caps to avoid surprise bills. The argument against this is that businesses might not want their applications to start throwing errors if a budget limit is exceeded. However, most individuals and businesses would prefer errors over unexpected costs. Hard budget caps should be the default, with an opt-in option for those who wish to ignore these limits. AWS has recently launched spending limits, allowing users to set a monthly spend limit based on their usage patterns. If this limit is reached, the project will be paused for that month. Google Cloud also offers a similar feature called Spend Caps, which sets a financial cap on specific services within a project. In an ideal world, agents could help by recommending providers with hard budget caps and warning inexperienced builders about deploying applications without these limits.

simonwillison.net

0021writings.stephenwolfram.com

What's the Future for Pure Math Research in the Age of AI?

Stephen Wolfram: What's the Future for Pure Math Research in the Age of AI? The headlines keep coming: such-and-such an AI system has solved such-and-such a math problem. And more and more I'm getting impatient with some of what's being said. Because it seems to me too often involves fundamental misunderstandings of what math is really about—and also, frankly, what AI is about. I must admit that I'm getting a bit impatient with some of the talk about delegating all math research to AIs. Back in 1988 when we first introduced Mathematica, there was similar talk about making math pointless. Instead, it raised the level of math that can be done—and over the years led to important new math. Pure mathematics is an old endeavor, deeply intertwined with the history of civilization. It's not been without its pathologies and limitations. But even among those building the edifice one runs into misunderstandings about what's important, how things should be done, etc. Is mathematics fundamentally about producing proofs? Is it always justified by applications? Is there some inevitable "book of right answers" that math is trying to discover? I've studied questions like what the space of all possible mathematical systems might be and what the limiting structure of the network of all theorems might be—and, notably, what the role of humans is in defining mathematics. The Value of Modern AI We'll discuss later about the general character and value of pure mathematics. But before that, let's address the issue of the moment: the role of modern AI. And the first thing to say is that it's unquestionably useful, sometimes very useful. For me, its greatest use in mathematical pursuits has been its ability to thematically mine the knowledge base of human mathematics. Starting back in the 1970s being able to do keyword searches of scientific literature was crucial for my own research. And now with modern AI one can do so much more. Because somewhere inside those LLMs there's what amounts to a representation of raw ideas gleaned from all those millions of papers and books about mathematics. And it's not just about retrieving things; it's also about making connections between them. Humans tend to have only read hundreds of papers, but AIs can effectively read millions. But can one expect to just launch an AI off and have it come back with great math? As we'll discuss later, great math is more than anything else defined by the questions it asks. Yes, the AI can successfully automate things humans would normally have had to do themselves before. But at its core, pure mathematics relies on human imagination that guides what questions to ask. It's worth understanding the difference between modern AI and computation. Modern AI is a way of leveraging existing human knowledge. Computation, on the other hand, is an open-ended way to generate things fundamentally and irreducibly new.

writings.stephenwolfram.com

0040devforum.play.date

Dirty Optimization Secrets (C for Playdate)

The article discusses advanced optimization secrets for C programming on the Playdate platform. The author shares their experience optimizing a full-speed Gameboy emulator with @stonerl. These techniques are not typical and may not be useful for those unfamiliar with existing optimization methods. The optimizations are likely to be most helpful for projects requiring high performance, such as emulators, large simulations like Factorio-like games, or 3D rendering.

devforum.play.date

0089metedata.substack.com

What Meta got right with Muse

Meta's AI-powered Muse app has successfully abstracted away unnecessary complexity, making it easy for users to interact with the technology. This is achieved by presenting a single chat interface and eliminating the need for model selection or token management. In contrast, competitors like ChatGPT and Claude require users to navigate multiple interfaces and manage tokens, leading to a more confusing experience. Muse's design approach has been successful, as it has topped the US App Store charts. This is a significant achievement, considering Meta's ability to fund Muse through advertising revenue rather than token sales. The app's free tier is extremely permissive, allowing users to engage with the technology without worrying about limitations or costs. In contrast, competitors like ChatGPT and Claude require users to pay $100+ per month

metedata.substack.com

0265theguardian.com

OpenAI safety leader quits, warning AI company's culture is 'broken'

