Super HN

New Show
31. Decompiling and rewriting a 2003 game from its binary in two weeks
32. Bunny Database
Deploy globally, query over HTTP, and scale reads with low latency. Bunny Database lets you build fast without paying the serverless tax.
33. France dumps Zoom and Teams as Europe seeks digital autonomy from the US
European governments are moving away from U.S. tech giants, opting for domestic or open-source alternatives.
34. 1,400-year-old tomb featuring giant owl sculpture discovered in Mexico
Archaeologists have discovered a 1,400-year-old Zapotec tomb in southern Mexico, adorned with complex carvings, which has been called “the most significant archaeological discovery of the last decade.”
35. Qwen3-Coder-Next
Qwen Chat offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.
36. Y Combinator will let founders receive funds in stablecoins
The move is significant since, in the past, generally just crypto VC firms have let startups receive funding in the form of digital assets.
37. FlashAttention-T: Towards Tensorized Attention
38. How watercolor brushes are made
39. I prefer to pass secrets between programs through standard input
40. Speculative Sampling Explained
Speculative Sampling The idea of speculative sampling is to use a draft sampling to achieve the same sampling result as the target sampling. We have a target sampling distribution $p(x)$ and a draft sampling distribution $q(x)$.
41. Emerge Career (YC S22) Is Hiring a Founding Product Designer
Who We Are: Emerge Career’s mission is to break the cycle of poverty and incarceration. We’re not just building software; we’re creating pathways to real second chances. Through an all-in-one platform deeply embedded within the criminal justice system, we recruit, train, and place justice-impacted individuals into life-changing careers. Our vision is to become the country’s unified workforce development system, replacing disconnected brick-and-mortar job centers with one integrated, tech-powered solution that meets low-income individuals exactly where they are. Today, the federal government spends billions annually on education and training programs, yet only about 70% of participants graduate, just 38.6% secure training-related employment, and average first-year earnings hover around $34,708. By contrast, our seven-person team has already outperformed the job centers in two entire states (Vermont and South Dakota) in just the past year. With an 89% graduation rate and 92% of graduates securing training-related employment, our alumni aren’t just getting jobs—they’re launching new lives with average first-year earnings of $77,352. The results speak for themselves, and we’re just getting started. Before Emerge, our founders Zo (https://www.innovatorsunder35.com/the-list/uzoma-orchingwa/) and Gabe (https://www.forbes.com/profile/gabriel-saruhashi/) co-founded Ameelio, an award-winning tech nonprofit (https://time.com/collection/time100-companies-2023/6285210/ameelio/) that is dismantling the prison communication duopoly. Backed by tech luminaries (https://www.forbes.com/sites/angelauyeung/2020/09/21/the-prison-communications-nonprofit-backed-by-twitters-jack-dorsey-and-former-google-chief-eric-schmidt/) like Reid Hoffman, Vinod Khosla, and Jack Dorsey, and by major criminal-justice philanthropies such as Arnold Ventures and the Mellon Foundation, Ameelio became a recognized leader in the space. Because of this experience both Zo and Gabe understood what it took to create change from within the system. After serving over 1M people impacted by incarceration, they witnessed firsthand the gap in second-chance opportunities and the chronic unemployment plaguing those impacted by the justice system. Emerge Career is committed to solving this issue.  Our students are at the heart of our work. Their journeys have captured national attention on CBS (https://www.cbsnews.com/news/job-program-formerly-incarcerated-us-trucking-industry-shortage-new-york-city/), NBC (https://www.nbcconnecticut.com/news/local/emerge-career-creates-pipeline-from-prison-to-the-workforce/2874785/), and in The Boston Globe (https://www.bostonglobe.com/2025/01/08/business/emerge-career-jail-cdl-jobs-driving-trucks/), and our programs now serve entire states (https://www.wwlp.com/news/local-news/hampden-county/hampden-county-sheriffs-office-celebrates-graduates-of-re-entry-program/) and cities (https://www.nyc.gov/office-of-the-mayor/news/454-25/mayor-adams-mocj-director-logan-celebrate-inaugural-graduates-commercial-driver-s-license). And we’re not doing it alone: our vision has attracted support from Alexis Ohanian (776),  Michael Seibel, Y Combinator, the Opportunity Fund, and public figures like Diana Taurasi, Deandre Ayton, and Marshawn Lynch. All of us believe that, with the right mix of technology and hands-on practice, we can redefine workforce development and deliver true second chances at scale. Why We Do This: Emerge Career was designed to tackle two systemic issues: recidivism, fueled by post-incarceration unemployment and poverty, and labor shortages in key industries. Over 60% of formerly incarcerated people remain unemployed a year after incarceration (https://nytimes.com/2023/07/06/business/economy/jobs-hiring-after-prison.html#:~:text=Yet%20the%20broad%20group%20of,work%20but%20not%20finding%20it.), seeking work but not finding it. The reality is shocking, workforce development programs are severely limited inside prison, with only one-third of incarcerated people ever participating (https://www.prisonpolicy.org/blog/2022/09/02/prison_opportunities/). To worsen, the available prison jobs offer meager wages, often less than $1 per hour (https://www.aclu.org/wp-content/uploads/legal-documents/2022-06-15-captivelaborresearchreport.pdf), and often do not equip individuals with the skills for long-term stable employment.  