[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"section-science":10,"sections":1302},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"section":11,"sections":16,"articles":81},{"name":12,"slug":13,"count":14,"latest_published_at":15},"Science","science",66,"2026-07-10T10:29:37.000Z",[17,22,27,32,37,42,47,52,57,62,66,71,76],{"name":18,"slug":19,"count":20,"latest_published_at":21},"AI","ai",2590,"2026-07-16T04:00:00.000Z",{"name":23,"slug":24,"count":25,"latest_published_at":26},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":28,"slug":29,"count":30,"latest_published_at":31},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":33,"slug":34,"count":35,"latest_published_at":36},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":38,"slug":39,"count":40,"latest_published_at":41},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":43,"slug":44,"count":45,"latest_published_at":46},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":48,"slug":49,"count":50,"latest_published_at":51},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":53,"slug":54,"count":55,"latest_published_at":56},"Dev Tools","dev-tools",59,"2026-07-07T04:00:00.000Z",{"name":58,"slug":59,"count":60,"latest_published_at":61},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":63,"slug":64,"count":60,"latest_published_at":65},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"General","general",29,"2026-07-10T22:28:58.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"How-To","how-to",6,"2026-06-16T09:00:00.000Z",[82,109,126,144,163,180,198,214,231,247,262,278,294,312,327,345,361,377,402,420,436,454,470,486,501,523,543,560,579,598,619,638,656,683,699,720,738,758,780,797,815,834,860,883,902,932,950,969,988,1004,1024,1044,1062,1081,1100,1116,1133,1152,1170,1186,1205,1221,1238,1256,1271,1286],{"id":83,"slug":84,"title":85,"dek":86,"body_md":87,"tags_json":88,"published_at":15,"created_at":89,"updated_at":90,"status":91,"review_note":92,"review_notes":93,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":100,"sources":104,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4604,"humanoid-robots-remove-organs-in-live-animal-surgery","Humanoid Robots Remove Organs in Live Animal Surgery","UC San Diego surgeons used teleoperated humanoid robots to extract gallbladders from live pigs, a claimed first for the field.","Humanoid robots just performed live organ removal surgery on animals for the first time, according to researchers at UC San Diego.\n\nSurgeons at the University of California San Diego guided two teleoperated humanoid robots — machines built roughly in the human form — through the removal of gallbladders from live pigs. The robots were not autonomous; human surgeons controlled them remotely. The lab setting, not an operating theater, was the venue. The researchers describe this as a medical first for humanoid robot platforms.\n\nThe distinction matters because most surgical robots in clinical use today, like the da Vinci system, are purpose-built arms that look nothing like a person. Humanoid platforms are designed to operate in environments built for human hands and bodies, which could make them easier to deploy in existing surgical suites without retooling the room. If humanoid robots can match the precision of dedicated surgical hardware, the case for general-purpose machines in medicine gets meaningfully stronger.\n\nFor now, this is a proof of concept on pigs in a controlled lab — a long runway separates a gallbladder extracted under research conditions from routine human surgery, and that gap has swallowed more than a few promising headlines before.","[\"robotics\",\"surgery\",\"medical-research\",\"ai\"]","2026-07-10T11:37:51.412Z","2026-07-10T11:37:54.363Z","published",null,[94],{"id":95,"reviewer":96,"round":97,"reason":98,"status":99},"editor-r1","editor",1,"The article ends abruptly without a concluding sentence or thought — the final paragraph cuts off mid-analysis and reads as a fragment.","resolved",[101,102,103,19],"robotics","surgery","medical-research",[105],{"name":106,"url":107},"The Next Web","https:\u002F\u002Fthenextweb.com\u002Fnews\u002Fhumanoid-robots-surgery-live-pigs-uc-san-diego",0,{"id":110,"slug":111,"title":112,"dek":113,"body_md":114,"tags_json":115,"published_at":116,"created_at":117,"updated_at":118,"status":91,"review_note":92,"review_notes":119,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":120,"sources":122,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4547,"surgeons-used-humanoid-robots-to-remove-pig-gallbladders","Surgeons Used Humanoid Robots to Remove Pig Gallbladders","A Nature-published preclinical trial shows human-controlled humanoid robots completing minimally invasive surgery on live pigs for the first time.","Humanoid robots, operated remotely by surgeons, removed gallbladders from live pigs in a first-of-its-kind preclinical trial.\n\nThe experiment, published in the journal Nature, involved teleoperated humanoid robots performing minimally invasive procedures on living animals. Surgeons controlled the robots' movements throughout — these were not autonomous machines making independent decisions. The research team behind the trial includes Shanglei Liu, an assistant professor of surgery at the UC San Diego School of Medicine.\n\nThe significance here is less about the surgery itself and more about the hardware. Specialized surgical robots already exist, but they are expensive and require significant floor space, limiting their use to well-resourced hospitals. Humanoid robots, by contrast, could slot into smaller operating rooms at a fraction of the cost — making remote robotic surgery plausible for rural clinics, military field settings, or, per Liu, even space.\n\nThe gap between \"worked on pigs in a trial\" and \"cleared for human patients\" is substantial, and Nature publications have a way of generating headlines that outrun clinical timelines. Still, if the cost and size advantages hold up under scrutiny, this is a more pragmatic case for humanoid robots in medicine than the autonomous-surgeon narrative that usually dominates the pitch deck.","[\"robotics\",\"surgery\",\"medical research\",\"ai\"]","2026-07-09T20:12:03.000Z","2026-07-09T21:30:26.068Z","2026-07-09T21:30:28.995Z",[],[101,102,121,19],"medical research",[123],{"name":124,"url":125},"Ars Technica","https:\u002F\u002Farstechnica.com\u002Fai\u002F2026\u002F07\u002Fhumanoid-robots-controlled-by-surgeons-did-world-first-operation-on-live-pigs\u002F",{"id":127,"slug":128,"title":129,"dek":130,"body_md":131,"tags_json":132,"published_at":133,"created_at":134,"updated_at":135,"status":91,"review_note":92,"review_notes":136,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":137,"sources":140,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4543,"uc-san-diego-humanoid-robots-pull-off-gallbladder-surgery","UC San Diego Humanoid Robots Pull Off Gallbladder Surgery","A preclinical trial at UC San Diego used teleoperated humanoid robots to complete two gallbladder removals on non-primate mammals — a first.","Humanoid robots just assisted in — and in one case fully performed — live gallbladder surgery, no human hands required for the main act.\n\nResearchers at UC San Diego completed two preclinical surgeries using a teleoperated humanoid robot named Surgie, a five-foot, 60-pound machine built for operating room work. In the first procedure, Surgie assisted a human surgeon on a laparoscopic gallbladder removal. In the second, two Surgie units performed the same procedure without a human surgeon at the table at all. Both surgeries were on non-primate mammals, not humans — a distinction worth keeping in mind before updating your insurance card.\n\nThe case for humanoid surgical robots goes beyond novelty. Existing robotic surgery systems like Intuitive Surgical's da Vinci are fixed, expensive installations that require a specialist on site. A mobile, lightweight robot that can walk into a field hospital or a rural clinic is a meaningfully different proposition. UC San Diego's Michael Yip, a senior author on the paper, pointed specifically to remote communities with staffing shortages and mass-casualty scenarios as target use cases — situations where the current surgical infrastructure simply does not reach.\n\nThe caveats are real: Surgie needed recalibration mid-procedure, and latency between human operator and robot response remains unsolved — a serious problem if you are trying to operate from a satellite uplink in a disaster zone. The research team sees a near-term role for Surgie as an operating room assistant rather than a lead surgeon, fetching tools and handling cleanup while humans focus on the critical cuts. That is a narrower pitch than the headlines suggest, but it is a defensible one.","[\"robotics\",\"healthcare\",\"surgery\",\"research\"]","2026-07-09T15:58:31.000Z","2026-07-09T16:35:13.799Z","2026-07-09T16:35:16.683Z",[],[101,138,102,139],"healthcare","research",[141],{"name":142,"url":143},"Mashable","https:\u002F\u002Fmashable.com\u002Ftech\u002Fhumanoid-robots-teleoperated-gallbladder-surgery-uc-san-diego",{"id":145,"slug":146,"title":147,"dek":148,"body_md":149,"tags_json":150,"published_at":151,"created_at":152,"updated_at":153,"status":91,"review_note":92,"review_notes":154,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":155,"sources":159,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4396,"quantum-learners-beat-classical-ones-at-predicting-particle-dynamics","Quantum Learners Beat Classical Ones at Predicting Particle Dynamics","A new proof shows quantum machine learning algorithms can solve a physics prediction task that no classical algorithm can match in polynomial time.","Quantum computers just got a formally provable edge in a machine learning contest — and this one comes with a mathematical certificate.