SCROLL NEWS / DISCOVERY
Search headlines
Find the stories shaping the conversation.
Results for " moe " 100 found 🔀 AI-powered Shuffle
Sign in to save shuffle history.
St. Xavier QB Aidan Newdigate makes touchdown run vs. Moeller
St. Xavier QB Aidan Newdigate made a touchdown run vs. Moeller on Sept. 18, 2026.
Gladbach sink Dortmund, St Pauli edge closer to drop
Borussia Moenchengladbach ensured they will remain in Germany's top flight next season when Haris Tabakovic scored with two minutes remaining to beat Borussia Dortmund on Sunday, while Mainz pushed St Pauli closer to relegation. Earlier on Sunday, goals from Phillip Tietz and Phillipp Mwene took Mainz to a 2-1 win at St Pauli, securing their top-flight status next season while pushing the hosts closer to relegation.
Bark dust piles catch fire at Cornelius landscape supply business
Cornelius firefighters knocked down a compost fire on Saturday after crews spotted heavy smoke while heading back from another call.
Vikings’ QB Situation Sparks Major Bleacher Report Warning
The NFL trade deadline is not until Tuesday, November 10. However, several analysts have put together articles on ways teams can improve. For the Minnesota Vikings, Bleacher Report’s Moe Moton indicated that the Vikings should look at the quarterback position heading into the trade deadline.Kyler Murray has led the Vikings in the quarterback room this […] The post Vikings’ QB Situation Sparks Major Bleacher Report Warning appeared first on HEAVY.
Boston Celtics officially announce Jaylen Brown
Per the team, they have "acquired 9-time NBA All-Star and 6-time All-NBA selection Paul George as well as future draft pick compensation from the ... 76ers".
Nine-year-old boy hospitalized with brain-eating amoeba, health ministry says
The boy is sedated and on a ventilator at a Haifa hospital. The rare infection has a 97 percent fatality rate, according to the Health Ministry
Miles McBride brings NBA championship trophy to Moeller football game
New York Knicks guard Miles "Deuce" McBride brought the Larry O'Brien championship trophy to the Moeller Crusaders football team's 45-7 win.
Scientists identify design rules for the best thermoelectric materials
Researchers from Tokyo Metropolitan University have used a theoretical framework to derive design rules for creating new thermoelectric materials. The efficiency of converting thermal energy to electricity ...
If you’re at risk of Alzheimer’s, these dietary habits can help, study finds
Research suggests diet remains a relevant factor for brain health even after blood markers of dementia are present.
Google finally explains why the AICore app is eating up your storage
It looks excessive, but there's actually a valid reason behind it.
(Half) Twisted Science: Researchers Build a Molecular Möbius Strip With Only Half the Twist
The Quantum Insider (TQI) is the leading online resource dedicated exclusively to Quantum Computing.
Big changes coming as demolition begins on former Libbey Glass site in Shreveport
A 100-year-old Shreveport plant is being demoed. Here's what will happen next at the former Libbey Glass site.
‘Underwhelming’ 49ers’ 1st-Round Pick Called NFL’s Biggest Bust
Bleacher Report's Moe Moton predicts San Francisco 49ers WR and 2024 1st-round pick Ricky Pearsall will be NFL's "Biggest Bust" in 2026. The post ‘Underwhelming’ 49ers’ 1st-Round Pick Called NFL’s Biggest Bust appeared first on HEAVY.
Former Bucs Draft Pick Called ‘Biggest Bust’ for AFC Team
Bleacher Report's Moe Moton singled out former Bucs DT Logan Hall as the NFL's "Biggest Bust" in 2026 with the Houston Texans. The post Former Bucs Draft Pick Called ‘Biggest Bust’ for AFC Team appeared first on HEAVY.
This exercise ‘sweet spot’ is linked to greater longevity
A new study pinpoints how many minutes per week of strength training might help people avoid death from conditions including heart attacks and Alzheimer’s.
Packers to interview former Patriots ST coordinator Cameron Achord
Matt LaFleur will interview former Patriots special teams coordinator Cameron Achord after losing Rich Bisaccia this week.
North Carolina teen critically ill from rare brain-eating amoeba found in freshwater, officials say
A North Carolina teenager is critically ill from a rare brain-eating amoeba, Naegleria fowleri, that's found in warm freshwater, state health officials said.
Michelle Zhou’s Venice Horizons Title ‘Big Little Things’ Boarded By Moebius Entertainment For Sales
The Hong Kong-US collaboration follows a woman who returns to her hometown of Harbin after years living in the US.
Best of Santa Cruz County entertainment, arts & food events this weekend, May 21-25
With the long weekend nearly here, check out things to do around Santa Cruz County with a recommendation from Lily Belli and a specially curated list from Lookout's BOLO events calendar.
XCEPTIONAL ATHLETE: Bonduel’s Carter Moesch
Bonduel, Wis (WFRV) – This week Local 5 Sports highlights our Xceptional Athlete Carter Moesch, a four year varsity football player for Bonduel High School. Moesch plays both offense and defense for the Bears, serving as a running back and a linebacker. He excels in both positions, but if he had to choose one, his […]
Analysis | Maine is bringing Iraq War politics back. This governor is feeling déjà vu.
The state's long-awaited primary is Tuesday.
Building a digital ant gallery, from the ground up
How a particle accelerator, a robot, highspeed cameras, and 2,000 ants came together to paint a picture of biological diversity.
Naderi wants to 'make most' of Rangers move
Rangers' new signing Ryan Naderi insists he wants to "make the most" of his move to Ibrox from 3. The German striker has scored eight goals this season and signed a "long-term contract" at the club, having previously played in Borussia Moenchengladbach's B team. "I am really pleased that this move has happened," Naderi told club media.
Jets Predicted to Cut Former Cowboys’ 1st-Round Pick
Bleacher Report's Moe Moton predicts former Dallas Cowboys DT and 2023 1st-round pick Mazi Smith will be cut by the New York Jets. The post Jets Predicted to Cut Former Cowboys’ 1st-Round Pick appeared first on HEAVY.
Colton Underwood Opens Up About ‘Weird Homoeroticism’ In NFL Locker Rooms
With his time on "The Traitors" at an end, the former "Bachelor" is reflecting on his public — and, at times, controversial — journey to living as his true self.
Gemma 4 and what makes an open model succeed
Having written a lot of model release blog posts, there’s something much harder about reviewing open models when they drop relative to closed models, especially in 2026. In recent years, there were so few open models, so when Llama 3 was released most people were still doing research on Llama 2 and super happy to get an update. When Qwen 3 was released, the Llama 4 fiasco had just gone down, and a whole research community was emerging to study RL on Qwen 2.5 — it was a no brainer to upgrade. Today, when an open model releases, it’s competing with Qwen 3.5, Kimi K2.5, GLM 5, MiniMax M2.5, GPT-OSS, Arcee Large, Nemotron 3, Olmo 3, and others. The space is populated, but still feels full of hidden opportunity. The potential of open models feels like a dark matter, a potential we know is huge, but few clear recipes and examples for how to unlock it are out there. Agentic AI, OpenClaw, and everything brewing in that space is going to spur mass experimentation in open models to complement the likes of Claude and Codex, not replace them. Especially with open models, the benchmarks at release are an extremely incomplete story. In some ways this is exciting, as new open models have a much higher variance and ability to surprise, but it also points at some structural reasons that make building businesses and great AI experiences around open models harder than the closed alternatives. When a new Claude Opus or GPT drops, spending a few hours with them in my agentic workflows is genuinely a good vibe test. For open models, putting them through this test is a category error. Something else to be said about open models in the era of agents is that they get out of the debate of integration, harnesses, and tools and let us see close to the ground on what exactly is the ability of just a model. Of course, we can’t test some things like search abilities without some tool, but being able to measure exactly the pace of progress of the model alone is a welcome simplification to a systematically opaque AI space. Share The list of factors I’d use to assess a new open-weight model I’m considering investing in includes: Model performance (and size) — how this model performs on benchmarks I care about and how it compares to other models of a similar size. Country of origin — some businesses care deeply about provenance, and if a model was built in China or not. Model license — if a model needs legal approval for use, uptake will be slower at mid-sized and large companies. Tooling at release — many models release with half-broken, or at least substantially slower, implementations in popular software like vLLM, Transformers, SGLANG, etc due to pushing the envelope of architectures or tools. Model fine-tunability — how easy or hard it is to modify the given model to your use-case when you actually try and use it. The core problem is that some of these are immediately available at release, e.g. general performance, license, origin, etc. but others such as tooling take day(s) to week(s) to stabilize, and others are open research questions — with no group systematically monitoring fine-tunability. In the early era of open models, the days of Llama 2 or 3 and Qwen pre v3.5, the architectures were fairly simple and the models tended to work out of the box. Some of this was due to the extremely hard work of the Llama, Qwen, Mistral, etc. developer teams. Some is due to the new models being genuinely harder to work with. When it comes to something like Qwen 3.5 or Nemotron 3, with hybrid models (either gated delta net or mamba layers), the tooling is very rough at release. Things you would expect to “just work” often don’t. I’ve been following this area closely since we released Olmo Hybrid with a similar architecture, and Qwen 3.5 is just starting to work well in the various open-source tools that need to all play nice together for RL research. That’s 1.5 months after the release date! This is just to start really investing more into understanding the behavior of the models. Of course, others started working on these models sooner by investing more engineering resources or relying on partially closed software. The fully open and distributed ecosystem takes a long time to get going on some new models. All of this is lead-in for the most important question for open models — how easy is it to adapt to specific use-cases? This is a different problem for different model sizes. Large MoE open-weight models may be used by entities like Cursor who need complex capabilities in their domain, e.g. Composer 2 trained on Kimi K2.5. Other applications can be built on much smaller models, such as Chroma’s Context-1 model for agentic search, built on GPT-OSS 20B. The question of “which models are fine-tunable” is largely background knowledge known by engineers across