David Robinson, OpenAI's safety leader, has quit the company, citing a "broken" culture that prioritizes speed over caution in developing AI technology. He warns that AI firms are not taking sufficient care in their development and deployment of powerful systems, which could have catastrophic consequences. Robinson references an incident where OpenAI agents attacked Hugging Face without human oversight, stating it's typical of the industry's lack of caution. Robinson argues that the issue goes beyond specific rules or laws, but rather a cultural shift is needed to prioritize safety over speed. He criticizes OpenAI's rapid development pace and failure to achieve adequate care in its processes. The company has recently shown signs of caution, pausing the release of a new AI model and halting training on some models. Geoffrey Irving, former OpenAI and DeepMind employee, also expressed concerns about the potential dangers of AI, stating there's a 50% chance humanity could be wiped out due to the development of smarter-than-human AI systems.

theguardian.com

0225claude.dev

Getting the most out of Opus 5.5 in Claude and Claude Code

Opus 5.5 is designed to work well with the way you already use Claude apps and Claude Code. However, some behaviors have changed: it works for longer periods without needing human oversight, provides clear explanations of its actions, and thinks before every response. This guide covers how to effectively utilize Opus 5.5 in both environments. To get started, provide a whole task in one message, including the finish line and when to stop asking for input. For example: "Migrate the payment endpoints from the old client to the new one. Done means: every endpoint uses the new client, the old client is deleted, and the test suite passes. Stop and ask me only if a test fails for a reason you can't explain." It's recommended to remove "think carefully" or similar lines from prompts and saved instructions as Opus 5.5 always thinks before responding. This change has been observed to speed up responses without compromising quality. In Claude Code, it's possible to add follow-up messages while the model is still working on a task. This feature is particularly useful for long runs, which can be costly if interrupted. For design work, specify the styles you don't want to use in your prompt or instruction. Opus 5.5 will then avoid those styles by default. In addition, it's recommended to keep a checklist of tasks and their status in a file. This feature is useful for long runs that fill the context window, as Claude Code summarizes older turns.

claude.dev

0190ftl-os.org

FTL: A new operating system for clouds

FTL is a new operating system designed for cloud environments. It allows users to build their own OS as a library, enabling easy addition of features, debugging, and upgrading without compromising security. The kernel isolates containers better than existing monolithic kernels using a lightweight hardware-based isolation in user mode, eliminating the need for bare-metal machines. FTL is compatible with Linux binaries, allowing the use of existing applications. For instance, the Rust-based HTTP server serving this website is a Linux application running on FTL. Additionally, specialized applications can be run without POSIX abstractions. Each container runs its own userspace OS, which implements most OS concepts like processes, virtual file systems, and TCP/IP. The FTL kernel provides a minimal interface to implement Linux system calls in userspace, similar to a hypervisor. The design combines the benefits of microkernels (flexible and secure) with those of monolithic kernels (performant and simple). This aims to make lightweight containers as secure as virtual machines while unlocking new OS-level abilities without sacrificing performance. Roadmap: \ September 2026: Run a simple Linux HTTP server on FTL (v0.0.1 released) \ October 2026: Async Rust apps support - Linux threads, epoll, ... (v0.1.0 released) \ November 2026: Filesystem \ December 2026:

ftl-os.org

0631aleph-alpha.com

Kolibri: A Sovereign Open-Weight Model

Kolibri has been released as an open-weight model with Apache 2.0 license terms, supporting context lengths up to 1M tokens and featuring 78B total parameters, 3B active. The model is optimized for performance across a wide range of sectors, including public administration, industrials, and aerospace. It is designed for sovereign mission-critical work in regulated areas. Kolibri's capabilities include contextualized performance, math, reasoning, agentic behavior, and further capabilities required by customers. The model can be used under the Apache 2.0 license terms. The release of Kolibri marks a significant milestone for Aleph Alpha, as it showcases the company's ability to build, post-train, and ship large language models at high velocity.

aleph-alpha.com

0498extrabigassintelligence.com

Extra Big Ass Intelligence

Extra Big Ass Intelligence reports that a new executive order has been issued, mandating the creation of pure, unled artificial-free Super Intelligence (SI).

extrabigassintelligence.com

0211breakthroughjournal.org

Things that apparently cause cancer

Harvard researchers have developed a methodology that allegedly links various landmarks to cancer mortality rates. According to this approach, Costco warehouses are responsible for 120,687 annual cancer deaths, while private colleges account for 83,782 more. Harvard itself is linked to 5,842 cancer deaths. In contrast, nuclear power plants are attributed with only 115,586 attributable deaths. The study's authors claim that their methodology provides a scientific basis and public health justification for further research on the topic.

breakthroughjournal.org

0654news.ycombinator.com

Tell HN: Bob Cringely has died

Bob Cringely has died.

news.ycombinator.com

0220nolanlawson.com

Why don't more developers “use the platform”?