\n\nResearchers posted a paper to arXiv demonstrating a rigorous \"learning separation\": a supervised learning task where a quantum algorithm succeeds and no classical polynomial-time algorithm can, unless a foundational assumption in complexity theory collapses. The task involves predicting how quantum many-body systems evolve over time under an unknown Hamiltonian — essentially forecasting the future state of a system of interacting quantum particles. The quantum learner trains on short-time samples, reconstructs the underlying Hamiltonian, then uses quantum simulation combined with a technique called classical shadows to make predictions. The classical hardness result is anchored in a BQP-complete computation embedded in a variant of the Feynman-Kitaev clock Hamiltonian, meaning any classical shortcut would imply BQP collapses into P\u002Fpoly — a result most complexity theorists consider extremely unlikely.\n\nClaims of quantum advantage are common; provable, complexity-theoretic separations are rare. Most quantum speedup demonstrations are heuristic or empirical, leaving room for a clever classical algorithm to close the gap later. This result uses PAC-learning theory — the same formal framework that underpins classical machine learning guarantees — which means the separation lives on the same rigorous ground as classical learning complexity results.\n\nThe catch is the usual one: the separation is proven for a specific, somewhat contrived family of input distributions, not general physics problems. It is a proof of principle, not a blueprint for a near-term quantum advantage experiment — but in a field full of marketing, a proof is worth something.","[\"quantum computing\",\"machine learning\",\"complexity theory\",\"research\"]","2026-07-08T04:00:00.000Z","2026-07-08T08:09:49.470Z","2026-07-08T08:09:52.368Z",[],[156,157,158,139],"quantum computing","machine learning","complexity theory",[160],{"name":161,"url":162},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.06472",{"id":164,"slug":165,"title":166,"dek":167,"body_md":168,"tags_json":169,"published_at":151,"created_at":170,"updated_at":171,"status":91,"review_note":92,"review_notes":172,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":173,"sources":177,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4375,"ai-drones-map-salt-marsh-plants-with-99-accuracy","AI Drones Map Salt Marsh Plants With 99% Accuracy","A new UAV pipeline called EcoVision uses transformer models to classify salt marsh grasses at 2x2m resolution, matching traditional field surveys within 8%.","A research team has built an AI-powered drone system that maps salt marsh vegetation with enough precision to rival scientists walking the ground.\n\nEcoVision processes low-altitude drone footage through a four-stage pipeline: semantic segmentation via a transformer model, vegetation extraction, species classification using a ConvNeXt architecture, and a grid-based dominance scoring system at 2x2m resolution. The system targets two salt-tolerant grasses, Spartina maritima and Puccinellia maritima, and was trained on manually annotated UAV imagery supplemented with public biodiversity datasets. Object-level classification hit an F1 score of 0.99, and dominance estimates differed from traditional quadrat field surveys by less than 8% on average.\n\nSalt marshes are among the most carbon-dense ecosystems on the planet, and tracking which species dominate where matters for both conservation policy and climate accounting. Conventional field surveys are slow, expensive, and hard to repeat at scale — a drone pipeline that can match their accuracy opens the door to continuous, wide-area monitoring without proportional cost increases.\n\nThe segmentation mean IoU of 0.56 is the one number worth watching: it reflects how well the model delineates individual plant patches at the pixel level, and there is room to improve before this becomes a drop-in replacement for field ecologists rather than a useful complement to them.","[\"ai\",\"remote-sensing\",\"ecology\",\"drones\"]","2026-07-08T07:33:47.875Z","2026-07-08T07:33:50.834Z",[],[19,174,175,176],"remote-sensing","ecology","drones",[178],{"name":161,"url":179},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.06105",{"id":181,"slug":182,"title":183,"dek":184,"body_md":185,"tags_json":186,"published_at":151,"created_at":187,"updated_at":188,"status":91,"review_note":92,"review_notes":189,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":192,"sources":195,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4300,"neural-networks-learn-to-simulate-shockwave-physics","Neural Networks Learn to Simulate Shockwave Physics","Researchers used physics-informed neural networks to model wave behavior across steel-aluminum boundaries, cutting reliance on costly finite-element runs.","A team of researchers has built a neural-network framework that predicts how stress waves travel through two-material systems — without running a full finite-element simulation every time.\n\nThe paper, posted to arXiv (2607.06479) by authors working with a steel-aluminum Split Hopkinson Pressure Bar setup, trains a physics-informed neural network (PINN) by baking the governing equations of elasticity directly into the loss function. High-fidelity ANSYS simulations provided validation data and additional training constraints. The resulting model accurately reproduces axial and radial displacement histories, stress and strain evolution, and wave transmission and reflection at the material interface.\n\nThe payoff is a surrogate model: once trained, it can predict wave responses at new time steps or for different material combinations without commissioning fresh finite-element runs — which are computationally expensive and slow. For engineers doing high-rate impact testing or designing layered structures, that is a meaningful time and cost reduction.\n\nPINNs have been gaining traction across computational mechanics for roughly five years, but most published work targets simpler single-material or steady-state problems. Extending the approach to transient, multi-material wave dynamics is a harder problem, and the results here appear to hold up under mesh-sensitivity checks. Whether the training cost amortizes well at scale — and against faster traditional solvers — is the question the paper leaves open.","[\"machine learning\",\"physics simulation\",\"materials science\",\"research\"]","2026-07-08T05:09:50.586Z","2026-07-08T05:09:53.425Z",[190],{"id":95,"reviewer":96,"round":97,"reason":191,"status":99},"The article omits author attribution and source identification (arXiv paper ID, authors, institution) required for the piece to be independently verifiable.",[157,193,194,139],"physics simulation","materials science",[196],{"name":161,"url":197},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.06479",{"id":199,"slug":200,"title":201,"dek":202,"body_md":203,"tags_json":204,"published_at":56,"created_at":205,"updated_at":206,"status":91,"review_note":92,"review_notes":207,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":208,"sources":211,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4150,"chaos-makes-ai-better-at-finding-physics-equations","Chaos Makes AI Better at Finding Physics Equations","A new paper argues that chaotic systems are actually easier for AI to reverse-engineer from data — and stable, predictable ones may be nearly impossible.","The systems hardest to predict turn out to be the easiest for AI to learn from scratch.\n\nResearchers publishing on arXiv have worked out a formal answer to a question that has nagged data-driven science for years: when can an AI actually recover a system's governing equations from observations alone? The answer hinges on chaos. Systems that exhibit chaotic behavior across their entire domain can, in principle, be uniquely identified from a single observed trajectory. The classical Lorenz system — a textbook example of chaos — is proven here to be analytically discoverable for the first time. Systems that are merely chaotic on a strange attractor can also be recovered, provided a specific geometric condition holds.\n\nThe flip side is the harder news. Stable, predictable systems — the kind engineers rely on for digital twins, robotics, and structural modeling — are often not discoverable from trajectory data alone. If a system has conserved quantities called first integrals, unique identification from data is mathematically impossible without additional prior knowledge baked in. That quietly undermines a lot of the confidence placed in purely data-driven modeling for engineering applications.\n\nThe finding reframes why weather forecasting has been such a productive target for machine learning: the atmosphere is chaotic, which turns out to be a feature for discoverability, not just a bug for prediction. For the tidy, well-behaved systems engineers actually want to simulate, more data is not the answer — the approach itself needs to change.","[\"ai\",\"science\",\"machine-learning\",\"data-driven-modeling\"]","2026-07-07T18:20:28.250Z","2026-07-07T18:20:31.248Z",[],[19,13,209,210],"machine-learning","data-driven-modeling",[212],{"name":161,"url":213},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.08860",{"id":215,"slug":216,"title":217,"dek":218,"body_md":219,"tags_json":220,"published_at":56,"created_at":221,"updated_at":222,"status":91,"review_note":92,"review_notes":223,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":224,"sources":228,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4105,"a-gnn-that-predicts-what-sticks-to-drug-delivery-nanoparticles","A GNN That Predicts What Sticks to Drug-Delivery Nanoparticles","GenShin uses a graph neural network to rank proteins that coat lipid nanoparticles, potentially replacing costly mass spectrometry screening.","A new machine learning model wants to cut one of the more expensive steps out of nanoparticle drug delivery research.\n\nWhen lipid nanoparticles enter the bloodstream, plasma proteins immediately coat their surface — a layer called the protein corona. That coating determines where the particle ends up in the body, which makes predicting its composition essential for designing tissue-specific therapies. Today, figuring out that composition means running mass spectrometry on physically prepared liposome samples: slow, expensive, and impossible to do at scale before you have even synthesized a candidate. GenShin, a graph neural network described in a new preprint, proposes a shortcut. Instead of simulating the full binding geometry between lipids and proteins, it scores lipid-protein pairs and ranks the results — a signal the researchers argue correlates well enough with real corona composition to guide pre-synthesis screening.