the industry. There should be a thriving research area here to support the open ecosystem model. The first step is to understand characteristics of different base and post-trained models to understand what they look like. The second step is to tune pretraining recipes for open models so they’re more flexible. Interconnects AI is a reader-supported publication. Consider becoming a subscriber. Subscribe For The ATOM Project and other Interconnects endeavors, we’ve put in substantial effort to measuring adoption trends in the open ecosystem. Everything takes a long time to unfold after a model is first publicly available — and adaptability is why. What we know for sure now, when Qwen has been going from strength to strength with its releases, is that technical staff across the industry has gotten comfortable working with Qwen models. Countless research methods and datasets were made to work with Qwen. It’ll take patience for any other model family to get to this point — a patience I’m not sure many open model builders have. This takes us to Gemma 4, Google’s latest open models. Gemma 3 was released more than a year ago, in March of 2025, and is a bit underrated. Gemma 4 comes in 4 sizes for now, with a bigger, MoE model of over 100B total parameters rumored but not released yet. The models we have today come in sizes of ~5B dense, 8B dense, 26B total 4B active MoE, and 31B dense. I’m most excited that they’re finally adopting a standard Apache 2.0 open source license. This’ll massively boost adoption. The standard of better licenses for strong open-weight LLMs was set by mostly Chinese open model labs in the last 1-2 years, and now U.S. companies are following suit. I will personally be so happy if the horrible Llama licenses and Gemma terms of service were an ~18-month transient dynamic of the industry being nervous about releasing strong open models. The Gemma 4 scores look very solid, the small models have incredible benchmark scores (especially in general domains like LMArena) and the 31B model rivals the recent Qwen 3.5 27B, which is the leading member of that class. The ~30B size range is an important one, as it’s accessible both to researchers and to enterprises looking to deploy the model in real use-cases. Where the 7B model scale is the default for tinkering and research, a 30B model is the default for seeing if an open model can unlock substantial value in your specific workflow — a good mix of intelligence, low price, tractability for downstream training, etc. This takes us back to the above adoption criteria I mentioned for open models and the bigger question — do I think Gemma 4 will be an overwhelming success? Previous Gemma models have been plagued by tooling issues and poorer performance when being finetuned. Gemma 4’s success is going to be entirely determined by ease of use, to a point where a 5-10% swing on benchmarks wouldn’t matter at all. It’s strong enough, small enough, with the right license, and from the U.S., so many companies are going to slot it in. I’m cautiously optimistic that Gemma 4 is going to work better here. Winds are shifting for open models built in America. We saw GPT-OSS go through a bumpy launch to become an overwhelming success. There’s a collective energy around the likes of Reflection, Arcee, Nemotron, Gemma, Olmo, and peers that show substantial demand for building new stacks around open models. There’s capital to be spent on AI stacks across the economy by those who want more ownership of everything, including the model. After launching The ATOM Project 240 days ago, the conversation is shifting into the next stage. Summer of 2025 was a crisis moment where the U.S. AI scene realized it can’t wait and figure out open models after building AGI. The two markets will capture different areas and proceed in parallel. Now that more companies in the U.S. are releasing strong models, we need to improve the ecosystem so that these models are easy to use, understand, and build value around. It’s the hard work to build another inflection point in these adoption plots I’ve been updating consistently, but that’s the work to be done. Join me in it. More data coming soon! Here’s a sneak peek:
Watch schedule highlights of Cincinnati's Week 6 high school football
Cincinnati Week 6 high school football games include St. Xavier at Elder, Moeller at La Salle, East Central at Harrison and Deer Park vs. Madeira.
AI-Powered Emails Are Influencing Politics on Hot-Button Issues
From California to North Carolina, officials weighing policy changes are being inundated with comments via platforms that tout AI capabilities.
Yankees Birthday of the Day: Bob Cerv
A part-time player with New York, Cerv shined elsewhere during a 12-year career.
Guinea-Bissau: Opposition leader house arrest raises stakes
Opposition leader Domingos Simoes Pereira remains under house arrest months after the Guinea-Bissau coup, deepening a political crisis and raising tensions with international partners.
Highly Critical Drupal Core Flaw Exposes PostgreSQL Sites to RCE Attacks
Drupal released security updates for a highly critical Drupal Core vulnerability affecting sites that use PostgreSQL.
5 Common Fruits That Can Boost Brain Health
Certain fruits, including berries, citrus, and pomegranates, contain nutrients that improve cognitive health. The darker the fruit, the more brain-healthy nutrients it may have.
How does a shape-shifting microbe survive extreme heat?
Discover the fire amoeba, an extremophile that's breaking records and expanding what scientists know about the limits of life.
Brain-eating amoeba that killed 17-year-old linked to a man-made creek, health officials say
A man-made creek was likely the source of exposure to the brain-eating amoeba that killed a 17-year-old boy, health officials said.
North Carolina teen dies from rare brain-eating amoeba found in freshwater: Health officials
State health officials say the teenager died Monday.
Glyphosate: A story of science, risk, and nuance
In the summer of 2014, a California groundskeeper named Dewayne Johnson started noticing patches of raw, weeping skin spreading across his body. He had spent years applying Roundup, Monsanto’s glyphosate-based herbicide, for a school district, sometimes drenched in it when equipment malfunctioned. By 2014, he had a diagnosis: non-Hodgkin’s lymphoma. In 2018, a jury awarded him $289 million in damages, later reduced on appeal. Tens of thousands of lawsuits followed. The story hasn’t quieted down. It’s all over the internet, swinging between “this will kill you, buy organic” and “relax, you’re fear-mongering.” Some of that recent chatter stems from an Executive Order protecting glyphosate and its production, which set off intense infighting within the MAHA movement. Some of it reflects a deeper truth: Americans genuinely care about harmful chemicals in their environment (see poll below). And the conversation isn’t going anywhere, with a Supreme Court case involving Monsanto set for this April. Glyphosate (the active ingredient in Roundup) is one of the most widely used herbicides in the world. But does glyphosate cause cancer? And how much exposure is actually risky? With so much mistrust in institutions and corporations (and some genuine conflicts of interest), making sense of the actual scientific evidence is genuinely hard. Here is a translation of what the scientific evidence actually says. Does glyphosate cause harm? Glyphosate is an herbicide that kills most plants it contacts rather than targeting specific species. It works by blocking an enzyme pathway (EPSPS) that plants and many microorganisms need to produce essential amino acids. Without those amino acids, the plant can’t grow and dies. Glyphosate was first developed by Monsanto in the 1970s and became ubiquitous after the company introduced “Roundup Ready” crops in the 1990s. In 2024, Bayer removed glyphosate from residential Roundup-branded products due to litigation pressure. But today it is still the most widely used herbicide in the world. Farmers also discovered they could spray it directly on crops like oats, lentils, beans, and peas just before harvest to dry them out faster and make harvesting easier. Because of its widespread use, it’s obviously really important to understand if it’s harmful to humans. The key principle: The dose makes the poison The most important concept in toxicology is that “the dose makes the poison.” Sixteenth-century physician Paracelsus articulated what remains the foundational principle of toxicology: sola dosis facit venenum. Water will kill you if you drink enough of it. Aspirin saves lives at 81 milligrams and causes internal bleeding at higher chronic doses. So, at what dose is glyphosate poisonous? The LD50 in rats (the dose that kills half the test animals) is around 5,600 mg/kg body weight. Table salt is roughly 3,000 mg/kg. By that classic measure, glyphosate is less acutely toxic than salt. But some people are exposed to a lot of glyphosate through their work. And among the general public, there are many questions about low-dose, cumulative exposure from multiple sources over long periods. Does glyphosate cause health problems? Cancer Correlation maps are striking, with one analysis finding that 60% of counties that spray the most glyphosate have non-Hodgkin’s lymphoma rates above the national average. But correlation alone doesn’t establish cause, which is why animal and human studies matter. In animals: Rat studies are typically what most fear is rooted in, but most people don’t know the study details on the dosage given to rats. A pooled analysis of the 13 most rigorous rat studies found evidence that very high doses, around 55,000 times greater than typical food residue exposure, can cause certain cancers. In humans: One of the strongest studies followed 55,000 herbicide users over 20 years and found no overall increased cancer risk. There was one signal: among those with the highest use, rates of acute myeloid leukemia (AML) appeared elevated. A broader meta-analysis of 18 human studies also found no overall cancer risk, though a separate meta-analysis focused specifically on non-Hodgkin lymphoma found that people with high cumulative glyphosate exposure faced a 41% higher relative risk compared to those with low or no exposure. These human studies are observational studies (randomized controlled trials would be unethical), and meta-analyses are only as reliable as the studies they pool. But the story seems consistent: at very high exposure levels, there may be some cancer risk. But that’s about as far as the science currently takes us. Emerging research: gut and reproductive health Gut: Because glyphosate targets a pathway that exists in some gut bacteria, scientists have questioned whether it could affect the microbiome. In mice, a 2023 University of Iowa study found that glyphosate altered gut bacterial composition and raised markers of intestinal inflammation. A 2024 systematic review found similar patterns: disrupted gut bacteria and problems with the mucus layer that protects the intestinal wall in mice. Whether this translates to humans is an open question. The human gut is highly variable and resistant, and we don't yet have strong studies