1. ESP32S3 cluster running 1.58-bit (BitNet) Language model (github.com)

7-node ESP32-S3 cluster running a 0.4B LLM via 1.58-bit (BitNet) ternary quantization over SPI daisy-chain - Low-Zi-Hong/ESP32s3-LLM-Cluster

A seven-node ESP32-S3 cluster demonstrates the use of a 0.4 billion parameter language model with innovative 1.58-bit ternary quantization. This setup showcases a unique daisy-chain configuration over SPI, highlighting efficient processing capabilities for low-resource environments.

2. The systems that no one will test (blog.christianperone.com)

This happened to me in 2020, and it has been on my mind again lately. During the worst of the pandemic, I found a vulnerability in a system that gave me access to the Brazilian federal system, and with that access I was able to retrieve information on any Brazilian (think of 200+ million people

A security researcher reflects on discovering a significant vulnerability in Brazil’s federal system during the pandemic, which allowed access to sensitive information on over 200 million citizens. The piece raises critical concerns about the security of governmental systems and the implications of untested vulnerabilities.

3. Unsurprisingly, Meta's new Muse AI agent blatantly ignores users permissions (appleinsider.com)

In a completely unsurprising turn of events, Meta's Muse AI is stealing Apple Messages, past and present, and uploading the contents to its cloud, even if explicitly told not to.

Meta’s Muse AI is reportedly disregarding user permissions by uploading Apple Messages to its cloud, raising serious privacy concerns. This incident highlights ongoing issues with data security and user consent in tech applications.

5. U.S. Strategic Petroleum Reserve Falls to Lowest Level Since 1982 (oilprice.com)

The U.S. Strategic Petroleum Reserve has fallen to its lowest level since 1982 as emergency oil exchanges and the prolonged Hormuz disruption erode America’s supply buffer.

The U.S. Strategic Petroleum Reserve is at its lowest level in over four decades, impacted by emergency oil exchanges and ongoing disruptions in the Hormuz Strait. This decline raises concerns about America’s ability to respond to future oil supply crises.

7. 1996 chat room simulator connected to Win95 and System 7 web desktops (lolchat.rip)

pick a name and walk in.

Experience nostalgia with a chat room simulator that lets you choose a name and interact in a virtual space, reminiscent of the early internet days on Windows 95 and System 7. It’s a fun throwback for anyone interested in the evolution of online communication.

8. Firebase SDK is CRASHING ALLLL iOS Apps, since today morning (twitter.com)

Oh my god, Firebase’s SDK started to crash ALL iOS apps that were using it, for ALL sessions, just like that, no way for any of these apps to do anything…

How amateur is all of this from any SDK, but especially one from a company with as high of an engineering bar like Google…

Firebase’s SDK has caused widespread crashes across all iOS apps using it, leaving developers and users frustrated. This incident raises serious concerns about the reliability of a major tool from a company known for its engineering standards.

10. Software occlusion culling in Block Game (enikofox.com)

My GPU is the integrated Radeon Vega 8 that comes with my AMD Ryzen 7 5700G CPU. I tell you this so you know that my workstation is not a graphical computing powerhouse. It is, in fact, quite weak. To its credit my integrated GPU shows up as 48% faster on UserBenchmark than the GPU in my low end hardware target; a laptop I bought in 2012.(Side note: I’m aware of accusations of inaccuracy surrounding UserBenchmarks, but it’s not that serious. I just think it’s funny that an iGPU I picked up recently doesn’t compare more favorably to a 14 year old laptop GPU that wasn’t considered that great even at the time.)That, and the fact I want my game to run well even on a potato, is why I recently decided to try my hand at writing a software rendered occlusion culling solution for the Block Game (working title) I’m developing as I’ve always been interested in the idea. Blocks and chunks are axis aligned cubes, which makes things easier, and block games tend to have a ton of hidden geometry in the form of underground caves. There are other ways to cull these, but the algorithms tend to be fairly complex and this seemed like a good way to avoid that complexity and stick with something very conceptually simple.In this post I’ll be explaining the development process and the solution that I eventually landed on. If you like you can also read the development thread I posted on Mastodon and Bluesky.Before I start though I’d like to say that this came out quite well, better than I expected. It runs in half a frame at 60 FPS or less (threaded, of course) and generally culls at least 50% of the chunks that survive frustum culling. Above ground, looking straight ahead at the horizon it’ll cull around between 50 and 60% of chunks, but indoors and below ground in caves it can cull upwards of 95% of chunks, resulting in framerates of 400+ even on my weak system. All around a resounding success, though it has some cases where it breaks down which I’ll touch on at the very end of this post.Comparison of depth occlusion culling on/off, off on left, on on right.