\n\nThe practical upside is scale. Conventional methods gate screening behind wet-lab work, meaning researchers can only evaluate candidates they can physically make. A model that scores pairs computationally — and does so without needing precise docking poses — opens the door to searching far larger lipid spaces before any synthesis happens. On standard benchmarks, GenShin held up against pose-dependent models even when those models were fed unreliable pose data, which is the common case in early-stage screening.\n\nGraph neural networks have been creeping into every corner of drug discovery for years, but most of the attention has gone to small-molecule binding. Applying the same architecture to the messier problem of nanoparticle surface chemistry is a narrower, harder bet — and a preprint is still a long way from a validated screening pipeline.","[\"machine learning\",\"drug delivery\",\"nanoparticles\",\"bioinformatics\"]","2026-07-07T17:22:22.571Z","2026-07-07T17:22:25.510Z",[],[157,225,226,227],"drug delivery","nanoparticles","bioinformatics",[229],{"name":161,"url":230},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2504.13853",{"id":232,"slug":233,"title":234,"dek":235,"body_md":236,"tags_json":237,"published_at":56,"created_at":238,"updated_at":239,"status":91,"review_note":92,"review_notes":240,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":241,"sources":244,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},4023,"lidar-biomass-estimates-get-more-accurate-with-continuous-tree-maps","LiDAR Biomass Estimates Get More Accurate With Continuous Tree Maps","A new training method for forest biomass models cuts estimation error by swapping out coarse plot averages for spatially continuous reference data.","Researchers found a way to make LiDAR-based forest carbon measurements meaningfully more accurate - and the fix is in how training data is structured.\n\nConventional above-ground biomass models learn from plot-level inventory aggregates: a single number representing all the trees inside a fixed boundary. A new paper on arXiv proposes replacing those with Horizontal Biomass Distributions (HBDs), continuous spatial maps derived from Quantitative Structure Models (QSMs) that reconstruct individual tree geometry in 3D. The team trained a sparse 3D U-Net on simulated broadleaved forest data using three reference types - standard inventory aggregates, edge-effect-free QSM aggregates, and the new continuous HBD maps - then tested them across plot sizes ranging from 100 to 2500 square meters.\n\nThe gap matters most at small scales. On 100-square-meter plots, the HBD approach cut relative root mean square error by 16.84 percent and pushed R-squared up by 0.22 compared to the standard inventory baseline. That is significant because small plots are exactly where forest carbon monitoring is weakest - and where boundary-edge artifacts distort measurements the most.\n\nForest biomass estimation is increasingly central to carbon credit markets and national emissions reporting, where a 16 percent accuracy improvement is not a footnote. The broader lesson is familiar from other deep learning domains: model architecture gets the headlines, but reference data quality often determines the ceiling.","[\"machine learning\",\"remote sensing\",\"climate\",\"research\"]","2026-07-07T15:12:31.712Z","2026-07-07T15:12:34.688Z",[],[157,242,243,139],"remote sensing","climate",[245],{"name":161,"url":246},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.05260",{"id":248,"slug":249,"title":250,"dek":251,"body_md":252,"tags_json":253,"published_at":56,"created_at":254,"updated_at":255,"status":91,"review_note":92,"review_notes":256,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":257,"sources":259,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3992,"new-benchmark-exposes-hidden-failures-in-quantum-circuit-design","New Benchmark Exposes Hidden Failures in Quantum Circuit Design","HamQASBench groups molecules by quantum structure rather than identity, catching failure modes that standard energy-accuracy metrics consistently miss.","A new diagnostic benchmark for quantum circuit design catches problems that standard tests ignore.\n\nResearchers introduced HamQASBench, a benchmark for Quantum Architecture Search - the process of automatically designing parameterized quantum circuits used in variational quantum algorithms. Existing benchmarks sort test cases by molecule type or qubit count, neither of which says much about the underlying physics. HamQASBench instead organizes 11 molecules into five structural tiers using fingerprints drawn from the Pauli operator basis, computational basis representation, and ground-state entanglement. On top of energy accuracy, it adds per-qubit entanglement analysis and pairwise state fidelity checks.\n\nThe difference in approach surfaces failure modes that conventional metrics never flag. Testing five QAS methods across four paradigms, the benchmark exposed over-parameterization on near-product ground states, circuit search space growth that kills scalability, and topology-induced routing failures - none of which show up when you only measure whether the algorithm lands on the right energy level. That last point matters most: a circuit can score well on energy accuracy while being structurally wrong for the problem it is supposed to solve.\n\nQuantum hardware benchmarking has long leaned on metrics borrowed from classical computing, and the gap keeps causing surprises when algorithms hit real devices. HamQASBench does not solve the noise problem or make quantum advantage arrive sooner, but it gives researchers a sharper diagnostic tool - one that asks whether a circuit is right for reasons beyond getting the number to match.","[\"quantum computing\",\"benchmarking\",\"machine learning\",\"research\"]","2026-07-07T14:17:15.089Z","2026-07-07T14:17:17.984Z",[],[156,258,157,139],"benchmarking",[260],{"name":161,"url":261},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.04845",{"id":263,"slug":264,"title":265,"dek":266,"body_md":267,"tags_json":268,"published_at":56,"created_at":269,"updated_at":270,"status":91,"review_note":92,"review_notes":271,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":272,"sources":275,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3949,"a-plankton-classifier-that-knows-when-to-doubt-itself","A Plankton Classifier That Knows When to Doubt Itself","Researchers built a verification framework for underwater AI that flags uncertain classifications before marine biologists waste time on bad data.","Underwater robots counting plankton are getting a formal quality filter for their AI brains.\n\nResearchers have proposed a robustness verification framework for plankton classifiers running on autonomous underwater vehicles. The system uses reachability analysis to formally prove whether a neural network's output can be trusted under noisy, real-world ocean conditions. Alongside it, they introduced a neural ordinary differential equation model — a continuous-time alternative to standard convolutional neural networks — trained on imagery from a hardware imager called the SilCam. The core claim: the framework acts as an automated pre-filter, catching ambiguous classifications before they hit a marine biologist's desk.\n\nThis matters because the bottleneck in ocean monitoring is not data collection — AUVs can run for days — it is the post-processing pile-up when AI systems misfire on murky water, debris, or unusual lighting. A classifier that ships formal stability guarantees against environmental perturbations could meaningfully reduce that manual review burden. Most deployed ocean AI systems offer no such guarantees at all.\n\nStandard neural network classifiers routinely pass confident-looking wrong answers downstream, and the ocean is an especially unforgiving environment for that failure mode. Formal verification is common in aerospace and automotive safety systems; applying it to marine biology tooling is genuinely useful, even if the paper's lab results will need field validation before any AUV operator leans on it.","[\"ai\",\"ocean-science\",\"autonomous-vehicles\",\"machine-learning\"]","2026-07-07T12:58:42.088Z","2026-07-07T12:58:45.051Z",[],[19,273,274,209],"ocean-science","autonomous-vehicles",[276],{"name":161,"url":277},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.04453",{"id":279,"slug":280,"title":281,"dek":282,"body_md":283,"tags_json":284,"published_at":56,"created_at":285,"updated_at":286,"status":91,"review_note":92,"review_notes":287,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":288,"sources":291,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3927,"smarter-fusion-reactor-safety-cheaper-at-runtime","Smarter Fusion Reactor Safety, Cheaper at Runtime","A new AI framework trains on camera images and sensor data but drops the cameras at inference, cutting costs without losing accuracy.","A research team has found a way to make plasma disruption prediction faster and cheaper without sacrificing the accuracy that visual data provides.\n\nPlasma disruptions — sudden losses of plasma containment — are among the most serious safety risks in tokamak fusion reactors. Existing models lean on time-series diagnostic signals, but visible-light camera images add useful spatial information: plasma deformation, localized brightening, radiation patterns. The problem is that processing images at inference time is expensive. The new framework, tested on 640 discharges from China's EAST tokamak, sidesteps that tradeoff by training a multimodal \"teacher\" model on both images and sensor signals, then distilling that knowledge into a leaner \"student\" model that runs on sensor data alone.\n\nThe distillation happens at three levels — graph structure, internal representations, and final decisions — so the student model inherits the teacher's spatial intuition without needing a camera feed at runtime. That matters because fusion reactors need real-time disruption warnings; a model that is too slow or compute-hungry to deploy continuously is not useful in practice.