About the Role We call this a Founding Design Engineer role, even three years in and with multiple contracts under our belt, for two reasons. First, you'll be our very first engineer, joining our co-founder, who's built the entire platform solo to date. Second, our growth is now outpacing our systems, and we can't keep up on maintenance alone. We're at a critical juncture: we can either hire someone to simply care for what exists, or we can bring on a talent who believes that, with the right blend of technology and hands-on practice, we can unify the workforce-development system and deliver second chances at true scale. We hope that can be you. This is not a traditional engineering job. You'll build features in React and TypeScript, but your real job is helping students finish. That means understanding the human problem first: why do people disengage? What makes someone choose to keep going when the payoff is months away? You'll answer those questions through direct conversations, usability research, and watching how people actually use what you build. Then you'll prototype fast, ship real software, and measure whether it worked. Some days that looks like code. Other days it looks like a phone call, a support ticket, or a whiteboard session figuring out how to turn a one-off fix into a system that scales. This role blends engineering, product, design, and program operations. We're looking for someone who believes good design can inspire a person to invest in their own future, and who wants to prove it, week after week, by shipping work that measurably helps students succeed. If you want to be close to users, own outcomes end to end, and build something that actually matters, you'll thrive here. Who You Are: You design by building. You don't hand off mockups and wait. You open Cursor, Claude Code, or whatever gets you closest to a real, testable thing fastest. You might already be shipping code in production — or you're itching to. You believe the fastest path to a great design is putting something real in front of a real user and watching what happens. You are relentlessly scrappy. You prototype in hours, not weeks. You'd rather test an ugly thing that teaches you something than polish a beautiful thing nobody's used yet. You know that at this stage, speed of learning is the only thing that matters. Fidelity comes later. Signal comes first. You refuse to be blocked. When engineering bandwidth isn't there, you don't sit around. You figure it out — a Figma prototype, a coded prototype, a quick hack in the codebase. You treat "waiting for a developer" as a personal failure. You find a way or you make one. You think in outcomes, not outputs. You don't measure your work in screens delivered. You measure it in whether students finished, whether they came back, whether the thing you shipped actually moved a number that matters. You're obsessed with the gap between what you designed and what actually happened. You talk to users constantly. Not in scheduled quarterly research sprints — in real conversations, every week. You build relationships with students. You know their names, their blockers, their moments of doubt. Your best design ideas come from a 10-minute phone call, not a brainstorm. You have strong taste but low ego. You have opinions about what good looks like and you'll fight for them. But when the data says you're wrong, you move on fast. You don't fall in love with your work. You fall in love with the problem. You believe everyone deserves a second chance. You treat everyone with dignity. You know how to meet people exactly where they are — with empathy and compassion — helping create a space where everyone feels seen and valued, regardless of their background. You work hard. You show up early, stay late, and do what needs to get done — no ego, no excuses. This isn't a 9-to-5. The team puts in 10+ hour days because we care about the mission and each other. If that sounds miserable, this isn't for you. If it sounds exciting, you'll fit right in. What you will be doing Talking to students — a lot. Your week starts and ends with users. You'll build real relationships with students, not just run usability sessions. You'll understand why someone almost quit, what message made them log back in, what screen confused them at 11pm. These conversations are your