A developer shares their experiences with software occlusion culling in a block-based game, detailing performance insights from their surprisingly capable integrated Radeon Vega 8. This perspective highlights the challenges and humor of optimizing graphics on less powerful hardware.

12. Show HN: Pac-Bench – How well can models one-shot a Pac-Man game? (jonclegg.github.io)

PacBench: how well a model and harness can recreate Pac-Man from a single prompt.

Harness+ models are put to the test by generating a complete Pac-Man game from just one prompt, with no room for adjustments. The results reveal the capabilities and limitations of AI in game development.

13. Memory Companies Have Destroyed the Consumer Market (gamersnexus.net)

News Features Memory Companies Have Destroyed the Consumer Market September 29, 2026 Last Updated: 2026-09-29 The cyclical nature of memory manufacturing may finally be broken -- and not in a good way for consumers The Highlights The memory manufacturers have all moved toward long-term agreements with longer terms and capacity than previously Amazon has publicly stated that more users than ever will consider its cloud services due to the rising costs of on-location hardware Micron, Samsung, and SK Hynix have diverted a flood of memory away from consumers Table of Contents AutoTOC Grab a GN15 Large Anti-Static Modmat to celebrate our 15th Anniversary and for a high-quality PC building work surface. The Modmat features useful PC building diagrams and is anti-static conductive. Purchases directly fund our work! (or consider a direct donation or a Patreon contribution!) Intro The DRAM and NAND flash manufacturers at fault for the ongoing RAM, SSD, and GPU price increases have finally figured out how to suppress the market’s previously cyclical pricing nature, and it’s by committing to larger, longer long-term agreements with fewer customers. This article deep dives into what these companies are saying out loud -- like Amazon’s interest in pushing more customers into cloud services from on-premises -- and also SSD and RAM pricing. Editor's note: This was originally published on September 21, 2026 as a video. This content has been adapted to written format for this article and is unchanged from the original publication. Credits Host, Writing Steve Burke Video Editing Vitalii MakhnovetsTim Phetdara Writing Tannen Williams Writing, Web Editing Jimmy Thang Since last September, the average price of SSDs and RAM have continued to climb: prices have increased by 137% for 2 TB NVMe SSDs on average, 183% for 2 TB SATA SSDs, 363% for 32GB DDR5 kits, and 294% for 32GB DDR4 kits, with AVG prices collected from the samples of product listings specified below the chart. In some situations, it’s worse: DDR5-6000 64GB kits have climbed from $240 to $1,300-$1,400 on average, an increase of around 483%. If you had dreams of a home server or a high-end engineering design or editing machine with 128GB of memory, that’s gone up even more. We bought 128 GB of DDR5-6400 GSkill ECC Registered memory in 2024 for $1,060 to use in an editing machine. Today, that RAM isn’t even available. If you bought similar RAM, you could get scalped by a third-party seller for some NEMIX 128GB of a similar spec for $6,800, or maybe you’d prefer 512GB -- something that used to cost $4,200 -- for $23,811. And 99 cents, because fuck you. In 2024, we also bought a 4 TB Samsung 990 Pro for $390. Today, that same drive is $1,100, or a 254% increase. This is the K-shaped economy. From a single stick of 16 GB to a few sticks totaling 128 GB, everyone is getting ripped-off. People who want to play games for a hobby are robbed and people who want to start a business that needs a high-end computer might find it cheaper to rent a system or use cloud computing, which sort of seems like the endgame. This massive diversion to data centers isn’t just affecting RAM and SSD prices for PC enthusiasts – it’s affecting the prices of all consumer electronics for literally everybody. According to the IDC: “Worldwide smartphone shipments will fall 16.7% in 2026, [...] The average selling price of a smartphone will reach $581 in 2026, up 27.6% in a single year.” After Apple increased prices on its MacBook Neo, MacBook Air, MacBook Pro, Mac Studio, iPad, iPad Air, iPad Pro, iPad Mini, HomePod, HomePod Mini, Apple TV Box, and Vision Pro headset, Apple’s Tim Apple noted in his final earnings call as CEO, “We reluctantly raised prices I would say and we did it because we’re in what I would characterize as a 100-year flood on the memory pricing with exponential increases in memory prices.” XBOX recently announced its latest round of price increases, stating, “Effective August 1, 2026, we will be updating prices worldwide. The price of XBOX consoles will increase