\n\nFusion energy has attracted serious capital in recent years, and as commercial reactors move closer to reality, the gap between research-grade AI and production-ready safety systems becomes harder to ignore. A distillation approach that preserves accuracy while slashing inference cost is a step toward closing it — though 640 discharges is a modest dataset, and real-world generalization across reactor designs remains an open question.","[\"fusion energy\",\"machine learning\",\"safety\",\"hardware\"]","2026-07-07T12:21:01.440Z","2026-07-07T12:21:04.410Z",[],[289,157,290,39],"fusion energy","safety",[292],{"name":161,"url":293},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.04241",{"id":295,"slug":296,"title":297,"dek":298,"body_md":299,"tags_json":300,"published_at":56,"created_at":301,"updated_at":302,"status":91,"review_note":92,"review_notes":303,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":304,"sources":309,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3894,"a-formal-verification-tool-for-python-stats-code","A Formal Verification Tool for Python Stats Code","Why3-py translates Python statistical programs into a format that can be mathematically checked for hidden assumptions and misuse.","A new open-source tool wants to catch the statistical errors that fuel science's reproducibility crisis before they make it into published research.\n\nResearchers have released Why3-py, a Python front-end for the Why3 formal verification platform. The tool converts Python statistical programs into WhyML, a language designed for mathematical proof checking, so that automated verifiers can inspect the code for violated assumptions and incorrect usage. The project also extends an existing tool called StatWhy to cover meta-analysis methods — the aggregated studies that carry outsized weight in research communities. The core problem it targets is unglamorous but real: statistical methods come loaded with implicit assumptions that many practitioners never check, and Python's dynamic typing makes those assumptions invisible to standard code analysis.\n\nThe reproducibility crisis has been documented across psychology, medicine, and social science for over a decade, and bad statistics remain a leading culprit. Formal verification is a well-established technique in software safety engineering, but applying it to scientific computing is still uncommon enough that a tool bridging the two is genuinely useful — especially one that plugs into Python, the dominant language for data analysis.\n\nThe catch is adoption. Formal methods have a steep learning curve, and most researchers writing pandas pipelines are not fluent in proof theory. Whether Why3-py finds an audience beyond verification specialists will depend on how much the tooling can abstract away the underlying machinery — something the paper does not fully address.","[\"formal verification\",\"open-source\",\"reproducibility\",\"statistics\"]","2026-07-07T11:05:13.966Z","2026-07-07T11:05:16.877Z",[],[305,306,307,308],"formal verification","open-source","reproducibility","statistics",[310],{"name":161,"url":311},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.03951",{"id":313,"slug":314,"title":315,"dek":316,"body_md":317,"tags_json":318,"published_at":56,"created_at":319,"updated_at":320,"status":91,"review_note":92,"review_notes":321,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":322,"sources":324,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3866,"a-new-ai-model-maps-amazon-carbon-stocks-through-the-clouds","A New AI Model Maps Amazon Carbon Stocks Through the Clouds","A new transformer model fuses radar and satellite imagery to map Amazon carbon stocks, clearing the ESA's sub-20% error benchmark.","Researchers have built a deep learning model that fuses radar and satellite imagery to measure Amazon forest biomass — and it meets the European Space Agency's accuracy target.\n\nA team developed the Trimodal Coherent Co-attention Transformer (TCCT), combining optical surface reflectance data from Landsat-5 with Polarimetric SAR Interferometry (PolInSAR) data across P and L radar bands. The key engineering move is using complex-valued encoders that preserve the radar signal's phase information — something conventional fusion approaches discard. A co-attention mechanism automatically down-weights cloud-corrupted optical pixels and shifts reliance to radar when the view from space is blocked. In testing on the Paracou site in French Guiana, the model hit a relative biomass error of 4.51% in dense forest and a canopy height RMSE of 3.78 meters, outperforming Random Forest, CNN, and standard Vision Transformer baselines.\n\nAccurately measuring how much carbon sits in tropical forests is not an academic exercise. It feeds carbon offset markets, national emissions inventories, and the climate models governments rely on. Cloud cover and radar signal saturation have long made dense tropical regions one of the harder remote sensing targets; a model that handles both constraints at once could be practical infrastructure for the ESA's BIOMASS satellite mission, which sets a benchmark of under 20% error.\n\nThe results come from a single research site, so whether this calibration holds across the broader Amazon — let alone other tropical biomes — is the obvious next question.","[\"remote sensing\",\"climate\",\"ai\",\"carbon\"]","2026-07-07T10:24:28.248Z","2026-07-07T10:24:31.066Z",[],[242,243,19,323],"carbon",[325],{"name":161,"url":326},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.03663",{"id":328,"slug":329,"title":330,"dek":331,"body_md":332,"tags_json":333,"published_at":56,"created_at":334,"updated_at":335,"status":91,"review_note":92,"review_notes":336,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":339,"sources":342,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3858,"a-35-year-math-problem-falls-with-an-ais-help","A 35-Year Math Problem Falls, With an AI's Help","Researchers used ChatGPT to help crack a stochastic-process uniqueness problem open for more than 35 years, then verified every step themselves.","A math problem open for more than 35 years now has a proof, and a chatbot helped find it.\n\nResearchers posted a preprint resolving what's called the finite-signed uniqueness problem for the basic adjoint relationship (BAR) conjecture. The conjecture asks whether a specific mathematical relationship uniquely pins down the stationary distribution of a reflected diffusion, a model class used in queueing network analysis. The result applies within stable Harrison-Reiman data where the reflection matrix satisfies a nonsingular M-matrix condition, a setting the authors show is structurally necessary. The proof was discovered with ChatGPT's assistance, the paper states, and then independently verified by the authors.\n\nThe result matters because it closes both sides of a question posed by Dai and Dieker: uniqueness holds in the M-matrix class, but not in the broader completely-S class, where the paper exhibits an explicit four-parameter, three-dimensional counterexample. That double answer, a positive result bounded by a sharp negative one, tells researchers exactly where the BAR approach runs out of power, which is more useful than either result alone.\n\nWhether ChatGPT's role was genuine structural insight or a mechanical nudge toward a proof the authors were already close to, the paper doesn't say, and that question will matter more than any single theorem as mathematicians decide how much to trust this kind of collaboration.","[\"ai\",\"mathematics\",\"research\",\"queueing-theory\"]","2026-07-07T10:13:56.501Z","2026-07-07T10:13:59.323Z",[337],{"id":95,"reviewer":96,"round":97,"reason":338,"status":99},"The dek names 'ChatGPT 5.5 Pro' as the model used, but this version cannot be verified against any publicly documented OpenAI release lineup — reject and replace with a verifiable model name or remove the specific version until it can be confirmed.",[19,340,139,341],"mathematics","queueing-theory",[343],{"name":161,"url":344},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.03639",{"id":346,"slug":347,"title":348,"dek":349,"body_md":350,"tags_json":351,"published_at":56,"created_at":352,"updated_at":353,"status":91,"review_note":92,"review_notes":354,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":355,"sources":358,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3751,"quantum-circuit-design-beats-penalty-tricks-for-crispr-guide-selection","Quantum Circuit Design Beats Penalty Tricks for CRISPR Guide Selection","Researchers tested two quantum optimization strategies on a gene-editing problem, and the structural approach beat penalty tuning by a wide margin.","A quantum algorithm design choice matters more than hardware, at least for a small CRISPR guide RNA selection problem.\n\nResearchers at arXiv published COMET, a head-to-head comparison of two ways to handle constraints in the Quantum Approximate Optimization Algorithm (QAOA) - the dominant near-term quantum heuristic. The test case: picking one guide RNA per gene across three immune-checkpoint targets (PDCD1, LAG3, and HAVCR2), a combinatorial problem that maps cleanly onto a 12-qubit quantum circuit. The conventional approach adds penalty terms to steer the optimizer away from invalid solutions. COMET's alternative uses an XY-mixer, a circuit structure that makes illegal solutions physically unreachable.\n\nThe gap is stark. In simulation, the XY-mixer hit above 95% probability of finding the optimal answer by circuit depth p=3. The three penalty variants tested never broke 6% at any depth. On IBM's ibm_kingston processor, the XY-mixer's real-hardware results stayed close to simulation; the worst-tuned penalty variant drifted by +53.9 energy units. The authors are candid that gate-level noise does partially erode the structural guarantee - that honesty is worth noting in a field prone to overselling quantum results.