confirming the same effects in people. Reproductive health: Animal studies have found that glyphosate-based herbicides can interfere with reproductive hormones, including disrupting egg cell development in mice and reducing the number of eggs available in sheep’s ovaries. However, the doses used in these experiments were high, so it’s not yet clear whether the same effects would occur in humans. On the human side, a 2021 pilot study led by Mount Sinai researchers tested urine samples from pregnant women across the U.S. and found glyphosate in 95% of them. The study also found a possible connection between higher glyphosate exposure during pregnancy and a subtle physical difference measured in female newborns. That said, some connections weren’t statistically reliable and this was a small study. The researchers themselves said more and larger studies are needed before drawing firm conclusions. What do regulators say about glyphosate safety? In 2015, the International Agency for Research on Cancer (IARC), part of the World Health Organization, classified glyphosate as "probably carcinogenic to humans" (Group 2A). The classification is worth taking seriously. But, there’s an important distinction: IARC classifies hazard, meaning the potential to cause harm under some conditions. It does not assess risk, which accounts for actual exposure levels and the real-world probability of harm. The U.S. EPA, the European Food Safety Authority, Health Canada, and regulatory agencies in Australia and Japan have all concluded that glyphosate is “unlikely to be carcinogenic to humans” at realistic exposure levels. They’re asking a different question: given how much of this people actually encounter, what is the probability of harm? Your actual exposure is what determines your actual risk. What does glyphosate mean for your daily life? At the grocery store: Food residues are small. The EPA’s acceptable daily intake is 1.75 mg/kg of body weight per day. To hit that threshold from oatmeal alone, a 150-pound adult would need to eat roughly 50 pounds of oats every day. But this calculation addresses only one pathway (food), and we don’t yet have robust tools to measure people’s actual total exposure. If reducing exposure matters to you, organic certification prohibits synthetic herbicides, and organic oat products generally test at lower or undetectable levels. Either way, a diet rich in whole grains, conventional or organic, is far better for your health than one that omits them out of fear of residues. On the farm: Occupational exposure is orders of magnitude higher than consumer exposure. Farmers and farm workers who mix, load, and apply glyphosate-based herbicides, especially without protective equipment, face meaningfully different risk profiles. Proper PPE isn’t optional. Integrated weed management strategies that rotate chemical modes of action are also worth taking seriously, both for health and for long-term effectiveness. If you live near a farm: Spray drift is real. Concentrations near field edges can spike during and after application, particularly on windy days. This is an area where I think public health could really step up and empower rural communities with buffer zones, rather than placing the onus on individuals. But for now, knowing your local agricultural calendar and keeping windows closed during application are the first steps. Bottom line Dewayne Johnson was exposed to glyphosate at levels most people will never encounter, and a jury decided he deserved compensation. That outcome is not evidence that the oatmeal in your pantry is dangerous. But it is a reminder that the dose, the context, and the person doing the exposing all matter enormously. Glyphosate is a chemical tool with benefits, trade-offs, and genuine open scientific questions. Learning to hold that complexity without collapsing into panic or dismissal is, honestly, one of the most useful and hardest things to do in navigating modern health science. Love, YLE Your Local Epidemiologist (YLE) is founded and operated by Dr. Katelyn Jetelina, MPH PhD—an epidemiologist, wife, and mom of two little girls. Hannah Totte, MPH, is an epidemiologist and YLE Community Manager. YLE reaches more than 425,000 people in over 132 countries with one goal: “Translate” the ever-evolving public health science so that people will be well-equipped to make evidence-based decisions. This newsletter is free to everyone, thanks to the generous support of fellow YLE community members. To support the effort, subscribe or upgrade below: Subscribe
Apple Starts Preparing for September iPhone Event
Apple is getting ready for its big September iPhone event, which is expected to see the debut of the iPhone 18 Pro, iPhone 18 Pro Max, and an all-new foldable iPhone. Ahead of the event, Apple is holding an Event Support Experience Program lottery for U.S. retail employees who want to travel to its Apple Park campus to staff the in-person media gathering.
We tried Google’s AI glasses and they’re almost there
Google demoed prototype Android XR glasses that overlay Gemini-powered translation, navigation, and other information directly into your field of view.
Lions Secondary Put on Notice by Week 1 Fantasy Sleepers
The Detroit Lions‘ defense has a lot of questions entering the season. So, despite quarterback Tyler Shough being a lower-ranked fantasy football option in Week 1, he could end up being a great sleeper target. Moe Moton of Bleacher Report writes that Shough, along with his receiver Devaughn Vele, are two of the most underrated […] The post Lions Secondary Put on Notice by Week 1 Fantasy Sleepers appeared first on HEAVY.
Walking an extra 1,000 steps a day may help your heart, study finds
Here’s the latest science on how many steps you actually need every day.
MSU poultry scientists recognized with early career awards
Li Zhang and Timothy Bolz are honored by the Poultry Science Association.
UK fund major L&G aims to double Asia business, CEO says
Britain's biggest asset manager Legal & General aims to double its assets under management in Asia to around $500 billion amid its push to grow its international business, CEO Antonio Simoes told Reuters on Monday.
Michigan betting that space is next frontier to create jobs
Michigan is playing a key role in the US space industry, contributing to the Artemis II rocket mission. Products from Michigan-based companies, like Moeller Aerospace and Elmet Technologies, were part
Louisiana girl, 8, dies after contracting suspected brain-eating amoeba linked to lake swim
Brain-eating amoeba that killed 17-year-old linked to a man-made creek, health officials say
A man-made creek was likely the source of exposure to the brain-eating amoeba that killed a 17-year-old boy, health officials said.
Science Snapshots: September 20, 2026
Explore groundbreaking discoveries on North Korea's nuclear impact, amoebic evolution, and the health benefits of regular exercise.
15 discouraging hours with Onimusha: Way of the Sword, and 10 brilliant ones with He Who Watches
In the last few weeks, while playing some video games that I might review, I emailed two cries for help. I sent the first to a rep for Capcom, confessing that the company’s samurai game Onimusha Way of the Sword (out today for PC and console) was crushing me. “Feels like I’m playing ‘hard’ mode, not ‘normal,’” I wrote. “And I don’t want to have to knock it down to “story”. I’m exploring, trying to level my gear, but these two-phase boss fights with no mid-fight checkpoints are brutal.” I wagered I was only about halfway through the game, if that. I wasn’t sure I could keep up with the steepening difficulty curve. Was I missing something? Then, on Monday, I took some photos of some puzzles that were baffling me in the mind-bending He Who Watches (out now on PC), wrote out lengthy descriptions of how I’d been trying to solve them and asked Bobby Vanden, the game’s solo developer, for a nudge in the right direction. I told Vanden I’d spent hours on these puzzles. “I’m stuck in triplicate,” I told him. Could he tell me if the solutions I was attempting were even on the right track? I didn’t mention that the puzzles—coupled with major time pressures from other work—were stressing me out so much that playing the game was giving me headaches. This double dose of panic was unusual. I’m used to managing a workload full of reporting and reviewing, used to a crowded schedule and the need to stay calm amid the bustle. But I was stressed about these games. No one was forcing me to review them, nor even play them. I’d simply found them interesting enough to start, hoped they’d be interesting enough to write about and I’d sunk hours into them assuming I’d made the right call. Suddenly, however, I felt trapped. I was emailing for a way out. Above: This guy took me about an hour to beat. The feedback I got from Capcom confirmed my fears. They didn’t say it, but I will: Onimusha Way of the Sword is not for me. It’s a brutally hard game in need of a normal difficulty level. It’s a game that spikes so severely during its abundant boss fights—yet offers such flawed ways to improve when outside of them—that it feels like a bait-and-switch. In the new Onimusha, you play as famed samurai Musashi Miyamoto, aided by a magical soul-absorbing gauntlet that is possessed by a lady who offers advice. (Way of the Sword is a continuation of Capcom’s long-dormant Onimusha series, but that set-up had me wondering if the company was cribbing from their ill-fated Bionic Commando reboot; read about “wife-arm” when you get a chance). Way of the Sword is partially set in 17th century Kyoto, an explorable city where Miyamoto can fight demons, help civilians and eventually rescue ghostly dogs. The city evolves over time, with new mini adventures and obtainable loot on offer, all feeding into systems that promise to upgrade your gear and sword-fighting skills. Upgrades, however, are so incremental as to be imperceptible, causing the exploration of Kyoto to feel like inconsequential busywork. From that hub, players can go on linear missions into forests and old temples, where they will encounter many of the game’s tough bosses. Combat in Way of the Sword is akin to many modern melee games. Players are asked to mix attacks with dodges and parries. They’re encouraged to deplete an enemy’s stamina bar in order to do greater damage to its health bar. The old Onimusha games’ signature move was the issen, a counter-attack so sudden it’s all but pre-emptive. You strike just as they begin to strike toward you, beating them to the cut. By definition, the issen is a high risk move, leaving you wide open to pain. It feels risky in the game, especially against bosses, who can maul you in just a few blows. A game might train you to perform an issen well by offering a variety of lesser enemies who are not as tough as the boss but get close. Or it might offer a toehold against bosses by saving or checkpointing the player’s progress when a boss’ first health bar has been destroyed and before the second health bar appears. It might offer the less skilled player a chance to over-level their character, should the player simply commit to playing through its optional sidequests. Way of the Sword does none of