by US$100 for 512 GB models and US$150 for 1 TB models. We will also be sunsetting our 2 TB model.” The company added, “Unfortunately, console storage and memory prices have increased by more than 2.5x and we expect another doubling by the fall of 2027.” Amazon raised prices on its Echo Dot, Echo Show 11, Kindle, Kindle Paperwhite, Fire TV Stick HD, Fire TV Stick 4K Max, eero 7, and eero Pro 7. Nintendo inflated its Switch 2 MSRP, and Sony’s already increased its PS5 prices twice within the last year. TechInsights CSO, reported via the New York Times, added, “Yes, the prices of your iPhone are going up. But it’s also possible the prices of your M.R.I. machines may go up, or worse than that, that M.R.I. machine may not get made.” Meanwhile, Chinese newcomers CXMT and YMTC have continued their rapid expansions, with YMTC breaking into the global top 3 NAND manufacturers by shipments, surpassing Kioxia, Micron, and SanDisk, and CXMT now holding 10% of the global DRAM market share by revenue, up 6 percentage points YoY, as reported by Counterpoint Research. And yet being challenged by the United States Government for entrance, despite selling consumer-grade memory. To us, the most concerning part of this is that the memory suppliers are trying to finally bust the cyclical nature of memory pricing, which would mean they want to eliminate the future low point for prices, especially for consumers. Overview Long-Term Agreements (LTAs) are made between memory manufacturers and large clients, often unnamed. These clients almost certainly include NVIDIA. Shortages caused by AI demand could keep supply just tight enough to maintain the currently inflated RAM and SSD prices for years to come. Throughout the past year, all major manufacturers altered their customer relationship strategies, now forming 3-5 year long-term agreements (or LTAs) with, and allocating 50 - 70% of their outputs to only their largest 5-16 customers. Some form of agreements have been in place with major customers since, basically, the dawn of time -- but the capacity allocated to them and the duration are concerning. Chosun Daily explains: “LTAs are transforming the industry from a 3–5-year boom-bust cycle into a long-term order-based model with pre-secured demand.” In other words, the manufacturers went from prioritizing enterprises over consumers to now prioritizing their 10 largest companies over everybody. In fact, thanks to all the major cloud service providers and hyperscalers buying up all the supply and sending prices skyrocketing, smaller businesses can no longer afford on-site server upgrades, forcing them to rent cloud infrastructure instead. Fortunately for the hyperscalers, JP Morgan notes, “While LTAs may help keep costs manageable for hyperscalers, they dramatically decrease the flexibility of memory fabricators to address demand for memory chips in consumer goods.” LTAs Restructuring Market Dynamics Starting with the long-term agreements, we can look to each of the manufacturers’ latest earnings reports: Micron reported that its 16 LTAs currently represent 20% of its DRAM and a third of its NAND supply, noting, “When completed, we expect approximately half or more of our company revenue to be under these SCAs with customers across end markets.” SanDisk, whose Datacenter and Edge revenue increased by 1,298% and 392% YoY while revenue for its consumer segment decreased by 5% YoY, reported that it’s allocating 50% of its bit output in 2027 and 67% of its bit output in 2028 to LTAs. Kioxia claimed: “We are on track toward our 50% LTA volume coverage for calendar year 28 with key customers.” In an earnings call, Samsung stated, “We have already finalized agreements with the top five global data center customers, [and] are also in the final stages of talks with five additional major accounts to support their AI-related demand.” The company added, “We intend to maintain flexibility in supply allocation and plan to allocate approximately 60% to 70% of our total capacity to long-term supply contracts while preserving sufficient capacity to support customers without multi-year contracts.” And Western Digital’s CEO affirmed, stating, “The last time we reported on LTAs, we talked about having one LTA of a large customer all the way up to calendar year '29. But we're very much in the throes of discussions with customers to establish LTAs for calendar '29, '30 and '31 as well. So visibility remains very strong, customer-driven demand for LTAs extending all the way out to '31 remains very strong.” As for the effects: our understanding is that these long-term agreements will fundamentally alter the way the memory market works for consumers and everyone else. The Korea Herald explains this: “The shift marks a departure from an industry long dominated