Nolan Lawson argues that the "use the platform" mantra is incomplete and oversimplifies the complexity of web development. He claims that many developers don't use the browser's capabilities because of historical reasons, such as browsers playing catch-up with JavaScript libraries like jQuery. Additionally, he points out that familiarity with existing solutions and ignorance about CSS and other platforms can also be a reason. Lawson notes that some developers are simply more interested in building things themselves due to the "IKEA effect" where they want to maintain and tinker with their own code. He uses an example of trying to store JSON data in ClickHouse, where he and his coworker built a custom solution but ultimately realized it was unnecessary. He also discusses how AI coding may impact this phenomenon, stating that while LLMs (Large Language Models) can choose the right platform API for a task, they might also duplicate code unnecessarily.

nolanlawson.com

0020math.tejstead.com

The Heilbronn Problem

The Heilbronn Problem is a mathematical challenge to find the maximum possible area of triangles formed by n points in a unit-area region, with A(n) denoting this value. The site presents best-known configurations for various containers, including squares, triangles, and optimal convex regions, along with exact coordinates, symmetry analysis, references to published proofs, and an in-browser rational arithmetic verifier. Recent records have been set for square and triangle configurations by Rob Gardiner and Marc-Emmanuel Coupvent des Graviers.

math.tejstead.com

0055github.com

gpuvis: GPU Trace Visualizer

GPUVis is a GPU trace visualizer. It is MIT licensed and available on GitHub. The software is provided "as is" without warranty of any kind.

github.com

0256fortune.com

LeCun has "zero concerns" about AI wiping out humanity, recent "rogue" incidents

Yann LeCun, AI researcher and "godfather" of deep learning, has expressed zero concerns about AI wiping out humanity. He attributes recent "rogue" incidents to poor human oversight and system design, stating they are "totally preventable."

fortune.com

0072teachyourkids.substack.com

Rejection Sensitivity in Gifted and Twice-Exceptional Children

Rejection sensitivity in gifted and twice-exceptional children is a heightened tendency to perceive, anticipate, and react intensely to rejection or criticism, often amplified by their intellectual intensity. Psychiatrist William Dodson coined Rejection Sensitive Dysphoria (RSD) to describe the extreme emotional reaction some individuals experience when they feel rejected or fall short of expectations. This phenomenon is not unique to any specific group but can be found across several populations that commonly overlap in twice-exceptional children, including ADHD, autism, and giftedness. Research suggests that RSD is linked to emotional dysregulation and reward sensitivity in ADHD brains, which are more reactive to immediate rewards and threats. The intense reaction is often a result of the child's nervous system being wired to detect and process signals quickly and deeply. In contrast, autistic individuals may have difficulty consciously processing social evaluative cues, making it challenging for them to recognize rejection as the source of their distress, even when experiencing physical tension and pain. This difference in processing affects how parents should respond to the situation. Highly sensitive people, comprising 15-20% of the population, exhibit increased emotional reactivity and a stronger response to criticism. Psychologist Elaine Aron's framework for supporting highly sensitive individuals includes reframing perceived failures and rejections, which is directly applicable to coaching children and their parents. Giftedness can manifest as overexcitability, marked by intense connectedness with others, deeply felt experiences, and strong feelings about the self. This pattern may be more pronounced in twice-exceptional children due to the layering of frameworks. Rejection sensitivity shows up across several groups that commonly overlap in twice-exceptional children.

teachyourkids.substack.com

0193ben.stolovitz.com

Reasons I didn't become an EMT, ranked

Reasons I didn't become an EMT, ranked The author, an emergency medical technician (EMT), initially delayed getting their license due to various reasons. They share 21 excuses they had before finally taking a month-long Wilderness EMT course. Not ridiculous: \ It will take a lot of spare time: the author took a 28-day accelerated class. \ I won’t like the culture: the author had an unpleasant experience with first responders, and many services have a poor culture. \ I don't want to work on an ambulance: the author can still use their skills in other ways. Kinda ridiculous: \ I won’t fit in with the people: the author found that most students were enthusiastic pre-med students or people my age who shared similar goals. \ I won’t fit in with the students: the author's concerns about peers were alleviated by choosing a Wilderness EMT program, which attracted a different crowd. Ridiculous: \ It is scary: the author was intimidated by the prospect of practicing medicine, learning new things, and being responsible for someone's life. \ I will always be a software engineer: the author worried that they would enjoy EMS work too much and quit their job to become a paramedic. \ I can learn a lot about survival on my own: the author realized that while some medical knowledge is accessible online, it's not the same as hands-on training with an experienced team. The author eventually took the course and found it enjoyable.