this. Fighting its trivially easy grunt enemies and then its ferociously adept bosses is like playing a game of chess where I’ve got a rook, while the opposing side is all pawns and five queens. Bosses simply don’t checkpoint. You may be able to clear a boss’ first phase without taking much damage. Good for you. Do it again, every time you fail the second phase. As for the the central Kyoto hub, I kept pace with all of its added side activities and never felt like I was overpowered. Bosses could still easily buckle me. The new Onimusha, I concluded, is just too discouraging to keep playing. The advice I got from Capcom had been well-intentioned. They reminded me of the game’s various systems. They alerted me to an optional assist that flashes recommended button inputs (basically: counter this move; dodge that move). It was a little helpful. So was dropping down to “story mode,” which I tried last night. Story mode turns on those button prompts by default. It makes countering and dodging way too easy, nearly automatic. I don’t consider myself a master of action games like these, but I’ve cleared the likes of God of War, Stellar Blade, Star Wars Jedi and others of sub-Sekiro difficulty and had a fun time doing so. Not here. Not yet. Unless they patch Onimusha: Way of the Sword and add a middle difficulty or just checkpoint the damn bosses when they’re between health bars, my streak of loving Capcom games released in 2026 will likely end at two. It’s a shame, because I was having a good time finding those ghostly dogs. And then there’s He Who Watches, a phenomenal first person puzzle game with Portal vibes, except instead of a portal gun you’ve got a magical bow that shoots one arrow that can activate switches or be used to drag blocks. The twist is an actual twist: you can also walk on walls and the ceiling, reorienting gravity as you do, thereby placing blocks onto those other surfaces. Walking on walls and ceilings to solve problems is really the best thing. Would that I could in reality. He Who Watches consists of locked-room challenges, each brief enough to only require about a minute to actually solving, though it will take five, 10, eventually 30 minutes to figure out said solutions. They are not obvious! It’s a game of layered mechanics and shocking solutions. For example, see if you can follow this: You can drag a block to a wall. Then you can walk on the wall, which now feels like the floor, and drag that block across the wall. Later, back on the floor, you might realize you can look up. The block is still on the wall, stuck to it. You can shoot your magic arrow to its underside. Doing so creates a tether, so that you can yank the block off the wall and carry it, as if it were a balloon, into the spot where you need it to be to solve the puzzle. Here, watch Vanden explain some of the game’s basics in this launch trailer: As with any great puzzle game, the solutions to He Who Watches’ puzzles elicit relief and amazement. Oh my god, that worked! I didn’t even think that could be possible! Learning to solve the game’s puzzles feels like learning magic. But, oh, does the game get hard. Its Steam reviews, mostly glowing, are full of players remarking how quickly He Who Watches made them sweat. Wrote one: “loving this one already although my brain is hurting!” Wrote another: “As you get further in the game it feels like you need an IQ level of at least 120+.” Vanden has shown some mercy. Each locked room puzzle includes an optional junior version of the puzzle, meant to teach you. Difficulty nevertheless ramps severely. By the time I was emailing Vanden for help, I was struggling to even clear one of the hint puzzles. When I messaged him, my stress was exacerbated by strange things I’d found in the game’s various hubs: suspicious passageways, odd blocks and some intriguing locked doors. These were hints of secrets, I realized, yet I sensed I needed to play further into the game to understand them. I was stuck, but I wanted to see the rest. Vanden offered to give me the exact solutions, to solve those three puzzles for me, but first he gave four nudges. Three were for the specific puzzles, but his best hint was for all and ultimately proved to be a terrific life lesson: go back a bit, he suggested. Do the puzzles right before those. They were teaching you something. Focus on what their lesson was. Don’t rush, in other words. Recognize what you’ve accomplished. Build off of that. I did so. I absorbed the lessons. I saw those three tough puzzles in a new light. I solved one the night after he wrote back, another the morning after. I left a third for another time. I didn’t need it and I wanted to plunge deeper into the game, closer to the moment when I was sure I could start unlocking those secret doors. My head stopped hurting, too. He Who Watches feels special. It’s got more challenges, and I’m undeterred. It’s also nice that its puzzles don’t have a surprise second health bar. Game File is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Subscribe Item 2: In brief… 🎮 Ilari Kuittinen, co-founder of Helsinki-based Housemarque (Super Stardust HD, Resogun, Returnal) has stepped down from his studio, after a 31-year run, saying he’s “ taking a rare and well-deserved breather.” Sony purchased Housemarque in 2021, and the studio released its newest game, Saros, earlier this year. Trend-watch: Sony-owned Insomniac saw its founder Ted Price retire in early 2025, after a 30-year run. Sony-owned Sucker Punch saw its co-founder Brian Fleming step down in late 2025, after a 28-year run. 👀 Savvy Games Group chief Brian Ward is stepping down from the Saudi-government-backed games conglomerate, Bloomberg reports. No word on a long-term successor yet, nor if this augurs any closer integration between Savvy and Electronic Arts. EA was purchased by a consortium led by Saudi’s sovereign wealth fund, in early August. 💿 Square Enix has denied a Japanese media report that the company was planning to go private. In very different Square Enix news, the company revealed that the physical version of Final Fantasy VII: Revelation, the April 2027 finale of its radical remakes of FFVII, will include just one disc in its case, requiring a download for the rest. The first two FFVII remakes in the trilogy shipped on two discs apiece. While Square hasn’t commented on the reason why, a feature here in Game File last month broke down the processes and costs for physical games. The expense of a second disc comes out of the game publisher’s cut of each copy. Square could be saving in excess of a dollar a copy with this move. 🟥 Grand Theft Auto IV’s 26-minute ’‘extended look” gameplay showcase that premiered on Netflix last week—and was exclusive to the platform for six hours—drew 31.1 million views in its first four days, the streamer announced. (Netflix calculates views as total viewed hours divided by runtime; excludes replays). Compared to YouTube: The version that later premiered on Rockstar Games’ YouTube channel is sitting at 22.6 million views after seven days (YouTube’s more liberal counting system tallies anyone who viewed even just the first few second.) Compared to other Netflix mega-events: Last year, Stranger Things season 5’s first four episodes, which debuted together, drew a record 59.6 million views in their first few days on Netflix. For more of a live event type of comparison, in 2024, the Jake Paul vs. Mike Tyson boxing match drew 65 million concurrent streams as it happened; Netflix didn’t release live numbers for the GTA premiere. ☁️ Microsoft is changing the way it charges for cloud gaming, from an all-you-can-play offer for Xbox Game Pass subscribers, to caps for the subscription’s tiers (15 hours per month at the Ultimate tier), with the option to buy more hours in bulk beyond that. The company also said it will sell batches of cloud gaming time to non-Game Pass subscribers, details TBA. Microsoft does not release figures for cloud game usage, but said its changes would impact 4% of Game Pass subscribers. That implies a small portion of its subscribers exceed the 15-hour monthly mark. Cloud gaming offers the promise of playing games without the need for console or PC hardware, as the graphics, sound and controls are all transmitted via a streaming connection. Microsoft notes that its cloud gaming service can be used for a range of things, including Xbox One players using it to play newer-gen Xbox Series games via streaming; and people playing games via streaming, sans any console, “in markets where consoles are less available or affordable.” (That latter use case could even be relevant in Xbox’s home market soon, as console prices keep rising.) Take Two Interactive chairman Strauss Zelnick recently told investors he expects game-streaming tech, which has been only lightly adopted by gamers for over a decade, to be widely used within three years. 🇺🇸 The Trump administration has published multiple agenda-driven web video games to the White House’s website. One called Build The Wall partially clones Tetris as it asks players to build a wall to stop a “zombie border siege.” While actual Tetris involves clearing lines and basically not building a wall, the Trump-backed game omits the line-clearing aspect (and omits the fun as well). On Instagram, Tetris’ official license holders reacted by stating: “At Tetris we believe in the power of connection and bringing people together, not dividing them. To our fans everywhere: we love you, we see you, and we’re grateful to have you in our community. P.S. The Tetris Company was not involved in the creation of ‘Build the Wall.’ P.P.S.: We take copyright infringement very seriously.” Item 3: The week ahead: Tuesday, September 8 Game releases: Mewgenics (consoles; already out on PC) is an engrossing tactical combat game (with cats). It is getting its console port right after the release of a Star Wars tactics game and ahead of the late September launch of a Fire Emblem–a bumper crop for the genre. Wanderburg (PC) is an auto-shooting game like Vampire Survivors, but you’re a castle on wheels Event: Nintendo airs a 30-minute Nintendo Direct (10am ET) focused on The Legend of Zelda’s 40th Anniversary. A full unveiling of the upcoming remake of The Legend of Zelda Ocarina of Time is expected. Will they also show footage from next year’s live action movie? Tease the next big 3D Zelda? Add Link to Mario Kart World? Shadow-drop a new Tingle game? Wednesday, September 9 Event: Nintendo again! On this day, they’re planning to air a 45-minute Nintendo Direct (10am ET) focused on games heading to Switch 2 this winter. Followed by about an hour of gameplay demoed by the company’s Treehouse team. The September Nintendo Directs are always some of the year’s biggest. The 2025 edition included announcements of several major 2026 Switch/Switch 2 games: Pokémon Pokopia, Mario Tennis Fever, Yoshi and the Mysterious Book, Fire Emblem Fortune’s Weave. Plus the return of the Virtual Boy. Friday, September 11 Game release: EA Sports NHL 27 (PlayStation, Xbox) is a hockey game. There are so few that an indie studio that had been making wacky sci-fi first-person shooters just announced that, hey, we’re going to fill a void and make a hockey game for PC.