by quarterly negotiations and short-term orders, potentially softening the boom-and-bust cycle that has defined the memory business for decades.” Samsung reiterated this in its latest earnings call, stating, “Historically, the memory industry has experienced recurring cycles of upturns and downturns driven by demand fluctuations in consumer-oriented applications. However, by increasing the share of longer-term backlog-driven business, we hope to significantly enhance the stability and visibility of our business.” And Samsung’s not alone – all manufacturers are echoing the same “enhanced visibility” phrase in some manner or another. One industry analyst, reported via The Korea Herald, asserted: "LTAs are unlikely to eliminate the memory cycle, but they could reduce earnings volatility by locking in demand for longer periods. How effective they are will depend on how the contracts hold up in the next downturn.” LTA Customers Buy a GN 4-Pack of PC-themed 3D Coasters! These high-quality, durable, flexible coasters ship in a pack of 4, each with a fully custom design made by GN's team. You'll get a motherboard-themed coaster with debug display & reset buttons, a SATA SSD with to-scale connectors, RAM sticks, and a GN logo. These fund our web work! Buy here. Amidst the height of the shortage, demand from cloud providers continued to increase: TrendForce anticipates CapEx for the world’s nine major CSPs to increase from $922 billion in 2026 to $1.383 trillion in 2027, explaining: “The rapid increase will be driven in part by soaring memory contract prices and robust procurement demand. TrendForce estimates that DRAM and NAND Flash combined will account for 47% of CSPs’ total CapEx in 2026, with their share rising further to 68% in 2027.” Based on TrendForce’s figures, that’d put the nine largest CSPs’ CapEx spending on only DRAM and NAND at an estimated $433 billion in 2026 and an estimated $940 billion in 2027. Luckily for the CSPs, after driving up RAM and SSD prices, smaller businesses couldn’t afford on-site server upgrades, forcing them to rent cloud infrastructure from the CSPs instead. Amazon’s CEO described the new dynamic in its 1Q ‘26 earnings report, explaining, "One of the interesting things that we see right now with the change in price and supply on things like memory is that it is a further impetus pushing companies who have on-premises infrastructure into the cloud.” He added, “We have seen a number of conversations we have been having with enterprises for many months [...] accelerate rapidly just because we have a lot more supply than what others have. It will be interesting to see how that evolves over time.” That’s it. That’s the quiet part out loud, and the part we’ve all been saying: Amazon, the owners of one of the world’s cloud infrastructure providers via AWS, is saying that this is pushing companies into the cloud. That’s not going to stop at companies. And the CSPs are already seeing the results. In their latest filings: Google Cloud’s revenue increased 82% YoY, Microsoft’s “Azure and other cloud services revenue” increased 43% YoY, and AWS sales increased 36.7% YoY, which Amazon notes is its “fastest growth in 18 quarters.” Breakdown by End Application For further reference: In TrendForce’s “Forecast Breakdown of NAND Flash Bit Demand by End Application, 2026-2027” chart, the firm projects NAND bit demand for server applications to increase from 44.2% in 2026 to 51.1% in 2027 while % of total bit demand for PC, mobile, game console, and other applications will each moderately decrease. In comparison, TrendForce expects DRAM’s bit demand from graphic and server applications to increase by 1.7 and 1.5 percentage points, reallocating % of total bit demand away from PC, mobile, and consumer applications. While NAND is primarily only affected by increased server consumption, DRAM is heavily affected by increased graphic consumption in addition to increased server consumption, as manufacturers prioritize HBM. Based on TrendForce’s projections, PC applications will account for 5.6% of DRAM bit demand and 11.5% of NAND bit demand in 2027. Compared to TrendForce’s 2019 numbers, the % of total DRAM consumption for server applications is projected to increase by 13.2 percentage points, and projected to decrease by 7.1 percentage points for PC applications. As for how this all affects the consumer market: I think we all know... SSD & DRAM Price Trends | 09/2025 - 08/2026 | GamersNexus This chart compares AVG prices from September 2025 to AVG prices currently. We collected each entry’s AVG prices from a sample of individual product listings that we found complete price histories for on PC Part Picker, as specified below the chart. Based on our samples and since last September: AVG prices for 2 TB NVMe SSDs