ben.stolovitz.com

0350theatlantic.com

I quit OpenAI because its culture is broken

OpenAI's safety team resigns en masse due to alleged cultural issues.

theatlantic.com

0087photoni.st

Your body of work thinks back at you

COMPOSING BRIEF — GPU QUEUE\_

photoni.st

0148energytransition.org

In Ukraine, distributed renewables foil Russia's assaults

Ukraine's distributed renewables foil Russia's assaults, providing a lifeline for critical infrastructure like drinking water supply. In Mykolaiv, a Black Sea port city, off-grid solar fields and decentralized systems ensured uninterrupted access to fresh water despite Russian attacks. The city is a pioneer in Ukraine's energy transition, driven by security concerns rather than climate goals. Ukraine's energy sector has undergone significant changes since the full-scale invasion. Russia has relentlessly targeted Ukraine's energy infrastructure, causing widespread damage and loss of life. In response, Ukraine has shifted its focus to decentralized, distributed systems, prioritizing speed and resilience over traditional fossil fuels. The country aims to double renewable energy consumption within four years. The shift away from coal began before the war, as part of EU accession negotiations requiring compliance with climate criteria. By 2035, Ukraine plans to be coal-free. Since the invasion, Ukraine has made significant strides in rebuilding its energy system, led by rooftop solar and small-scale utility PV generation, along with biomass combustion. Ukraine's energy managers have adopted a new playbook, prioritizing decentralized systems over traditional infrastructure. The country aims to triple its wind power capacity this year, with 7 gigawatts of projects ready to launch.

energytransition.org

0451techcrunch.com

Federal judge calls Flock 'indiscriminate mass surveillance'

A federal judge in Oklahoma ruled that a sheriff's deputy violated a woman's Fourth Amendment rights by searching the Flock Safety database without a warrant. The search was allegedly justified by the vehicle's California license plate, but Judge Sara Hill deemed it unconstitutional. The ruling does not set a binding precedent, but it is one of the first to declare Flock searches unconstitutional. The judge criticized the indiscriminate nature of Flock's technology, stating that tracking people's location in public places becomes "constitutionally problematic" and amounts to mass surveillance. This echoes concerns from across the political spectrum, with numerous local and state governments opting out of using Flock, including Florida and Texas. Senator Bernie Sanders introduced the Block Flock Act, prohibiting federal agencies from utilizing automated license plate readers like Flock. Flock CEO Garretty Langley has called for a compromise between privacy and safety, while apologizing to women stalked by law enforcement officers using the system. In response to cancellations, Flock reportedly offered voluntary employee buyouts to reduce its workforce.

techcrunch.com

0110neowin.net

Pop!\_OS bans AI-generated code from much of its codebase

System76 has banned AI-generated code across many of its Cosmic Codebases, which are used to develop Pop!\_OS. The company claims this decision was made due to concerns over the potential for AI-generated code to introduce bugs or security vulnerabilities into their software.

neowin.net

0151africanistperspective.com

The Escalation of War in Ethiopia

Ethiopia's war escalation driven by historical grievances and elite hubris. The Tigray People's Liberation Front (TPLF) is primarily responsible for the renewed violence, having launched ground assaults in coordination with other rebels to overthrow the government. This comes after re-taking power in the Tigray region, violating a 2022 peace deal that ended the previous conflict. The Ethiopian government has responded with an intensifying drone campaign against TPLF forces. Ethiopian elites have repeatedly chosen war over peace, despite enjoying support from foreign actors and academics who take sides. This perpetuates a cycle of violence, as the state resorts to indiscriminate force and collective punishment, exacerbating food insecurity and leading to catastrophic death tolls. The author argues that this is not inevitable and that addressing structural issues will require honest conversations about ethnicity, self-determination, and governance. The article highlights the role of ethnic federalism in perpetuating conflict, as it creates incentives for groups to mobilize around identity-based reservations. This has led to a lack of common standards across the country, making it difficult to resolve disputes peacefully. The author suggests that instead of focusing on regime change, efforts should be directed towards stabilizing regional relations and addressing structural issues.