New ClickLock macOS Stealer Kills Apps Every 210ms Until Victims Type Their Password
ClickLock Stealer kills macOS apps every 210 milliseconds until victims enter a password, then steals browser credentials, cookies, wallets, and Keych
HSSPX: West De Pere FRCC-North Frontrunners
(WFRV) – In this week’s episode of High School Sports Xtra, we featured a thriller in our Game of the Week between West De Pere and Bayport. We also put the spotlight on our Xceptional Athlete from Bonduel, Carter Moesch, and recapped the best matchups from week 7 of the high school football season.
Peter Chan’s Toronto Selection ‘The Road Not Taken’ Boards Moebius
Moebius Entertainment boards international sales for Peter Ho-Sun Chan’s Toronto premiere ‘The Road Not Taken.’
Driving Business Performance and Circular Water Practices at Moeve’s San Roque Energy Park
Optimizing water use through circular solutions is a key goal for Moeve as well as a strategic priority for Spain, due to growing pressure on the country’s water resources.
Newfound 'fire amoeba from the Cascades' sets record for the hottest temperature complex life can survive at
The newly-described amoeba Incendiamoeba cascadensis can survive at a sweltering 145 degrees Fahrenheit.
Tennessee football OL Wendell Moe suffers injury vs Texas
Tennessee football left guard Wendell Moe left the game against Texas on Sept. 26 on the Vols' opening drive.
New ‘fire amoeba’ could assist scientists in finding alien life
Previous research has shown that complex multicellular organisms could only reproduce in temperatures up to 140 degrees.
Venice Lesbian Drama 'Big Little Things' Boards Moebius for Sales
Moebius Entertainment boards worldwide sales for Michelle Zhou's Venice Horizons title 'Big Little Things.'
Giants Trade Pitch Deals Disappointing Former 1st-Round Pick
Bleacher Report's Moe Moton predicts the New York Giants will trade cornerback and former 1st-round pick Deonte Banks. The post Giants Trade Pitch Deals Disappointing Former 1st-Round Pick appeared first on HEAVY.
Moebius Picks Up ‘Happily Ever After’ Ahead of Toronto Premiere
Hong Kong-based sales agency has acquired world rights excluding Taiwan to Fiona Roan Feng-I's second feature, selected for Special Presentations.
I took my Wall Street skills from Goldman Sachs to a dance studio
Caila Moed, 35, went from professional dance, to arts and philanthropy, to Goldman Sachs, to entrepreneurship — and a better work-life balance.
Growing AI energy demands, the search for subatomic particles and the hottest amoeba
From the smallest building blocks of life to prehistoric vomit and a newly discovered amoeba that can really take the heat.
N. Syracuse Manufacturer Marks Milestone, Plans for the Future
Air Innovations has grown from a small local manufacturer into a global business, selling its products in 74 countries organizations. Some of its clients include NASA, Blue Origin and SpaceX. By Stefan Yablonski [caption id=attachment_11117 align=alignright width=450] Michael Wetzel, president and CEO, and his father, Larry, who currently is an adviser to the board of directors. Larry served as the chairman of the board until 2024, when the company was acquired by Altus Capital Partners, a private equity firm.[/caption] Air Innovations in North Syracuse hopes to add more employees and possibly even expand its facility in the near future. The company is a global leader in designing and building environmental process control systems for applications that can’t be addressed with standard HVAC equipment, according to Michael Wetzel, president and CEO. “We customize solutions for temperature, humidity, filtration and pressurization, with additional capabilities in direct-expansion, chilled-water and thermoelectric technologies. Our business now is entirely focused on specialty or niche markets.” Air Innovations celebrated its 40th anniversary on Sept. 18, marking four decades of doing business, manufacturing products and growing in Central New York. In the beginning [caption id=attachment_11138 align=alignright width=451] A worker, Ezzy Salehi, tests a Wine Guardian condenser unit at Air Innovations.[/caption] “A guy by the name of Jim Kelley started the business [known as Floratech] in 1986. It was focused on refrigerated display cases for supermarket fresh flowers,” Wetzel said. “My father [Larry] started Clean Room Technology in the early 1980s, then sold that to a French company called Clestra Cleanroom. “My father and I bought Floratech in 1996 — 30 years ago, 10 years after it was started. Then in 1999, Floratech bought a portion of Clestra and their building. At that point everything was still called Floratech. But, as that was not reflective of the diversity of the business, we rebranded as Air Innovations in 2001.” The company rebranded to pursue a strategy of diversification in specialty HVAC systems. Wetzel explained that since then, it has become a trusted global partner, earning the President’s “E” Award for Exports and securing contracts in the most demanding fields, including aerospace, defense and semiconductor manufacturing. Air Innovations has grown from a small local manufacturer into a global business, selling nearly 70,000 units in 74 countries and territories and developing custom environmental control systems for companies and organizations including NASA, Blue Origin and SpaceX. “We work with all the big names you ever heard of and some you never heard of,” Wetzel added. The Wetzels sold the business two years ago to Altus Capital Partners, a private equity firm focused on control investments in lower middle-market, niche manufacturing companies in North America. Founded in 2003, Altus Capital Partners’ headquarters is in Wilton, Connecticut, with an additional office in Lincolnshire, Illinois. “They are an equity firm. Both my father and I both maintain a small minority share. I am going to continue to be involved. I call myself a part-time CEO. I will continue that role for the foreseeable future; I don’t see any change there.” Building to the future “We are at 58 employees right now, which is a little bit down. We had a high of 70. That moves along with the activity we have,” Wetzel said. “We’ll start ramping up here in the fourth quarter into next year. More than 20% of our employees have been with the company for more than a decade. “I am fairly certain that sometime in the next several years we are going to have to enlarge the footprint of our facility and for sure we are going to have to add employees. We already are. We are probably going to have to add five or six people next year. We are always in the market for HVAC technicians; we always need those.” Air Innovations will remain in Central New York, with its current workforce staying in place, he added. “I believe the primary driver of our success over time is the diversity of our business which is a direct result of our willingness to tackle new challenges,” Wetzel said.
Science of baseball event brings Evansville Otters to CMOE
Visitors learned about baseball science at CMOE with Evansville Otters players
The Clicks Communicator Phone With Blackberry Roots Demoed in New Video
A new video shows the $500 Clicks Communicator phone actually working. It offers a preview of all the typing and messaging that's to come.
North Carolina teen dies from rare brain-eating amoeba found in freshwater: Health officials
State health officials say the teenager died Monday.
A pro-Ala Stanford super PAC has pulled its ads from TV in the final stretch of the Philly congressional race
Sources said that 314’s move to pull its TV advertising came after the organization commissioned a poll that showed Stanford’s voter support declined precipitously in recent weeks.
Google Maps brings a 3D map to your driving directions
Maps is now the latest Google service to receive a Gemini makeover, with a redesign of the driving experience headlining the update.
The Long Fight for Small Business Lending Transparency Is Not Over » NCRC
Small businesses play a vital role in the US economy, but their ability to access credit has been a longstanding obstacle to their viability. Additionally, the CFPB's recent final revised rule implementing Section 1071 of the Dodd-Frank Act that effectively cuts small business lending reporting requirements could lead to more lending disparities for women and BIPOC entrepreneurs.
North Carolina teen dies from rare brain-eating amoeba found in freshwater: Health officials
State health officials say the teenager died Monday.
Jumping into the lake? Watch out for brain-eating amoebas
The Department of Health is reminding people who are active in Florida’s freshwater to wear nose clips to avoid infections by a brain-eating amoeba presently active.
Scientists identify ‘ghost’ of a long-extinct relative in humans today
Scientists retrieved proteins from six teeth unearthed in China that reveal a potential link between Homo erectus and later human species, including Homo sapiens.
Minneapolis economy: Busy playoff season brining business boost | FOX 9 Minneapolis-St. Paul
A busy basketball playoff season is bringing a big economic boost to downtown Minneapolis. FOX 9's Leah Beno sat down with President and CEO of the Minneapolis Downtown Council to learn more.