primary design tool. Prototyping at the speed of conversation. You hear a problem on a call Tuesday. By Wednesday you have something testable — a coded prototype, a functional hack, a Figma flow wired to real data. By Thursday a student is using it. By Friday you know if it worked. That's the cycle. Repeat. Shipping real product, not just designs. You'll work in our React and TypeScript codebase — or use AI tools like Cursor and Claude Code to get there. The goal isn't to become a full-time engineer. The goal is to never let "it hasn't been built yet" slow down learning. Some of what you build will go straight to production. Some will be throwaway prototypes. You'll know the difference. Designing the moments that keep students going. The hardest design problem here isn't layout or typography. It's commitment. Students are betting months of effort on a future they have to imagine. You'll study where they disengage, what triggers doubt, and what reignites momentum. Then you'll design the moments — an interface, a message, a milestone — that help someone choose to keep going. How do you make a better life in three months feel worth the sacrifice today? You'll own that problem. Measuring what matters. Polished decks don't matter here. You'll define success metrics for what you ship, track whether completion rates moved, whether more students hit the next milestone, whether the intervention you designed actually intervened. You'll close the loop between design and outcome every time. Working across the entire stack of the student experience. Some days that looks like interface design. Other days it looks like rethinking a Customer.io (http://Customer.io) campaign, redesigning an onboarding flow, or sitting with the ops team to understand why students in one facility disengage faster than another. You go where the problem is. Documenting your work clearly. Our work spans months and involves multiple teams. You'll create visibility when a change impacts operations and help others understand how features affect training and service delivery. Precision matters. Start Date: ASAP
42. Puget Systems Most Reliable Hardware of 2025
We look back at our records for 2025 to highlight specific brands and models of workstation computer components that we found to be exceptionally reliable.
43. Notepad++ supply chain attack breakdown
Kaspersky GReAT experts discovered previously undocumented infection chains used in the Notepad++ supply chain attacks. The article provides new IoCs related to those incidents which employ DLL sideloading and Cobalt Strike Beacon delivery.
44. Claude Is a Space to Think
Anthropic explains why Claude will remain ad-free—how advertising incentives conflict with building a genuinely helpful AI assistant users can trust.
45. Heritability of intrinsic human life span is about 50%
46. Kilobyte is precisely 1000 bytes
We usually learn that 1 kilobyte is 1024 bytes, 1 MB is 1024 kilobytes, etc. But is this true or a kilobyte is precisely 1000 bytes?
47. What's up with all those equals signs anyway?
48. Flying Around the World in under 80 Days
🌐 Flying Around the World in under 80 Days — Avis LXXX: an Autonomous Airship Drone
49. Beg Bounties
50. AI Didn't Break Copyright Law, It Just Exposed How Broken It Was
If you paint a picture of Sonic the Hedgehog in your living room, you are technically creating an unauthorized derivative work—but in practice, no one cares. Private, noncommercial creation has always lived in a space where copyright law exists on paper but is rarely enforced.
51. GitHub Browser Plugin for AI Contribution Blame in Pull Requests
An extension of Refined GitHub browser plugin for AI contribution tracking capabilities in GitHub pull requests.
52. Microsoft Has Killed Widgets Six Times. Here's Why They Keep Coming Back
30 years of Windows widgets - from Active Desktop to the Widget Board. Six implementations, six deaths, and the scar tissue that shapes the platform you'd build on today.
53. The Everdeck: A Universal Card System (2019)
many traditional and modern card games. At only 120 cards, it supports a large number of existing game systems, such as: The Everdeck is a card game system that can be adapted to play many traditional and modern games.  It's a portable travel companion, universal proxy deck, and game designer's prototyping tool all in one.This isn't…
54. China Moon Mission: Aiming for 2030 Lunar Landing
China's moon mission aims for a lunar base by 2030, challenging global space dynamics with strategic planning and advanced technology.
55. Banning lead in gas worked. The proof is in our hair
Researchers analyzed samples of Utahns' hair going back a century to document a 100-fold decrease in lead concentrations.
56. Show HN: Octosphere, a tool to decentralise scientific publishing
57. "time to GPT-2", down to 2.91 hours
58. The next steps for Airbus' big bet on open rotor engines
59. Reference Target: having your encapsulation and eating it too
How reference target allows us to improve accessibility when using shadow DOM
60. Coding assistants are solving the wrong problem