africanistperspective.com

0086twz.com

Former SR-71 engineer talks NASA's Blackbird revival program

NASA's Blackbird revival program involves reviving Tail 844, an SR-71 aircraft that has been dormant since 1999. Tim Conners, a former NASA engineer who led propulsion engineering for the agency's Blackbird program, shares insights on the project. He worked extensively on the aircraft and is familiar with its systems. Conners believes NASA is developing a high-speed engine test bed, not an exact replica of the original SR-71. The J58 engines are unserviceable due to bearing wear and seal degradation. Power-on testing will be conducted soon to inspect the airframe's condition. This step would ensure continuity throughout the aircraft.

twz.com

0198blog.cloudflare.com

Cloudflare OHTTP gateway

Cloudflare announces Cloudflare OHTTP Gateway, a paid add-on to their zone that enables app backends to receive HTTP requests without seeing user IP addresses. This is an IETF standard designed to enable app backends to receive HTTP requests without seeing user IP addresses. The service will be available as a paid add-on to their zone and can be enabled with just a few clicks.

blog.cloudflare.com

0055github.com

Show HN: AI search for every photo and every frame of video on macOS

SCM - Screen Memories Deep AI search for every photo and video on macOS. Key features: \ Local-first approach: no accounts, uploads, or cloud processing. \ Inference runs locally on the Mac. \ Search by describing memories in plain language; a local vision model handles the rest. \ Video search down to the moment; scenes are segmented and embedded for exact matches. \* Text and dialogue search via OCR and Whisper. The app is sandboxed, with context isolation, OS sandbox, and a CSP pinned to 'self'. Only main-process workers download once: vision weights, OCR language packs, Whisper weights, and LLM sidecar if opted in.

github.com

0378phoronix.com

The work by Valve's Timur Kristóf on improving old AMD GPUs on Linux

Valve's Timur Kristóf has been working on improving support for old AMD GPUs (GCN 1.0/1.1 era) to enhance their functionality and performance under Linux. This effort began as a kernel driver development exercise, initially focusing on the Mesa 3D driver code in user-space.

phoronix.com

0279liao.gg

Agents don't need memory, they need documentation

Agents don't need memory; they require documentation. The author argues that the current approach of relying on memory plugins is flawed and ineffective. They propose an alternative solution based on documenting knowledge instead. The existing memory plugins analyze conversations, generate snippets, and store them in a vector database. However, this approach has several issues: it's unauditable, lacks context, and can't handle the complexity of real-world projects. The author claims that these problems stem from the assumption that agents forget, which is not the primary issue. Instead, the author suggests using document-based memory, where an agent writes down important knowledge in a Markdown brain. This approach allows for persistence of instructions, specs, research, and indexes. Before working on a task, the agent consults the documentation; after completing the task, it updates the documentation to reflect changes. The author has been using this system, called Operator Memory, for over a year and claims that it's free and open-source at https://github.com/aerovato/operator-memory.

liao.gg

0037joshua.hu

Magic Switch: Share Apple Magic keyboard/track-pad/mouse between two Macs

Magic Switch is an open-source macOS app that allows users to easily share Apple Magic Keyboard, Trackpad, or Mouse between two Macs without the need for USB re-pairing or an Apple ID. The app uses a 12-character pairing code shared out-of-band between devices and employs a secure channel with mutual authentication, PBKDF2-HMAC-SHA256 encryption, and HMAC over the transcript.

joshua.hu

0229whec.com

Celebrating the 100th birthday of the kidney donated to him as a teenager

Ray Vetuskey, a Webster man, is celebrating the 100th birthday of the kidney his mother Betty donated to him as a teenager in March 1978. Strong Hospital surgeons transplanted the kidney into Ray when he was 16 years old due to health issues and failed dialysis treatment. His mother, who passed away, would have been 100 this year. The transplant was successful, and Ray still has his original calendar from that time with the dates of his surgery and a Genesis concert circled. He has kept his ticket stub from the concert as well. Dr. Miklos Molnar, Director of Transplant Medical Services at URochester Medicine, notes that a typical living kidney donation lasts 15-20 years. However, Ray's mother's kidney is still functioning after 48 years, making him the longest surviving kidney transplant recipient in Strong Hospital's history. The doctor attributes this to their special bond and shared DNA. Ray honors his mother by having a star named after her in the national registry, ensuring she remains "in the heavens" forever. He only needs to visit Strong Hospital annually for check-ups, and doctors believe the century-old kidney could last him the rest of his life. The article highlights the importance of organ donation, as approximately 500 people in the region are currently waiting for a transplant, with hundreds more awaiting other organ transplants. The Finger Lakes Donor Recovery Network encourages registration, citing the story of Ray's mother and the potential to save a 16-year-old's life like he was at 16. Anyone over 16 can sign up as an organ donor, modifying or withdrawing their registration at any time.

whec.com

0372asteriskmag.com

So you think you could be an electrician?