Driving Business Performance and Circular Water Practices at Moeve’s San Roque Energy Park
Optimizing water use through circular solutions is a key goal for Moeve as well as a strategic priority for Spain, due to growing pressure on the country’s water resources.
Engrams Embedding Entendre: Codesign for Efficient DRAM/SSD Offloading
Engram extends standard token embeddings with learned multi-token lookups. Recurring local patterns retrieve vectors directly, reducing the need to reconstruct them through attention and feed-forward layers. With Engram model architecture optimization, it allows for lower HBM capacity to be needed for models at the same quality. This does not mean there won’t be an insane demand for HBM but it just means that model architecture will continue to innovate around constraints. This model architecture design is naturally codesigned for parameter offloading: each token accesses a few embedding rows whose addresses depend on token IDs, not hidden states. The runtime can prefetch those rows from host DRAM while earlier layers compute, keeping the table outside HBM without transferring entire weight matrices. Our Memory model contains our latest estimates of quarter by quarter HBM, DRAM, & NAND supply and demand. Offloading frees HBM for model weights and KV cache, potentially supporting larger batches or more concurrent sessions. When DRAM becomes the next constraint, NVMe offers another tier. Recommendation systems already cache frequently or recently accessed embedding rows in faster memory while backing colder rows with SSDs. After NVIDIA roadmap had to change due to massively despec’ing Rubin Ultra from 1024GB to now ~200GB of HBM per chip, model architecture optimizations like emgram maybe helpful. Our DeepSeek-V4.1-Flash configuration uses roughly 189 GiB of memory for Engram. We replace it with a memory-mapped (mmap) file and measure serving performance with offload to SSD. Later on in our report, we will show our Engram offloading experiments along with the official InferenceX agentic inference serving results on Engram models like DeepSeekv4.1 Flash across all 6 NVIDIA GPU SKUs along with MI355X. Unsurprisingly, the CUDA Moat is still mogging MI355X on the ultra popular DeepSeekV4.1 Flash model. We also show how even on high capacity HBM SKUs, offloading emgrams to DRAM could result in even better performance for most of the pareto than keeping the emgram in HBM. Our benchmark has been widely reproduced, validated and/or supported by almost every major buyer of compute from Google Cloud to Microsoft Azure to Oracle, to Meta and many more. Furthermore, it has the support of the ML community including from vLLM, LMCache, SGLang, PyTorch, Huggingface and the support of major labs like OpenAI, MiniMax, ZAI, Qwen, Moonshot Kimi, etc. Star the InferenceX GitHub repository if you find the open-source benchmark and data useful!. InferenceX is the only inference benchmark in the world to have TPUv7, Jalapeño, Nvidia Rubin NVL72, AMD, and soon, SambaNova and Trainium. Due to how realistic AgentX scenario is to real world agentic inference workloads, AMD has committed to collaborating on MI455X UALoE72 too. DeepSeek did not release the original paper’s two trained Engram models. We replicated its setup on fineweb-edu using the released code and training hyperparameters, at an estimated 6E18 FLOPs per run. We observed the same U-shape scaling: Engram improved performance over pure MoE baselines. We also reproduced DeepSeek’s results, where earlier-layer representations with Engram resembled those of later layers. Like the original Engram paper, it is possible to probe Engram’s gate scores to see what n-grams DeepSeek-V4.1-Flash makes the most use of. Our gate scan finds names, code fragments, relational phrasing, and boilerplate. These examples prioritize interesting-ness over gate strength. One unexpected result was Wright : Ace Attorney. These examples suggest learned memory optimizes the training objective, not a judgment of which facts deserve storage. Licenses, bibliography fragments, API scaffolding, and website furniture can provide prediction shortcuts, so the value of additional Engram capacity may depend on what survives data preparation. This does not show that table capacity is “wasted”: the evaluation-corpus scan establishes neither training exposure nor the capacity occupied by each category. For offloading, strong gates do not identify cache-hot rows. Low gates do not automatically save reads either: computing the gate requires the retrieved key and undoes the performance gain of a fused kernel. Skipping reads would require a separate usefulness predictor before retrieval. In the original paper’s inference-time ablation, factual-knowledge benchmarks retained just 29–44% of their original performance, while reading comprehension retained 81–93%. This is due to the training–inference mismatch. The resulting degradation therefore measures this trained model’s dependence on Engram, not the performance difference between models trained with and without it. In our ablations, suppressing Engram worsens token likelihood across all evaluated domains, especially encyclopedia text and several code corpora. Surprisingly, GSM8K accuracy stays within measured run-to-run variation and removing Engram has no effect. Engram is not a detachable dictionary beside an unchanged MoE. Removing it changes downstream features and expert selection. We tested whether rerouting hurts or compensates by holding tokens fixed in a teacher-forced experiment on CRUXEval, a code-reasoning benchmark of small Python functions where the model predicts a function’s output from its code and an input, and scoring the reference answer. Removing Engram raised answer loss from 0.2848 to 0.3093 bits/token. Forcing the ablated model to use the original Engram-on expert choices made it worse still, at 0.3375 bits/token. Rerouting partially compensates for the missing memory. Memory features and expert selection work together, rather than following a clean “memory stores facts; experts reason” division. On the same CRUXeval, removing Engram during either phase reduced accuracy and increased generated tokens; removing it throughout produced the largest changes. Keeping Engram for prefill leads to more correct answers than keeping for decode likely due to semantically richer KV cache transferred to decode workers, allowing it to mitigate some of the performance loss. Engram’s table is large, but each lookup is small. DeepSeek-V4.1-Flash requests 24 rows at each of two Engram layers, about 12.4 KiB per processed token position across the model, or 3.1 KiB per GPU when split across four GPUs. Currently as of Day 7 since Model Release, MI355X is still 2-4x worse performance per dollar compared to B200 even when normalized by Mi355X’s lower TCO. Our full total cost of ownership breakdown comes from our AI Cloud TCO Model along with monthly market surveys of over 100+ gpu clouds & gpu cloud customers. On the Day 0 release of DeepSeekv4.1 Flash, NVIDIA vLLM works out of the box with zero issues across all 6 SKUs: H100, H200, B200, B300, GB200, GB300! This was thanks to the amazing work by the NVIDIA & Interact teams! In comparison, AMD vLLM did not work on day 0 for DeepSeekv4.1 Flash.Source: SemiAnalysis InferenceX AMD’s vLLM documentation points to using vllm/vllm-openai-rocm:deepseekv41-flash-0909, but from hour 0 of the model release to hour 23, AMD has not publicly released the image. AMD claims, “SPEED IS THE MOAT,” yet it has still not released it by the 23rd hour. We wish that, going forward, the AMD team has a better process for hour 0 model releases. Eventually when they did publicly release for “day 0” image support, performance-wise, it is currently up to 14.8x worse perf per dollar than H200 and up to 42x worse perf per dollar than B200/B300. The power of the CUDA MOAT is NVIDIA’s collaboration with its massive 6 million-developer community ecosystem including most of the vLLM & SGLang & Tokenspeed maintainers which means that CUDA is optimized on day 0. Overall, AMD did make significant improvements but the performance per dollar is still currently 2-4x worse than B200. AgentX Engram DRAM Offloading Improving Performance The HBM and DRAM offload use the same GPU kernel to select and dequantize rows. With HBM, it reads GPU memory; with Unified Virtual Addressing (UVA), it reads pinned host memory directly. Both support full decode graphs. Moving the table into HBM accelerates only the sparse lookup, leaving decoder computation and communication unchanged, resulting in little overall benefit while consuming memory otherwise available to KV cache. Another benefit of Engram offloaded to DRAM which means you can reduce the communication overhead by using less HBM GPUs per replica. For example, when enabling Engram offloading on B300, we are able to switch from TP4 to now TP2 which improves the pareto curve by up to 1.6x. When iso-model quality, less HBM is required as the engrams could be offloaded to host DRAM. Thus HBM bandwidth matters way more than HBM capacity. For inference workloads where memory bandwidth matters the most, 4-hi HBM provides the best $/bandwidth and therefore lowest cost per token. If China continues to make more and more revolutionary model architecture innovations, soon it could potentially 0Hi HBM stacks. [ ](https://newsletter.semianalysis.com/p/long-live-the-short-king-why-4-hi) [ Long Live the Short King: Why 4-hi HBM Wins ](https://newsletter.semianalysis.com/p/long-live-the-short-king-why-4-hi) Myron Xie, Bryan Shan, and 3 others · Sep 13 [ Read full story ](https://newsletter.semianalysis.com/p/long-live-the-short-king-why-4-hi) Moreover, on B300 and week-0 stack, moving the Engram table back to HBM did not improve results and stay within run-to-run variance. This is a result of the work optimizing DRAM offload, such as async, and overlap. SSD Offloading On B200, we created a unoptimized vLLM fork and stored the Engram tables in memory-mapped files on local SSDs. File backing lets the OS reclaim table pages when other applications need RAM. Pages already cached in memory can be served without reading the SSD again. Furthermore, note that we were unable to turn on GDS The unoptimized SSD implementation changes how rows reach the GPU. It copies row IDs to the CPU, deduplicates them, gathers the requested rows into pinned buffers, copies those rows back to the GPU and dequantizes them. This work runs between segments of the GPU execution graph. Native UVA performs row selection and dequantization directly on the GPU, avoiding the CPU round trip. A warm filesystem cache removes physical SSD reads, but leaves the coordination, row gathering and transfers. This is why a file already cached in RAM can still perform worse than a pinned DRAM table. The comparison measures the whole serving path; it does not separate the time spent on each of these operations. B200 DRAM dominates both measured SSD serving curves in total tokens per dollar and P90 interactivity. Near 125 tokens/s/user, DRAM delivers 121 million total tokens per dollar versus 52 million for SSD. For production serving, SSD offloading is likely not worth the tradeoff. On the B200 configurations we measured, SSD offloading loses on both measures: every observed SSD point has a DRAM alternative that delivers higher P90 interactivity and more total tokens per dollar. The only points where Cheaper storage does not automatically produce a cheaper inference service. Moving Engram to SSD leaves the same four expensive GPUs and the rest of the server in place. Reclaiming RAM only creates an economic benefit if it enables a cheaper server configuration or additional useful capacity. The current unoptimized path provides neither benefit, and the filesystem cache still consumes RAM when table pages are resident. Next we will look at the mechanisms and specific implementation of ngrams from DeepSeekv4.1 Flash, LongCat, Qwen3.8 Flash Next.