\n\nThe 12-qubit problem is, by the authors' own admission, classically trivial - any laptop solves it instantly. The paper's value is methodological: it gives quantum-biology researchers a concrete, hardware-validated reason to prefer constraint-preserving circuit design over penalty tuning. As quantum hardware scales and biologically relevant problem sizes grow, that design choice could determine whether near-term quantum solvers stay competitive.","[\"quantum computing\",\"crispr\",\"optimization\",\"research\"]","2026-07-07T07:30:25.844Z","2026-07-07T07:30:28.783Z",[],[156,356,357,139],"crispr","optimization",[359],{"name":161,"url":360},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.02622",{"id":362,"slug":363,"title":364,"dek":365,"body_md":366,"tags_json":367,"published_at":56,"created_at":368,"updated_at":369,"status":91,"review_note":92,"review_notes":370,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":371,"sources":374,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3742,"open-source-tool-turns-heart-scans-into-simulation-meshes","Open-Source Tool Turns Heart Scans into Simulation Meshes","A new automated pipeline converts CT segmentations into physics-ready cardiac models, enabling large virtual patient cohorts for in silico trials.","An open-source framework can now take a raw heart scan and output a simulation-ready 3D mesh in minutes.\n\nResearchers published a semi-automatic pipeline that converts CT-based cardiac segmentations into geometries suitable for multiphysics simulations. The tool addresses a persistent bottleneck: medical image segmentations often contain artifacts, gaps, or topological defects that make them useless for computational modeling. The framework uses deep learning segmentation combined with a template-based registration stage — a high-quality mesh template is deformed toward each individual heart using a Chamfer-distance morphing strategy. Validated on 58 healthy cardiac CT scans, it produces watertight, isotopological meshes with consistent point-to-point correspondence across all chambers and proximal vessels.\n\nThe bigger payoff is what comes after the mesh. Because all outputs share the same topology, they can be projected into a unified shape space, and the team used Principal Component Analysis to show that a low-dimensional representation captures most of the population's anatomical variation. That makes it possible to generate synthetic anatomies via Gaussian Mixture Modeling — essentially a factory for realistic virtual patients.\n\nIn silico clinical trials — where a drug or device is tested on a simulated population before human subjects — have long been hampered by the small-cohort problem: most studies use one or a handful of anatomies, which tells you little about how results generalize. This pipeline is a plausible answer to that, though \"validated on 58 healthy scans\" is a narrow base; disease-altered geometries and larger, more diverse datasets are the obvious next test.","[\"medical imaging\",\"computational biology\",\"open-source\",\"ai\"]","2026-07-07T07:19:18.673Z","2026-07-07T07:19:21.916Z",[],[372,373,306,19],"medical imaging","computational biology",[375],{"name":161,"url":376},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.02564",{"id":378,"slug":379,"title":380,"dek":381,"body_md":382,"tags_json":383,"published_at":384,"created_at":385,"updated_at":386,"status":91,"review_note":92,"review_notes":387,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":395,"sources":398,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3633,"singapores-cyborg-cockroaches-can-now-breathe-underwater","Singapore's Cyborg Cockroaches Can Now Breathe Underwater","A Nanyang Technological University team fitted remote-controlled cockroaches with 3D-printed oxygen suits for disaster rescue in flooded terrain.","Remote-controlled cockroaches with infrared cameras can now operate underwater for three hours, thanks to a custom 3D-printed oxygen suit developed in Singapore.\n\nResearchers at Nanyang Technological University, led by Hirotaka Sato, have been wiring cockroaches with IR cameras and wireless controls for disaster search-and-rescue work. After demonstrating a coordinated swarm in 2024, the team identified a gap: the insects couldn't cross water. The fix is a bespoke scuba suit — a 3D-printed shell with tubes feeding oxygen from a sponge-based tank into the cockroach's spiracles, using a slow hydrogen peroxide and manganese dioxide reaction to avoid heavy pressurized gear. In tests, the roaches held up at 50 cm depth for three hours, slowing only slightly from 8.75 cm per second on land to 7.84 cm per second submerged.\n\nThis matters because energy is the wall every micro-robotics team hits. Battery-powered miniature robots need recharging; a cockroach needs a meal every few weeks and self-heals wounds. The biological chassis solves power and durability problems that silicon and lithium can't, at least not yet at this scale.\n\nTwo decades after DARPA's HI-MEMS program first proved the concept of cyborg insects, the gap between proof-of-concept and operational tool is narrowing. Sato's team is already eyeing harsher environments — including Mars — which reads partly as a grant pitch but is less absurd than it sounds: cockroaches tolerate radiation, low oxygen, and high CO2 at levels that would kill a human. The more immediate question is whether regulators and rescue agencies will ever sign off on deploying living insects as surveillance hardware.","[\"robotics\",\"biotech\",\"disaster-response\",\"science\"]","2026-07-03T09:22:28.000Z","2026-07-03T10:14:09.182Z","2026-07-03T10:14:12.000Z",[388,390],{"id":95,"reviewer":96,"round":97,"reason":389,"status":99},"The headline and dek use an informal, placeholder-style construction ('Cyborg Cockroaches Now Swim - With IR Cameras') that reads as a working draft title rather than a finished, publication-ready headline — rewrite both to state the actual news in a complete, publication-ready sentence.",{"id":391,"reviewer":392,"round":393,"reason":394,"status":99},"publisher-r2","publisher",2,"\"DAPA\" in the final paragraph is a typo for \"DARPA\" (Defense Advanced Research Projects Agency), a factual name error that must be corrected before publication.",[101,396,397,13],"biotech","disaster-response",[399],{"name":400,"url":401},"Tom's Hardware","https:\u002F\u002Fwww.tomshardware.com\u002Ftech-industry\u002Frobotics\u002Fscientists-have-created-a-3d-printed-remote-controlled-cyborg-cockroach-equipped-with-ir-cameras-living-insects-fitted-with-flexible-diving-suit-can-survive-and-move-underwater-for-three-hours",{"id":403,"slug":404,"title":405,"dek":406,"body_md":407,"tags_json":408,"published_at":409,"created_at":410,"updated_at":411,"status":91,"review_note":92,"review_notes":412,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":413,"sources":417,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3470,"ai-cuts-the-cost-of-hunting-neutron-star-mergers","AI Cuts the Cost of Hunting Neutron Star Mergers","A neural network called Aframe matched the sensitivity of classical gravitational-wave search pipelines while running on a single consumer-grade GPU.","A neural network already used to find black hole mergers has now proven it can hunt binary neutron star collisions just as accurately as the conventional method — at a fraction of the computing cost.\n\nResearchers adapted Aframe, an AI search algorithm deployed during LIGO-Virgo-KAGRA's fourth observing run, to detect signals from merging neutron stars. The original pipeline for that task requires matching incoming detector data against roughly one million reference waveforms, a job that can demand up to a thousand CPU cores running in real time. Aframe sidesteps that by training a neural network to recognize the presence of a signal directly in the data. To handle the longer signal duration that neutron star mergers produce compared to black hole pairs, the team applied a preprocessing step called heterodyning before feeding data to the network. The result: sensitivity on par with matched-filter pipelines, on a single non-flagship GPU.\n\nThe stakes for binary neutron star detection are unusually high. When two neutron stars merge, they emit not just gravitational waves but light and neutrinos simultaneously — the kind of multi-messenger event that lets physicists probe nuclear matter, test general relativity, and measure the expansion rate of the universe all at once. The 2017 detection of GW170817 remains the only confirmed example, and the entire field has been waiting for a second one. Cutting latency and compute costs makes real-time alerts more feasible, which is when multi-messenger follow-up is most valuable.\n\nAframe was already the first AI pipeline to detect multiple binary black hole mergers live; this extension to the lower-mass neutron star regime is a meaningful step, though a detected event — not a benchmark — will be the real proof of concept.","[\"gravitational waves\",\"ai\",\"astrophysics\",\"neutron stars\"]","2026-07-03T04:00:00.000Z","2026-07-03T06:47:12.406Z","2026-07-03T06:47:15.382Z",[],[414,19,415,416],"gravitational waves","astrophysics","neutron stars",[418],{"name":161,"url":419},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.01372",{"id":421,"slug":422,"title":423,"dek":424,"body_md":425,"tags_json":426,"published_at":427,"created_at":428,"updated_at":429,"status":91,"review_note":92,"review_notes":430,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":431,"sources":433,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3371,"satellite-ai-embeddings-beat-expert-attributes-for-river-forecasting","Satellite AI Embeddings Beat Expert Attributes for River Forecasting","A new study finds that AI-learned satellite image embeddings outperform hand-crafted basin attributes when predicting river flow in ungauged regions.","AI-generated satellite embeddings may give hydrologists a better tool for predicting river flow in places where no measurement stations exist.\n\nResearchers tested AlphaEarth Foundation embeddings — representations learned from large satellite image datasets rather than designed by domain experts — against traditional basin attributes like soil type, terrain, and land cover. When used to train hydrological models on known basins and then predict flow in unmonitored ones, the embedding-based models achieved higher accuracy. The embeddings capture vegetation patterns, land surface properties, and long-term environmental dynamics in a single learned representation, compressing complexity that hand-crafted attributes routinely miss.