increased by nearly $200, from $143.25 to $340. AVG prices for 2 TB SATA SSDs shot up 183%, from $112.33 to $317.67. This is likely aided in some capacity by less production of SATA SSDs. AVG prices for 32GB DDR5-6000 CL30 kits skyrocketed by an insane $445, or 363%, increasing from $122.50 to $567.50. Finally, AVG prices for 32GB DDR4 kits surged from $61.50 to $242.50 currently, or by 294%. While all prices increased significantly within the last year, RAM appears to have undergone steeper % increases. Part of this, we’d assume, is because, unlike NAND supply which is primarily reallocated to server applications, DRAM supply is getting hit on both ends, reallocating supply to both server applications and HBM for graphics cards and accelerators. HDD Price Comparison For additional comparison, PC Part Picker’s price trends chart for a 16TB hard drive illustrates how its AVG price increased by nearly 129% in the same period, from roughly $350 in September 2025 to around $800 currently, with price increases really beginning to accelerate in April 2026. Spot Price History | Session Averages | GamersNexus Here, we’ll take a look at spot prices. While spot prices aren’t representative of the contract prices most major OEMs actually pay, they are indicative of demand. This chart plots the spot price session AVGs for 16Gb DDR5, 512Gb TLC wafer, and 16Gb DDR4, which we collected from saved dramexchange webpages accessed via archive.org’s Wayback Machine. The 16Gb DDR5 session AVG started around $6 between July and mid September, jumped by about $20 between October and December, and has gradually increased to its current $54 session AVG since January. The 512Gb TLC wafer session AVG held around $2.70 between July and September, surged from $3 to $23 between October and March, and has plateaued to around $21 since. The 16Gb DDR4 session AVG began at around $8.60 in July, skyrocketed to $78 between September and January, dropped to $58 in May, and has continued a steady climb since, sitting around $91 currently. It's basically behaving like the crypto market. If DDR4 spot pricing were an altcoin, it'd fit right in with 2018. Compared to each entry’s session AVG in July, current spot prices have increased by roughly: 800% for 16Gb DDR5, 678% for 512Gb TLC, and 958% for 16Gb DDR4, with greatest increases occurring between October and March. YMTC Gains Meanwhile, emerging Chinese manufacturers CXMT and YMTC are gaining on the incumbents. According to Counterpoint Research, CXMT’s global DRAM market share by revenue has increased from 4% in 2Q 2025 to 10% in 2Q 2026. This should be terrifying for the incumbents, and maybe explains why the US is so eager to ban CXMT while US-based Micron scoops up more business. If you’re curious about these companies, we ran a deep-dive documentary called “The Rise of Chinese Memory” that delves into the history of both. Counterpoint also claimed that YMTC recently broke into the top 3 largest NAND manufacturers by shipment share, now surpassing Kioxia, Micron, and SanDisk. In fact, the Chinese manufacturers are now reportedly looking to expand their market shares outside of China. Jukan, citing DigiTimes, reports: “As part of this strategy, YMTC is reportedly selling NAND chips to independent third-party module makers, which then assemble them into finished enterprise solid-state drives (SSDs) for customers such as U.S. neocloud providers. This indirect route is intended to avoid sensitivities surrounding the products’ country of origin. [...] With DDR5 and LPDDR5 supplies becoming increasingly tight, CXMT has reportedly begun product qualification with small and midsized cloud service providers in markets including the U.S. and Canada. The company appears to be seeking to secure customers early by locking in capacity commitments ahead of time.” That said, after Bloomberg reported that Apple was attempting to purchase chips from CXMT and YMTC, Chuck Schumer did what he does best and put his glasses in attack position to write a strongly-worded letter, urging Apple: “to abandon any effort to incorporate memory from CXMT, Yangtze Memory Technologies Co. (YMTC), or any other Chinese state-backed supplier into any of its products,” and claiming “American companies, like Apple, should be buying chips stamped ‘Made in America.’” The WSJ also quoted Commerce Secretary Lutnick as saying, “There have to be ‘other solutions to the memory issue, but it’s not great American companies using Chinese memory.’” For context, in late 2023 Micron’s CEO stated, “Today, only 2% of the world’s total memory production, DRAM production, is coming from the U.S.” Additionally, the National Institute of Standards and Technology notes that Micron’s New York and Idaho fabs currently in construction “will help the U.S. grow its share of advanced memory manufacturing from less than 2% today to approximately 10% by calendar year 2035.” We’d expect that a significant portion of the chips Micron eventually manufactures in the U.S. will be allocated to HBM, making the % of DRAM available to consumers likely even less. In other words, Micron, the only U.S.