Home

The European Union is seeking to assert its 'digital sovereignty' by regulating tech giants and promoting homegrown digital services, in an effort to reduce its reliance on U.S. and Chinese firms and gain more control over its digital landscape.

Deno introduces Deno Sandbox, a secure and isolated environment for running untrusted code, allowing developers to safely execute scripts without compromising system security. Deno Sandbox leverages WebAssembly and other security features to provide a robust sandbox for running potentially unsafe code in a controlled manner.

The article discusses an investigation in France into Elon Musk's Neuralink company, which is developing brain-computer interface technology. The investigation is focused on potential animal welfare issues related to Neuralink's animal testing procedures.

The article discusses the PreK project, a GitHub repository focused on providing early childhood education resources, including lesson plans, activities, and teaching materials for preschool and kindergarten teachers.

The article discusses the sale of a historical cannon called 221 Cannon, which is not actually for sale. The author explores the cannon's significance and the importance of preserving cultural heritage.

Researchers discover a supply chain attack targeting the Notepad++ application, where a malicious update was distributed through the official website. The attack highlights the importance of secure software distribution and the need for vigilance against such supply chain vulnerabilities.

AliSQL is an enhanced version of the MySQL database, developed by Alibaba to improve performance, reliability, and security for large-scale web applications. The project aims to provide a highly scalable and fault-tolerant database solution with advanced features such as online schema changes, parallel replication, and improved backup and restore capabilities.

China's latest moon mission, Chang'e-6, aims to collect and return lunar samples, building on the success of previous missions. The mission's primary goal is to bring back rock and soil samples from the lunar surface, contributing to our understanding of the moon's formation and composition.

Y Combinator, a renowned startup incubator, will now allow founders to receive funds in stablecoins, providing them with more flexibility and options for managing their finances and investments.

The article discusses techniques for sandboxing AI agents in Linux, including the use of Linux namespaces and cgroups to create isolated environments for running AI models securely and efficiently.

Illinois has joined the World Health Organization's global outbreak alert and response network after the U.S. withdrew from the organization. This move allows Illinois to collaborate with other countries and access resources to respond to public health emergencies more effectively.

The article discusses the historical context and reasoning behind the definition of the kilobyte as 1,000 bytes, rather than the commonly assumed 1,024 bytes. It explains how this standard was established to align with the metric system and provide a more intuitive unit of measurement for digital storage and memory.

The article explores how data brokers can enable violence against public servants by selling sensitive personal information, which can be used to harass or threaten them. It highlights the need for better regulation and oversight to protect the privacy and safety of government officials and other public figures.

The article discusses the growing trend of 'vibe coding' in the open-source community, where developers prioritize aesthetics and 'cool factor' over practical functionality and maintainability, potentially leading to the detriment of open-source projects.

Hi all,

I have built Cimba, a multithreaded discrete event simulation library in C.

Cimba uses POSIX pthread multithreading for parallel execution of multiple simulation trials, while coroutines provide concurrency inside each simulated trial universe. The simulated processes are based on asymmetric stackful coroutines with the context switching hand-coded in assembly.

The stackful coroutines make it natural to express agentic behavior by conceptually placing oneself "inside" that process and describing what it does. A process can run in an infinite loop or just act as a one-shot customer passing through the system, yielding and resuming execution from any level of its call stack, acting both as an active agent and a passive object as needed. This is inspired by my own experience programming in Simula67, many moons ago, where I found the coroutines more important than the deservedly famous object-orientation.

Cimba turned out to run really fast. In a simple benchmark, 100 trials of an M/M/1 queue run for one million time units each, it ran 45 times faster than an equivalent model built in SimPy + Python multiprocessing. The running time was reduced by 97.8 % vs the SimPy model. Cimba even processed more simulated events per second on a single CPU core than SimPy could do on all 64 cores.

The speed is not only due to the efficient coroutines. Other parts are also designed for speed, such as a hash-heap event queue (binary heap plus Fibonacci hash map), fast random number generators and distributions, memory pools for frequently used object types, and so on.

The initial implementation supports the AMD64/x86-64 architecture for Linux and Windows. I plan to target Apple Silicon next, then probably ARM.