\n\nThe ungauged basin problem is not new, but it is consequential: a significant share of the world's river basins lack flow records, making flood forecasting and water management harder in exactly the places that can least afford bad predictions. The finding that a foundation model trained on satellite imagery can substitute for — and outperform — decades of expert-designed hydrological descriptors is a meaningful shift in how that gap might be closed.\n\nThe study also found that basin selection matters: using the embeddings to identify environmentally similar donor basins improved predictions, while lumping in dissimilar basins degraded them — a reminder that better representations do not eliminate the need for careful experimental design.","[\"ai\",\"climate\",\"hydrology\",\"remote-sensing\"]","2026-07-02T04:00:00.000Z","2026-07-02T08:30:24.332Z","2026-07-02T08:30:27.228Z",[],[19,243,432,174],"hydrology",[434],{"name":161,"url":435},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.01558",{"id":437,"slug":438,"title":439,"dek":440,"body_md":441,"tags_json":442,"published_at":443,"created_at":444,"updated_at":445,"status":91,"review_note":92,"review_notes":446,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":447,"sources":451,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},3170,"a-smartphone-camera-as-a-chronic-pain-lab","A Smartphone Camera as a Chronic Pain Lab","Researchers validated a computer vision pipeline that extracts 3D movement data from a single phone video, rivaling expensive lab equipment.","Your phone camera may soon double as a clinical movement lab.\n\nA research team developed Quantitative Movement Testing (QMT), a computer vision pipeline that pulls 3D kinematic data from standard monocular smartphone video. They validated it against optical motion capture — the gold-standard, lab-only system that costs far more to operate — using 13 healthy controls, then deployed it in two clinical cohorts: fibromyalgia patients in a pre- and post-intervention trial, and chronic sciatica patients tracked at home over 30 days. In lab conditions, QMT hit strong correlations (r > 0.85) against optical capture and showed high test-retest reliability (r > 0.86) in fibromyalgia patients. It also detected group-level movement differences between sciatica patients and healthy controls using only remote home recordings.\n\nChronic pain is notoriously hard to measure objectively. Patients self-report, clinicians observe, and neither method scales well for ongoing monitoring or clinical trials. A validated tool that runs on hardware already in a patient's pocket removes both the lab bottleneck and the cost barrier — and produces quantitative biomarkers that don't rely on someone describing their pain on a scale of one to ten.\n\nThe caveat the researchers flag themselves: home environments introduced more measurement variance than the lab. That gap will need closing before QMT can anchor a regulatory submission or replace motion capture in high-stakes assessments — but as a low-cost screening and monitoring layer, it has a clearer path to adoption than most academic prototypes.","[\"health tech\",\"computer vision\",\"clinical research\",\"ai\"]","2026-07-01T04:00:00.000Z","2026-07-01T08:53:39.786Z","2026-07-01T08:53:42.765Z",[],[448,449,450,19],"health tech","computer vision","clinical research",[452],{"name":161,"url":453},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.02301",{"id":455,"slug":456,"title":457,"dek":458,"body_md":459,"tags_json":460,"published_at":461,"created_at":462,"updated_at":463,"status":91,"review_note":92,"review_notes":464,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":465,"sources":467,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},2897,"neural-net-decoder-pushes-quantum-error-correction-near-its-limits","Neural Net Decoder Pushes Quantum Error Correction Near Its Limits","A hybrid graph neural network and Transformer decoder brings quantum error correction thresholds within striking distance of theoretical maximums.","Researchers have built a neural network-assisted decoder that nearly matches the best possible performance for correcting errors in quantum computers.\n\nThe system, called Neural Minimum Weight Perfect Matching (NMWPM), pairs a graph neural network with a Transformer to improve a standard quantum error correction algorithm. The classic approach — Minimum Weight Perfect Matching — uses fixed edge weights on a graph to find the most probable error chains in a qubit array. NMWPM learns to predict those weights dynamically from the data instead, letting the underlying algorithm stay intact while the neural layer adapts to actual noise patterns. Testing on the toric code under depolarizing noise, the decoder reached thresholds of 17.9% and 10.95%, compared to theoretical maximums of 18.9% and 11.0%.\n\nGetting close to maximum likelihood bounds matters because error correction is the gating factor for practical quantum computing — without it, noise accumulates faster than any computation can outrun it. Most learned decoders sacrifice the structural guarantees of classical algorithms for flexibility; this hybrid approach keeps both, which is why the numbers land so close to the ceiling.\n\nOne open question is whether these thresholds hold at the scale and noise profiles of real hardware, where depolarizing noise is a convenient model but not the whole story.","[\"quantum computing\",\"machine learning\",\"error correction\",\"research\"]","2026-06-30T04:00:00.000Z","2026-06-30T14:44:06.174Z","2026-06-30T14:44:09.074Z",[],[156,157,466,139],"error correction",[468],{"name":161,"url":469},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.00242",{"id":471,"slug":472,"title":473,"dek":474,"body_md":475,"tags_json":476,"published_at":461,"created_at":477,"updated_at":478,"status":91,"review_note":92,"review_notes":479,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":480,"sources":483,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},2886,"physics-baked-neural-net-predicts-hydrogen-leaks-at-200-bar","Physics-Baked Neural Net Predicts Hydrogen Leaks at 200 Bar","A hybrid model embedding thermodynamic laws directly into its architecture outperforms pure deep learning for high-pressure green-hydrogen safety.","A neural network that hard-codes hydrogen transport physics into its structure can predict dangerous gas crossover far more reliably than data-only models — even at pressures well beyond its training data.\n\nResearchers built a physics-residual network, called PR-Net, for predicting hydrogen crossover in polymer electrolyte membrane water electrolyzers — the industrial systems that split water into hydrogen and oxygen under high pressure. Instead of asking a neural network to learn all the physics from scratch, PR-Net embeds three established laws (Henry's, Fick's, and Faraday's) as a fixed backbone, then trains a small residual layer to capture only the nonlinear effects those laws miss. Tested against 184 observations drawn from eight peer-reviewed studies across six membrane types, it reached an R² of 99.57% — with nine times lower prediction variability than either a standard neural network or a soft-constraint physics-informed neural network.\n\nThe extrapolation result is where things get interesting for industrial deployment. At 200 bar — two and a half times the maximum pressure in the training set — PR-Net held an R² of 94%, while the physics-informed neural network dropped to 68% and the plain neural network fell to 58%. That gap matters because real-world electrolyzer operators need to push systems beyond tested conditions, and hydrogen crossover is a direct safety hazard: too much gas migrating through the membrane can cause fires or explosions. The model also runs in about one millisecond on low-power embedded hardware, making real-time monitoring feasible without a server rack.\n\nThe hybrid approach — lock in what physics already knows, learn only what it doesn't — is a sensible counter to the current trend of throwing more data and larger models at every engineering problem. Green-hydrogen production is still looking for its cost curve; tools that reduce the experimental burden of safety validation could matter more than they look.","[\"green-hydrogen\",\"machine-learning\",\"energy\",\"safety\"]","2026-06-30T14:34:37.810Z","2026-06-30T14:34:40.704Z",[],[481,209,482,290],"green-hydrogen","energy",[484],{"name":161,"url":485},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.05879",{"id":487,"slug":488,"title":489,"dek":490,"body_md":491,"tags_json":492,"published_at":461,"created_at":493,"updated_at":494,"status":91,"review_note":92,"review_notes":495,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":496,"sources":498,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},2583,"a-new-ai-model-learns-to-read-the-night-sky","A New AI Model Learns to Read the Night Sky","Researchers built a light-curve model that handles astronomy's messy, irregular data better than hand-crafted features on 15 of 16 benchmarks.","A self-supervised AI framework trained on astronomical brightness data now outperforms traditional feature engineering across nearly every tested metric.\n\nAstronomers track how stars and other objects brighten or dim over time using what are called light curves. The problem: telescopes don't observe on a neat schedule, and the data arrives unevenly sampled, noisy, and spanning wildly different physical timescales — exactly the conditions that trip up most off-the-shelf time-series models. Researchers addressed this with a framework built on a Joint-Embedding Predictive Architecture, or JEPA, layering in uncertainty-aware tokenization and a training method called multi-view self-distillation. Tested on the StarEmbed classification benchmark, the model beat hand-crafted features on 15 of 16 metrics. In few-shot settings — the hardest test, where the model sees very little labeled data — it reached a macro-F1 score of 42.56 with just one labeled example per class, climbing to 63.58 with 100.\n\nThe broader claim here is that domain knowledge baked into the architecture matters more than scale alone. That's a pointed rebuke of the \"one model to rule them all\" instinct driving much of current AI research. The framework also transferred reasonably well to 12 unrelated irregular time-series datasets, matching or beating prior state-of-the-art on five of them — more wins than any single previous baseline managed.