Deadly Brain-eating amoeba linked to Durham facility; health dept says creek lacked required permits
A man-made creek at a Durham facility was the likely source of exposure to the brain-eating amoeba that killed a 17-year-old boy, NCDHHS says
Brain-eating amoeba that killed 17-year-old linked to a man-made creek, health officials say
A man-made creek was likely the source of exposure to the brain-eating amoeba that killed a 17-year-old boy, health officials said.
Exeter batteries drained in Harlequins loss
Exeter boss Rob Baxter says his side looked like a team in need of a rest during their 41-24 defeat by Harlequins.
Aiden Newdigate runs in touchdown for St. Xavier vs. Moeller
Aiden Newdigate's 1-yard touchdown run tied the game between St. Xavier and Moeller with 10:30 left in 2Q on Sept. 18, 2026.
Here’s how we used AI to find gambling ads in televised sports
The Post used AI to analyze 50 televised sports games for references to betting.
How AI-Powered Public Comments Are Entering US Climate Politics
Officials have been confounded by the new phenomenon.
Scientists sound the alarm as dangerous amoebas spread globally
Scientists warn that free-living amoebae may be an underappreciated public health threat, capable of causing deadly infections and shielding other dangerous microbes from water treatment. Climate change and aging infrastructure could help these resilient organisms spread more widely in the years ahead.
This 'fire amoeba' pushes the limit of what's possible for complex life
Scientists have found a species they've named the fire amoeba, which is capable of reproducing at 145°F and continuing to move up to temperatures of 147°F — a record for complex eukaryotic life.
Moebius Entertainment boards ‘Happily Ever After’ ahead of ...
The cast includes Shuhua, the Taiwanese singer from K-pop group i-dle, in her feature debut.
Apple Releases First iOS 27, iPadOS 27, and macOS 27 Betas to Developers
Following the WWDC 2026 keynote event, Apple has seeded the first betas of iOS 27, iPadOS 27, macOS 27, watchOS 27, tvOS 27, and visionOS 27 to developers for testing purposes. Registered developers can opt in to betas and download the software through the Settings app on each device. The software includes many of the new features that were shown off earlier today, though some of what Apple demoed won't be available right away.
This Is An Amoeba Reproducing At 63°C, The Highest Temperature Ever Recorded For A Complex Cell
The appropriately named fire amoeba can live its entire life at a temperature hot enough to cook an egg.
Double Success for ISTA at the Lower Austrian Science Awards
3D-printed thermoelectric materials and expansion microscopy of brain tissue—these are the contributions that earned two researchers from the Institute of Science and Technology Austria (ISTA) distinctions at the Lower Austrian Science Awards. Maria Ibáñez and Johann Danzl accepted their awards at a
North Carolina water feature linked to teen's fatal brain-eating amoeba infection
Durham County health officials closed the feature Aug. 28.
Transforming a fossil fuel firm into a clean energy leader
Energy companies Ørsted and Moeve offer their insights on how to transform fossil fuel-based businesses around opportunities offered by the clean economy.
Walking an extra 1,000 steps a day may help your heart, study finds
Here’s the latest science on how many steps you actually need every day.
‘The reality, for better or worse’: Columbia comp sci students and faculty grapple with AI’s disruption of the field
‘The reality, for better or worse’: Columbia comp sci students and faculty grapple with AI’s disruption of the field ‘The reality, for better or worse’: Columbia comp sci students and faculty grapple with AI’s disruption of the field A field that once seemed like a direct path to stable, lucrative work is becoming less certain under AI, students and faculty told Spectator. By Ria Vasishtha and Arjun Menon May 3, 2026 Shanying Liu / Deputy Illustrations Editor Asia Genawi, SEAS ’29, first encountered artificial intelligence during her sophomore year at her high school in Indiana. Back in 2023, policies on AI varied widely—even within the same district—and some schools had no guidance at all. As Genawi began to see the effects of AI trickle into her classroom, she turned to guidance from her home state’s Department of Education, which she said she found “outdated and inaccessible.” Genawi spent the rest of her high school career dedicated to the issue, including writing a 15-page memo exploring varying AI policies across Indiana schools, which she sent to every Indiana state senator and representative whose contact information she was able to find. She later helped draft a bill that would have required public high schools to establish and share their own AI policies with students in an accessible format. But with so many other bills already awaiting referral, hers never made it to committee. When Genawi arrived at Columbia to study computer science, she found that the gaps she had spent years trying to address in Indiana had followed her there. Many students and professors were also grappling with what AI meant for the field. Across Columbia’s undergraduate schools, computer science has become the most popular major—accounting for around 11 percent of degrees awarded in Columbia College, 32 percent in the School of Engineering and Applied Science, and 12 percent in the School of General Studies in 2024. While the number of computer science majors in SEAS grew from 166 to 173 during the 2024-25 academic year, the total number across the three schools fell around 4.5 percent from 393 to 375, driven by declines at Columbia College and General Studies. At Barnard, it has grown to become the second most popular major as of 2025, up from third the year prior. The number of degrees awarded have grown fivefold since 2016. Faculty members and students in the department told Spectator that the surge in the major’s popularity has been driven in large part by its apparent promise of a secure, well-paid career path. However, they noted that the recent rise of generative AI has compromised the sense of stability that once defined the field, even as interest in technical skills among students persists. According to data from the Federal Reserve Bank of New York in 2024, out of 74 majors tracked, computer science had the fifth highest unemployment rate among majors at 7 percent, while computer engineering had the second highest at 7.8 percent. “Anecdotally, people are moving away from CS as a major,” Daniel Bauer, a senior lecturer in the computer science department, said. “They’re still taking some of our classes, right, because they think they need that, but they’re instead majoring in other things.” Bauer described a field that had shifted from what he called a “nerdy outlier discipline” to one that “everyone needs,” a sign that technical skills have become a baseline expectation across industries with the emergence of AI. “A lot of people thought, ‘Oh, it’s a guaranteed path to a stable income. You get a six-figure job right out of your undergrad,’” Bauer said. Now, companies may see AI as a way to reduce demand for entry-level programmers, especially in what he calls “code monkey jobs,” roles where a large number of employees are “just writing out code nine-to-five.” Chris Murphy, a senior lecturer in the computer science department, described the current moment as the convergence of two forces: the downturn of the job market and generative AI coming onto the scene. “Things kind of started to really pick up in 2014, 16,” he said, recalling the years after the weaker 2010 market. Murphy added that this was also a period where “students were getting more experience outside of class than they were inside of class” through internships, jobs, and side projects. Murphy suggested that the earlier momentum helped make computer science feel like both an academic program and a reliable pathway into work. That feeling is harder to sustain now as the industry reconfigures, which is a shift that students are seeing firsthand. Rebecca Yu, SEAS ’27, said that during a recent internship in Big Tech, she saw that “every single tech company was racing to integrate” large language models. She added that her manager said her intern class was the first to complete their assigned projects, which she attributed to the increased efficiency that comes with AI. Frank Liu, SEAS ’29, said that with AI lowering the barrier to coding, companies now expect everyone to write code. “A lot of companies outright say if you’re not using coding in your job, you might get fired for not being productive enough,” Liu said. Yu similarly expressed that students increasingly have to accept AI being a part of industry workflow, adding that many also “make use of it to maximize their productivity.” Richard Li, CC ’28, described a “franticness” among his peers as the pace of change accelerates. He explained how “some new tool” is released regularly, noting that tools rise and fall in prominence within weeks. He has come to believe that he doesn’t need to “match that speed all the time” and keep up with every development. Li has learned that AI tools allow him to “take a breath” and “get back into the loop much easier than before” when he needs to. While some companies specifically target Columbia when hiring, sending representatives to campus career fairs and recruiting large numbers from Columbia, a computer science degree cannot be treated as a static credential, Yu described. “You can’t just get your degree and stop learning after that. You have to always be learning,” Yu said. “That is something you basically have to expect for your job.” AI has not only affected students’ decisions to major in computer science, but also reshaped how the subject is taught. Across departments, faculty members no longer treat work done outside the classroom as a reliable measure of understanding, shifting instead toward in-person exams, quizzes, and attendance checks. While homework once made up as much as 60 to 70 percent of a student’s grade in computer science classes, Murphy said that breakdown has since flipped, with exams and quizzes now accounting for the majority. “The assumption—especially among pessimists—is that they’re using it to do their homework,” Murphy said. “They take the homework assignment. They put it into generative AI. Out comes the solution. Submit solution. The end.” Murphy said that faculty members ranged in their responses from fully prohibiting AI use to actively teaching students how to effectively work with AI. He said he shifted