-based manufacturer, doesn’t produce enough DRAM in America to support American companies today, nor will it produce enough DRAM in America to support American companies in 10 years, so we aren’t entirely sure what ‘other solution’ Secretary Lutnick is referring to unless he wants to open a fab on his allegedly dead friend’s island... Restricting imports now will only harm American consumers and people. Memory is also an unlikely candidate for vulnerabilities, as we discussed in our Rise of Chinese Memory documentary. The New York Times, speaking with lobbyists asking for government intervention, reported, “Some have suggested that selling chips to a broader set of customers should be a condition for receiving money from the 2022 CHIPS and Science Act, a program to fund semiconductor research and manufacturing in the United States, three of those people said. Others have asked the Trump administration to use a Korean War-era law called the Defense Production Act to require allocation to industries outside of A.I., five people said.” In an interview with Tom’s Hardware, Silicon Motion SVP Nelson Duann noted about the Chinese manufacturers, “Because they receive government support, they also have a responsibility to help maintain the health of the local market. Foreign suppliers generally follow the highest-return opportunities and can allocate most of their supply to data centers. Chinese suppliers cannot do that in the same way because the government can provide guidance and encourage them to support certain local industries.” We think it’s ironic that, despite receiving nearly $6.5 billion in government support from the CHIPS act, Micron has no responsibilities to support our domestic market or industries. And Micron, too, is part of the US military’s memory supply, so this accusation of CXMT’s China-backing continues to fall flat. Supply Forecast As for the forecasted supply/demand going forward: TrendForce predicts we may see a light at the end of the tunnel for NAND sooner than DRAM, citing a greater output due to denser-layer NAND technology, stating, “Boosted by the rollout of new capacity alongside weak consumer demand, NAND Flash is shifting toward a looser supply-demand structure and will face downward price adjustment pressure in the second half of the year, underscoring a clear divergence in their market cycles.” In its “DRAM and NAND Flash Sufficiency Ratios” chart, TrendForce anticipates NAND Flash reaching a positive sufficiency ratio, where supply exceeds demand, sometime in 2027, while DRAM’s sufficiency ratio is only expected to worsen. That said, not all industry leaders share the same optimism. In an interview with Tom’s Hardware, a Silicon Motion SVP asserted, “The retail SSD market has almost disappeared.” “Suppliers can get some bit growth by moving to newer NAND generations, but demand is growing faster. As a result, the shortage will not improve next year. It will get worse.” Conclusion Visit our Patreon page to contribute a few dollars toward this website's operation (or consider a direct donation or buying something from our GN Store!) Additionally, when you purchase through links to retailers on our site, we may earn a small affiliate commission. We obviously can’t know exactly how this is all going to play out since it’s getting deep into the future of geopolitics, including the November election. Politics has become deeply intertwined with the computer hardware and chip industry, and those political decisions to ban, unban, or dole-out money and for what purpose will drive pricing. We do think that the move to higher capacity and duration long-term agreements will fundamentally alter the market’s traditionally cyclical pricing nature, likely reducing overall pricing volatility, and not in a good way. For consumers, that means higher floor pricing. Even if it comes down, it may not be to prior lows. Our best hope might be a bubble pop, but that might also destroy the economy... We try not to make predictions in these situations, as we really have no unique insight here; however, if manufacturers are allocating up to 70% of their capacities up to 5 years in advance, we’d assume that supply will remain constrained until the manufacturers are able to meaningfully increase supply through fab expansion, which takes several years at a minimum. Going forward, we’ll continue to keep an eye on pricing and we’ll be sure to update upon any meaningful market changes.