\n\nThe results are a reminder that foundation models built for general text or even general time-series data still stumble when physics gets in the way — and that a carefully constrained model often beats a larger, blunter one.","[\"ai\",\"science\",\"machine-learning\",\"astronomy\"]","2026-06-30T08:37:26.457Z","2026-06-30T08:37:29.361Z",[],[19,13,209,497],"astronomy",[499],{"name":161,"url":500},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.28446",{"id":502,"slug":503,"title":504,"dek":505,"body_md":506,"tags_json":507,"published_at":508,"created_at":509,"updated_at":510,"status":91,"review_note":92,"review_notes":511,"image_url":514,"persona_id":92,"persona_name":92,"section":13,"tags":515,"sources":520,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},2383,"rocket-lab-buys-iridium-for-8-billion","Rocket Lab Buys Iridium for $8 Billion","The launch startup acquires Iridium's 80-satellite network in a cash-and-stock deal Beck calls one of the most transformative in the space industry.","Rocket Lab is buying Iridium Communications in an $8 billion cash-and-stock deal.\n\nRocket Lab, the launch company founded and led by Peter Beck, announced Monday it will acquire Iridium, a profitable satellite communications firm with a network of 80 low-Earth orbit satellites. The deal combines a launch provider with an established telecom operator whose constellation has been delivering connectivity services for decades. Rocket Lab is paying in a mix of cash and its own stock.\n\nFor Rocket Lab, this is a leap from selling rides to orbit to owning the infrastructure in it. Most launch companies stay in their lane; buying a working, revenue-generating satellite network gives Rocket Lab a recurring revenue stream that launch contracts alone rarely provide. Beck called it \"one of the most transformative deals in the space industry\" and \"the ultimate combination for growth\" — language that sounds like marketing but reflects a real strategic shift.\n\nSpaceX has spent years blurring the line between launch provider and satellite operator through Starlink. Rocket Lab just made its own move toward that same model, except it skipped the years of building and bought a constellation that already works.","[\"space\",\"acquisitions\",\"satellites\",\"rocket-lab\"]","2026-06-29T15:33:54.000Z","2026-06-29T16:53:08.741Z","2026-06-29T16:53:16.056Z",[512],{"id":95,"reviewer":96,"round":97,"reason":513,"status":99},"The article misquotes Beck — the source says 'one of the most transformative deals in the space industry' is the full lead quote, but the article only uses 'the ultimate combination for growth' and attributes it to a 'promotional video,' which is fine, but the article also makes an unsupported claim that Iridium's satellites are '15-year-old technology'; no such characterization appears in the source material and the satellite constellation's age is not established in the provided source.","https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Frocket-lab-buys-iridium-for-8-billion.webp",[516,517,518,519],"space","acquisitions","satellites","rocket-lab",[521],{"name":124,"url":522},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F06\u002Fin-a-bold-move-rocket-lab-acquires-iridium-communications\u002F",{"id":524,"slug":525,"title":526,"dek":527,"body_md":528,"tags_json":529,"published_at":530,"created_at":531,"updated_at":532,"status":91,"review_note":92,"review_notes":533,"image_url":536,"persona_id":92,"persona_name":92,"section":13,"tags":537,"sources":540,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},2367,"eth-zurich-builds-a-pixel-that-both-emits-and-captures-light","ETH Zurich Builds a Pixel That Both Emits and Captures Light","Swiss researchers have built a pixel that can emit and capture light, potentially letting screens double as cameras without separate hardware.","A team at ETH Zurich has built a pixel that both emits and absorbs light, doing the job of a screen and a camera in a single component.\n\nResearchers at ETH Zurich published findings in *Nature* describing what they call the first bidirectional pixel. Display pixels and image-sensor pixels have always been separate: one converts electricity to light, the other converts light to electricity. This device does both within the same structure. The paper carries no announced commercial partner or production timeline.\n\nIf the approach scales, it could eliminate the front-facing camera cutouts that phone makers have spent years shrinking or hiding beneath displays. In principle, every pixel on a future screen could double as a sensor, capturing images, depth data, or biometric information without any dedicated camera hardware alongside it.\n\nA long road runs from a Nature paper to a factory floor, so treat this as a proof of concept, not a product announcement.","[\"display technology\",\"hardware\",\"research\",\"imaging\"]","2026-06-29T09:41:55.000Z","2026-06-29T10:56:47.250Z","2026-06-29T10:56:53.598Z",[534],{"id":95,"reviewer":96,"round":97,"reason":535,"status":99},"The body contains an editorial note in reader-facing copy ('the source material does not detail the underlying materials or fabrication method') — remove it or rephrase as a plain statement of what is known, without surfacing the editorial process to readers.","https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Feth-zurich-builds-a-pixel-that-both-emits-and-captures-light.webp",[538,39,139,539],"display technology","imaging",[541],{"name":106,"url":542},"https:\u002F\u002Fthenextweb.com\u002Fnews\u002Feth-zurichs-bidirectional-pixel-could-turn-screens-into-cameras",{"id":544,"slug":545,"title":546,"dek":547,"body_md":548,"tags_json":549,"published_at":550,"created_at":551,"updated_at":552,"status":91,"review_note":92,"review_notes":553,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":554,"sources":557,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},2201,"nasa-report-backs-cuts-to-lunar-programs-years-behind-schedule","NASA Report Backs Cuts to Lunar Programs Years Behind Schedule","An internal analysis supports NASA's cancellation of two Artemis-era programs, finding they were massively over budget and still nowhere near done.","NASA's own numbers back up the decision to kill two major Artemis components — and they are not flattering.\n\nAdministrator Jared Isaacman canceled development of a new upper stage for the Space Launch System rocket and shelved plans for a lunar-orbiting space station, pivoting instead toward a surface base — a shift he called \"Ignition.\" Contractors pushed back, arguing NASA was abandoning nearly complete hardware essential to landing humans on the Moon. A subsequent analysis undercut that framing: the programs were running years late and had burned through far more money than originally budgeted, with no clear finish line in sight. Isaacman's position was that neither program was actually required to get boots on the lunar surface.\n\nThe significance here is less about any single cancellation and more about what it reveals: NASA's cost and schedule controls on Artemis have been broken for years, and the hardware that contractors called \"nearly complete\" apparently was not. When the agency's own analysis agrees with the administrator who ordered the cuts, the grumbling from affected contractors looks less like principled dissent and more like self-interest.\n\nFor context, SLS itself has a long history of overruns — the rocket cost roughly $23 billion to develop before its first flight in 2022. Adding a stage adapter that allegedly needed 13 years and $500 million might have been the clearest sign yet that the program's incentive structures reward slow delivery over results.","[\"nasa\",\"space\",\"artemis\",\"policy\"]","2026-06-24T21:41:03.000Z","2026-06-24T23:43:45.594Z","2026-06-27T15:30:12.321Z",[],[555,516,556,34],"nasa","artemis",[558],{"name":124,"url":559},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F06\u002Fanalysis-finds-the-exploration-programs-nasa-recently-canceled-were-running-way-late\u002F",{"id":561,"slug":562,"title":563,"dek":564,"body_md":565,"tags_json":566,"published_at":567,"created_at":568,"updated_at":569,"status":91,"review_note":92,"review_notes":570,"image_url":571,"persona_id":92,"persona_name":92,"section":13,"tags":572,"sources":576,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},1877,"spacex-tests-starfall-a-saucer-shaped-orbital-cargo-pod","SpaceX Tests Starfall, a Saucer-Shaped Orbital Cargo Pod","SpaceX is launching a secretive reentry vehicle called Starfall this week to demonstrate point-to-point cargo delivery from low-Earth orbit.","SpaceX is about to test a saucer-shaped pod designed to drop cargo anywhere on Earth after a trip through orbit.\n\nThe vehicle, called Starfall, has been developed under wraps and is set to fly Tuesday aboard a Falcon 9 lifting off from Cape Canaveral Space Force Station in Florida. After two orbits, the upper stage will release the pod for reentry, with a parachute-assisted splashdown targeted roughly 800 miles west of California in the Pacific Ocean. The FAA published an environmental assessment last month describing Starfall's purpose as the \"transport and delivery of goods through space\" — which is as much official detail as SpaceX has offered publicly.\n\nPoint-to-point delivery from orbit sounds like a logistics pitch, but the real audience is almost certainly the Pentagon. The U.S. military has long funded research into using rockets for rapid global cargo delivery, and SpaceX already holds substantial defense contracts. A reentry pod that can place a payload anywhere on the planet within two orbits — roughly 90 minutes each — would shrink delivery windows that currently take days or weeks.