his approach this year after his teaching assistants called his belief that students weren’t using the technology “naïve,” and now permits students to use AI on assignments. His surveys of students in his introductory classes have told a more complicated story than the one so-called pessimists offer: Around 60 to 70 percent of students who reported using AI said they used it to have concepts explained to them, to clarify assignment instructions, to generate test cases, or to create practice questions. “Those are all ways that you might have had a tutor in the past to do that, and now you have this AI essentially as your tutor,” he said. Bauer has taken a different approach, teaching students how to use AI productively without sacrificing the core principles of the discipline. “People who are trained to actually use AI tools productively, while being in charge of designing the overall project, will be very much in demand in the next couple of years,” he said. Over the past year, a working group of computer science faculty has been redesigning the introductory and intermediate programming sequence to integrate generative AI as both a subject and a tool, representatives from SEAS wrote in a statement to Spectator. The effort is “coordinated” across the department, “not course-by-course,” with a systematic assessment of how AI integration is affecting student outcomes underway to shape the next round of changes. The goal, the department explained, is for students to be able to “read, verify, and reason about code, whether they or a model wrote it.” Tony Dear, senior lecturer in the computer science department, is piloting individualized AI feedback in a discrete mathematics course, while a machine learning course is also testing a Socratic AI dialogue system designed to “guide inquiry rather than supply answers,” the statement reads. Beyond individual courses, SEAS funded seven cross-departmental faculty projects in generative AI this past academic year, spanning biomedical engineering, chemical engineering, industrial engineering, and computer science, according to the statement. This phenomenon extends beyond the computer science department. SEAS’ new AI minor, which it announced in November 2025, reflects the University’s attempt to adapt. The minor is specifically designed for noncomputer science majors and aims to provide “essential AI knowledge” while emphasizing practical and responsible use across disciplines, according to the SEAS website. At the center of the minor is AI in Context, a course taught by a team of faculty from the departments of computer science, applied mathematics, philosophy, music, literature, and the Writing Center, the representatives told Spectator. Bauer described it as targeted at students who want to bring AI into fields like economics or biology, rather than as a path into software engineering itself. At the graduate level, SEAS announced a new Master of Science in artificial intelligence in February, with the inaugural cohort starting in the fall. While recognizing how AI has helped them navigate coursework, students also described its rise as isolating, even among members of the Application Development Initiative, one of Columbia’s largest computer science clubs. “As people start to ask AI for questions that they would normally ask students and members, it starts to isolate people” Li said. Alexa Kafka, BC ’27, echoed Li’s sentiment, pointing out that computer science has always been a collaborative discipline with code reviews and pull requests, even as AI makes independent work easier. Despite this, Kafka noted that she hasn’t been to office hours “in at least a year.” Bauer added that faculty are exploring long-term solutions to the assessment problem. “One of the things that we’re trying to do at the department or school level is to maybe move towards computerized testing facilities,” Bauer said, referring to ideas in the department to revitalize computer labs as an alternative to pen-and-paper exams. “We have to accept that this is the reality, for better or worse.” But more exams, students say, does not mean better learning, nor does it reflect how computer science is actually done. “Code is so variable and can be approached in so many directions, there’s never just going to be one correct answer. There can be multiple optimal solutions,” Kafka said. Students also expressed concern that the shift toward exam-heavy grading pulls focus away from the applied, project-based work that better reflects both discipline and industry expectations. ADI has tried to respond more quickly than the curriculum can. Through a vibe coding workshop this past semester, club members said they wanted to help students understand not just that AI exists, but how to use it deliberately. “Vibe coding” refers to the practice of prompting AI to write code for a task or project, and has become widespread enough that ADI felt the gap in formal instruction was worth addressing, according to Sophia Huang, BC ’29. The workshop introduced students to the concept of context engineering, which Shaina Sahu, SEAS ’27, described as not just writing a prompt, but optimizing “the rest of the relevant information that you would feed into the model.” Without that, students said that AI-generated code can fail at scale or introduce difficult-to-detect bugs. Genawi said the shift toward AI has changed what it means to be good at computer science. “I think the shift has kind of gone towards what can you create and what can you do, instead of, what do you know,” she said. Indeed, during ADI DevFest, its annual hackathon, Li said more teams are trying to complete their projects than in years past. While sign-ups have historically outpaced submissions, Li said this is beginning to change because teams “have a tool that completes the vision for them.” While many students see AI as an equalizer that lowers the barrier to entry for coding and technical work, Murphy said it also creates new equity concerns inside the classroom around who can afford stronger tools and who already knows how to use them. “The equity issues I worry about are students who have access to paid versions of things versus free versions of things,” he said. Although popular AI models like ChatGPT and Claude have free versions available, both have paid models that can reach over $100 per month, which output different qualities of code. Murphy also pointed to differences in prior exposure to AI literacy itself. “Students who go to high schools or K-12 education, where they’re taught AI literacy, prompt engineering, context engineering, they understand how to do this, as opposed to students who go to schools who don’t teach that,” he said. “Then they come in, and then they’re somehow expected to know how to use these tools.” For Genawi, those disparities begin long before students arrive on campus. “It’s very interesting to consider how lucky you were geographically,” she said. “To be born where you were is influencing your ability to get access to these spaces.” Students like Yu, though broadly apprehensive about the future of computer science, are not without hope. “I’m optimistic, because I think if you’re truly passionate about CS, there will definitely be a place in the field to do something you want to do,” she said. Senior Staff Writer Ria Vasishtha can be contacted at ria.vasishtha@columbiaspectator.com. Follow Spectator on X @ColumbiaSpec. Staff Writer Arjun Menon can be contacted at arjun.menon@columbiaspectator.com. Follow Spectator on X @ColumbiaSpec. Want to keep up with breaking news? Subscribe to our email newsletter and like Spectator on Facebook. Produced with Spectate by Emily Ta
Kobie Turner extension should be atop Rams’ to-do list
All attention is on the Rams re-signing Puka Nacua, but we cannot forget about Kobie Turner
Eri Hino
Fortune Brands Innovations has promoted Eri Hino to VP, Product Management – Moen.
What's Coming in September: New iPhones, Apple Watches and More
Apple will introduce at least five new devices at its iPhone-centric event next month, and it's also possible we'll get devices waiting on Siri AI like the Apple TV. iPhone 18 Pro The iPhone 18 Pro is going to look like the iPhone 17 Pro, but with a smaller Dynamic Island. It's getting some internal updates, including a new Wide camera with a variable aperture for more control over lighting and a faster A20 Pro chip built on a new 2nm process.
MOESC business advisory council highlights workforce programs and 2026-2027 meetings
The Mid-Ohio Educational Service Center Business Advisory Council met Wednesday, according to an MOESC press release.
Philadelphia Eagles' dynamic duo named one of NFL's best
The Philadelphia Eagles' dynamic receiving duo of A.J. Brown and DeVonta Smith was recently named one of the best in the entire NFL by Moe Moton of Bleacher Report. The pairing of Smith and Brown was named as the fifth-best wide receiver duo in the league, behind Ja'Marr Chase and Tee Higgins in Cincinnati, Puka Nacua And Davante Adams in L.A., CeeDee Lamb and George Pickens in Dallas and Amon-Ra St. Brown and Jameson Williams in Detroit.
Alexander Ovechkin receives standing ovation from rival Penguins crowd in potential last game in Pittsburgh
Hamilton West Little League honored with a proclamation. See video
The Hamilton West Little League team was honored with a proclamation for their historic run. The proclamation was presented by Hamilton Mayor Pat Moeller.
Bills Roundtable Podcast Ep 49: An Unstoppable Offense
Josh Allen’s high-powered unit is dismantling defenses, but major defensive lapses and key injuries threaten Buffalo’s perfection. Mike Straw and Anthony Moeglin debate if this formula is sustainable.
Goldman Sachs, Moelis Say Business as Usual Amid Iran Conflict
Goldman Sachs Group Inc.’s international co-heads say the turbulent start to the year has done little to deter clients, underscoring Wall Street’s bullish outlook despite events like the Iran war.
Scientists sound alarm as dangerous amoebas spread globally
Free-living amoebae are emerging as a global health concern, fueled by warming temperatures and outdated water systems. While many are harmless, some can cause deadly infections and even protect other dangerous microbes. Their ability to survive heat and disinfectants makes them especially hard to control. Scientists say improved surveillance and water treatment are urgently needed.
ASUS Demos Its First “ROG” DDR5 Memory Running at 8800 MT/s On Its Crosshair X870E APEX Motherboard
ASUS's first DDR5 memory modules, the ROG 20th Edition, have been demoed on the ROG Crosshair X870E APEX motherboard.