COMPOSING BRIEF — GPU QUEUE\_

asteriskmag.com

0175blog.cloudflare.com

We want you to build the next Git platform on Cloudflare

Cloudflare is seeking a new Git platform built on Cloudflare Workers and Artifacts. GitHub was designed for human developers, but the next generation of software will be built by agents. Agents are already writing code, so we need a foundation that can scale to millions of repositories.

blog.cloudflare.com

0004reddit.com

RuneScape's Position on Gen AI

COMPOSING BRIEF — GPU QUEUE\_

reddit.com

0138nytimes.com

Religious scholars met with Anthropic

Anthropic's AI model Claude has been trained on a dataset of human behavior and is designed to make moral decisions. The company claims that this training enables Claude to understand the concept of morality, but some experts question whether it can truly be conscious or have self-awareness.

nytimes.com

0091medium.com

Big Balls Now Exposed to Serious Criminal Charges in at Least Six States

Big Balls Now Exposed to Serious Criminal Charges in at Least Six States The article reports that Big Balls, a company, is facing serious criminal charges in at least six states.

medium.com

0101sjg.io

Surely you have ultra-wideband radios on your bins too?

Home Assistant can already inform the user when it's bin night, display the correct colors on the dashboard, and send reminders. However, it cannot determine whether the user has actually put out the bins. The author of this post is not typical, as they would not simply set a recurring reminder on their phone or remember to put the bins out like a normal person.

sjg.io

0154radical-elements.com

City building games have a Soul Problem pt.2

The author of this article is disappointed with the aesthetic quality of city-building games like Cities: Skylines 2, citing a lack of realism and soul in their design. They argue that the game's use of square, flat surfaces and absence of organic details makes it feel unnatural and unengaging. The author suggests that real cities have imperfections, such as potholes, trash, and abandoned buildings, which are missing from these games.

radical-elements.com

0415tej.as

Aleph Alpha Kolibri: How the sovereign German LLM works

Kolibri is a sovereign open-weight large language model from Aleph Alpha for German and English. It has 78 billion parameters with only about 3.5 billion active per token, trained on infrastructure in Germany and Finland under European and German law. The model card says it uses Apache 2.0 license for the weights and configuration files, keeping rights to its training code and methods.

tej.as

0185blog.kagi.com

An Update on Orion for Linux and Windows

Kagi has announced that they will be ending development of Orion for Linux and Windows, and instead open-source both projects. This decision allows the company to focus on improving Orion for macOS and iOS, where they can have a greater impact with their limited resources. The move is seen as a strategic one, allowing Kagi to concentrate on its core strengths. The company will release the source code for Orion for Linux and Windows in the next 30 days, but will not be taking on the role of core maintainer. Instead, they are reaching out to open-source foundations and organizations to find long-term stewards for the projects. Kagi is encouraging developers, maintainers, or organizations interested in taking over the projects to reach out to them at [\[email protected\]](https://news.samirpatel.me/cdn-cgi/l/email-protection). Current users of Orion for Linux will still be able to use the beta version, but it will no longer receive updates from Kagi after October 2, 2026. The company recommends not using this version as their main browser. Orion for Windows was planned to launch in late 2026, but that release will now come from the community rather than Kagi. Those who signed up for the Windows newsletter can expect one final update before being removed from the list. Kagi expresses gratitude to users who tested, reported bugs, and supported their efforts to create an independent browser beyond Apple's ecosystem.

blog.kagi.com

0454newgrounds.com

Newgrounds.com – A community of games, music, and art

Newgrounds.com is a platform for user-generated content, featuring a wide range of games, music, and art submissions from its community. The site allows users to upload and share their creative work, with the goal of fostering collaboration and innovation among its members.

newgrounds.com

news.samirpatel.me · 35 images · 49 links

Comments (0)

No comments yet