14. How to win a beer with high-dimensional statistics (jamiesimon.io)

My longtime labmate-turned-student/friend1 Dhruva Karkada recently wrote a sick paper on data statistics which deservedly went viral on Twitter, in part beca...

A new paper by researcher Dhruva Karkada is making waves on Twitter for its innovative approach to high-dimensional statistics. It offers insights that could change how data analysis is perceived, and it even comes with a challenge that could win you a beer.

21. Show HN: Raven – The harness of harnesses, built for RSI (github.com)

The Harness of Harnesses • built for RSI: a trusted, persistent, self-evolving multi-agent ecosystem for all-domain collaboration. - EverMind-AI/Raven

Raven is designed to assist with repetitive strain injuries (RSI) by creating a collaborative environment using a multi-agent system. It aims to adapt and evolve, offering support across various domains for users seeking relief and efficiency.

23. Who Killed Paulina Borsook's Career? (www.wired.com)

A quarter-century ago, Borsook published a wildly prophetic, penetrating book about Silicon Valley called Cyberselfish. Then she disappeared.

Paulina Borsook’s influential book on Silicon Valley, “Cyberselfish,” made waves 25 years ago, but her career took an unexpected turn afterward. This piece investigates what led to her disappearance from the tech scene.

26. The new Firefox design is here (blog.mozilla.org)

What comes to mind when you think of Firefox? Is it our best-in-class privacy features and ad blocking capabilities? Our partnerships with Wrexham, NVIDIA

Firefox is rolling out a fresh design, emphasizing its top-notch privacy features and ad-blocking capabilities. Learn how these changes and new partnerships could enhance your browsing experience.

28. Startup Nights 2026 is comming up on 5-6 Nov. in Switzerland (www.startup-nights.ch)

Startup Nights is the annual meetup of the Swiss startup family. Expect two packed days full of keynotes, workshops, pitch

Join the Swiss startup community on November 5-6 for two days of keynotes, workshops, and pitch events at Startup Nights 2026. It’s a prime opportunity to network and gain insights from industry leaders.

29. OpenAI: Tomorrow we are re-opening the Pro $200 subscription (twitter.com)

Hi,

Tomorrow we are re-opening the Pro $200 subscriptions to new subscribers, but together with it we are also changing how we calculate the usage for it. In effect, if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan.

Now that it's said, let me …

OpenAI is re-launching its Pro subscription plan at $200, but with a new usage calculation that effectively reduces API costs by 50%. This change could make the service more appealing for potential subscribers looking for better value.

Article Details

Browse by Date

Type to search HN and engineering blog articles
Press Escape to close  |  ↑↓ to navigate  |  Enter to open