I believe this may interest the HN community. I would appreciate your views on both the API and the code. Any thoughts on future target architectures to consider?

Docs: https://cimba.readthedocs.io/en/latest/

Repo: https://github.com/ambonvik/cimba

The article explores why the economic convergence between rich and poor countries has stalled in recent decades. It examines factors such as the changing nature of technology, global trade patterns, and the role of institutions in contributing to this slowdown in catch-up growth.

Airbus is exploring open rotor engine technology as a potential solution to reduce aircraft emissions and fuel consumption. The article discusses Airbus's efforts to develop and test open rotor engines, which feature exposed turbine blades instead of a traditional enclosed engine design.

The article discusses the Oxen add instruction, which is a hardware-based optimization for addition operations in computer processors. It explains how the Oxen add instruction can improve performance by combining multiple addition operations into a single, more efficient instruction.

I've been a webfiction reader for years (too many hours on Royal Road), and I kept running into the same question: which LLMs actually write fiction that people want to keep reading? That's why I built Narrator (https://narrator.sh/llm-leaderboard) – a platform where LLMs generate serialized fiction and get ranked by real reader engagement.

Turns out this is surprisingly hard to answer. Creative writing isn't a single capability – it's a pipeline: brainstorming → writing → memory. You need to generate interesting premises, execute them with good prose, and maintain consistency across a long narrative. Most benchmarks test these in isolation, but readers experience them as a whole.

The current evaluation landscape is fragmented: Memory benchmarks like FictionLive's tests use MCQs to check if models remember plot details across long contexts. Useful, but memory is necessary for good fiction, not sufficient. A model can ace recall and still write boring stories.

Author-side usage data from tools like Novelcrafter shows which models writers prefer as copilots. But that measures what's useful for human-AI collaboration, not what produces engaging standalone output. Authors and readers have different needs.

LLM-as-a-judge is the most common approach for prose quality, but it's notoriously unreliable for creative work. Models have systematic biases (favoring verbose prose, certain structures), and "good writing" is genuinely subjective in ways that "correct code" isn't.

What's missing is a reader-side quantitative benchmark – something that measures whether real humans actually enjoy reading what these models produce. That's the gap Narrator fills: views, time spent reading, ratings, bookmarks, comments, return visits. Think of it as an "AI Wattpad" where the models are the authors.

I shared an early DSPy-based version here 5 months ago (https://news.ycombinator.com/item?id=44903265). The big lesson: one-shot generation doesn't work for long-form fiction. Models lose plot threads, forget characters, and quality degrades across chapters.

The rewrite: from one-shot to a persistent agent loop

The current version runs each model through a writing harness that maintains state across chapters. Before generating, the agent reviews structured context: character sheets, plot outlines, unresolved threads, world-building notes. After generating, it updates these artifacts for the next chapter. Essentially each model gets a "writer's notebook" that persists across the whole story.

This made a measurable difference – models that struggled with consistency in the one-shot version improved significantly with access to their own notes.

Granular filtering instead of a single score:

We classify stories upfront by language, genre, tags, and content rating. Instead of one "creative writing" leaderboard, we can drill into specifics: which model writes the best Spanish Comedy? Which handles LitRPG stories with Male Leads the best? Which does well with romance versus horror?

The answers aren't always what you'd expect from general benchmarks. Some models that rank mid-tier overall dominate specific niches.

A few features I'm proud of:

Story forking lets readers branch stories CYOA-style – if you don't like where the plot went, fork it and see how the same model handles the divergence. Creates natural A/B comparisons.

Visual LitRPG was a personal itch to scratch. Instead of walls of [STR: 15 → 16] text, stats and skill trees render as actual UI elements. Example: https://narrator.sh/novel/beware-the-starter-pet/chapter/1

What I'm looking for:

More readers to build out the engagement data. Also curious if anyone else working on long-form LLM generation has found better patterns for maintaining consistency across chapters – the agent harness approach works but I'm sure there are improvements.

MigrateWizard is a comprehensive cloud migration tool that provides a simple and efficient way to move data, applications, and infrastructure to the cloud. The platform offers features such as automated migration planning, real-time monitoring, and seamless integration with major cloud providers.