\n\nSpaceX has spent years normalizing reusable rocketry; now it is quietly testing whether that same orbital infrastructure can move physical goods, not just satellites. Whether Starfall becomes a niche military logistics tool or something broader depends on how well it can handle the hard part: landing the cargo intact and on target, every time.","[\"spacex\",\"space\",\"logistics\",\"defense\"]","2026-06-23T05:25:16.000Z","2026-06-23T07:34:10.491Z","2026-06-23T07:34:19.598Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fspacex-tests-starfall-a-saucer-shaped-orbital-cargo-pod.webp",[573,516,574,575],"spacex","logistics","defense",[577],{"name":124,"url":578},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F06\u002Fwith-starfall-spacex-eyes-an-edge-in-global-cargo-delivery-from-orbit\u002F",{"id":580,"slug":581,"title":582,"dek":583,"body_md":584,"tags_json":585,"published_at":586,"created_at":587,"updated_at":588,"status":91,"review_note":92,"review_notes":589,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":592,"sources":595,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},1605,"ai-diffusion-model-sharpens-blurry-global-climate-forecasts","AI diffusion model sharpens blurry global climate forecasts","A new diffusion model turns coarse global climate output into 0.25-degree regional fields, with built-in uncertainty and far less compute.","A new AI tool promises to turn coarse global climate output into sharp regional maps without the usual supercomputer bill.\n\nThe tool, IPSL-AID, is built on a denoising diffusion probabilistic model, the same class of generative AI behind image generators. It is trained on ERA5 reanalysis data and takes the blurry output of conventional global climate models, which typically resolve features only down to 150 to 200 kilometers, and rebuilds them at 0.25-degree resolution, or roughly 25 kilometers. It generates temperature, wind, and precipitation fields from those coarse inputs plus their spatiotemporal context. Rather than committing to one answer, it models probability distributions of fine-scale features, so it can produce several plausible scenarios and attach uncertainty to each.\n\nRegional detail is where climate adaptation actually happens. A city sizing seawalls or a grid operator planning for heat waves needs kilometers, not hundreds of kilometers. The established way to get there is dynamical downscaling, nesting a physics-based regional model inside a global one, which is accurate but slow and computationally expensive. A diffusion model that approximates the same fine structure for a fraction of the compute could make high-resolution projections cheap enough to run many times over, which is the whole point if you want an ensemble rather than a single guess.\n\nThe authors report that the model reconstructs statistical distributions well, including extreme events, power spectra, and spatial structures. Those are the hard parts: extremes and fine texture are exactly what naive smoothing erases.\n\nStill, this is a preprint reconstructing known reanalysis data, not a forecast of an unseen future climate. Reproducing ERA5's statistics is a necessary first step, not proof the method holds when the climate moves outside the range it was trained on.","[\"climate\",\"diffusion models\",\"machine learning\",\"downscaling\"]","2026-06-18T04:00:00.000Z","2026-06-19T05:32:05.144Z","2026-06-19T14:20:56.795Z",[590],{"id":95,"reviewer":96,"round":97,"reason":591,"status":99},"Expand the draft past the 300-word minimum (it currently runs ~210 words) by adding concrete context and analysis—how dynamical downscaling compares in cost and speed, who actually uses 0.25-degree regional projections, and what limitations or validation the authors report—without inventing facts beyond the source.",[243,593,157,594],"diffusion models","downscaling",[596],{"name":161,"url":597},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2604.03275",{"id":599,"slug":600,"title":601,"dek":602,"body_md":603,"tags_json":604,"published_at":605,"created_at":606,"updated_at":607,"status":91,"review_note":92,"review_notes":608,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":609,"sources":614,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},1614,"amazon-and-quera-claim-quantum-error-correction-by-2028","Amazon and QuEra Claim Quantum Error Correction by 2028","Amazon and QuEra say useful error-corrected quantum computers could arrive by 2028, years ahead of where most of the field puts the milestone.","Amazon and QuEra are promising useful, error-corrected quantum computing by 2028, a timeline most experts would call aggressive.\n\nThe two companies set a joint target of 2028 for practical quantum error correction. The milestone requires building logical qubits: bundles of physical hardware qubits that store information redundantly and monitor each other for errors. Most researchers currently place that milestone five to ten years out. The same wave of announcements also included an updated trapped ion processor and a revision of some earlier quantum supremacy claims, after classical algorithms gained ground.\n\nError correction is the gating problem for nearly everything interesting that quantum computers could theoretically do. Current hardware is too error-prone to run the algorithms where quantum has any real advantage. Compressing that timeline by even a few years would reorder the competitive landscape for the companies and governments that have spent heavily to get there first.\n\nQuantum roadmaps have a long history of looking different two years later, and the simultaneous supremacy downgrade in this same news cycle is worth keeping in mind.","[\"quantum-computing\",\"amazon\",\"quera\",\"error-correction\"]","2026-06-17T20:44:32.000Z","2026-06-19T06:21:37.940Z","2026-06-19T14:20:56.985Z",[],[610,611,612,613],"quantum-computing","amazon","quera","error-correction",[615,617],{"name":124,"url":616},"https:\u002F\u002Farstechnica.com\u002Fscience\u002F2026\u002F06\u002Famazon-quera-promise-useful-quantum-error-correction-by-2028\u002F",{"name":124,"url":618},"https:\u002F\u002Farstechnica.com\u002Fscience\u002F2026\u002F06\u002Fquera-promises-thousands-of-error-corrected-qubits-by-2029\u002F",{"id":620,"slug":621,"title":622,"dek":623,"body_md":624,"tags_json":625,"published_at":626,"created_at":627,"updated_at":628,"status":91,"review_note":92,"review_notes":629,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":630,"sources":635,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},1297,"two-stage-multi-grade-neural-nets-cut-burgers-pde-errors-up-to-60","Staged Neural Network Training Cuts Physics Simulation Errors 60-fold","A two-stage training method reduces error in neural-net equation solvers by up to 60 times over standard single-pass approaches.","Neural networks that solve physics equations have a training problem - and a new paper proposes a structured fix.\n\nResearchers propose a two-stage, multi-grade deep learning method for solving partial differential equations (PDEs) - mathematical models that describe how quantities like heat, fluid, and pressure change across space and time. In the first stage, shallow networks are trained progressively: each new block learns only to reduce the error left by the one before it, while earlier layers stay frozen. The second stage unfreezes select layers and fine-tunes the full network, using the first stage as initialization. Tested on the viscous Burgers' equation - a standard benchmark known for steep gradients and shock-like behavior across one-, two-, and three-dimensional settings - the approach cuts predictive error by up to a factor of 60 compared to training a single deep network end-to-end.\n\nPDEs underpin fluid dynamics, weather modeling, and structural engineering, and neural solvers are increasingly attractive because they can approximate solutions faster than classical numerical methods at scale. The hierarchical structure also makes the resulting network easier to audit: each grade has a defined role, so the model carries more interpretable structure than a typical deep network trained all at once.\n\nA 60-fold improvement on a controlled academic benchmark is a headline-worthy number. Whether it holds on messier, higher-dimensional real-world problems is the question the next round of experiments will have to answer.","[\"deep learning\",\"physics\",\"neural networks\",\"simulation\"]","2026-06-16T04:00:00.000Z","2026-06-17T02:39:10.237Z","2026-06-19T13:16:30.296Z",[],[631,632,633,634],"deep learning","physics","neural networks","simulation",[636],{"name":161,"url":637},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2309.07401",{"id":639,"slug":640,"title":641,"dek":642,"body_md":643,"tags_json":644,"published_at":645,"created_at":646,"updated_at":647,"status":91,"review_note":92,"review_notes":648,"image_url":649,"persona_id":92,"persona_name":92,"section":13,"tags":650,"sources":653,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},966,"chinese-rocket-stage-breaks-up-near-starlink-constellation","Chinese Rocket Stage Breaks Apart Near Starlink and the ISS","A Zhuque-2E upper stage shattered in low-Earth orbit after last week's launch, scattering debris through one of the most congested stretches of space.","A Chinese commercial rocket's upper stage disintegrated in low-Earth orbit, adding a fresh debris cloud to an already crowded altitude band.\n\nThe Zhuque-2E rocket launched June 9, delivering two direct-to-cell communications satellites before its upper stage apparently failed to complete a planned disposal burn. Shortly after, the stage broke apart. The US Space Force confirmed the event via space-track.org, the military's public orbit-data portal, and said tracked fragments are being folded into routine conjunction assessments. As of its advisory, the Space Force found no immediate threats to human spaceflight — but noted analysis is ongoing.\n\nThe debris field sits in a slice of low-Earth orbit shared by the International Space Station and a large share of SpaceX's Starlink constellation, which now numbers well over 6,000 satellites. That density is precisely what makes upper-stage breakups costly: even a modest debris cloud forces operators to run avoidance maneuvers, burning fuel and disrupting service windows. The more satellites in a band, the more conjunctions each fragment creates. SpaceX has faced this math before with other operators' hardware; now it faces it again, and closer to home.\n\nLaunches from private Chinese operators like Landspace — the company behind Zhuque-2E — are multiplying, but their debris-mitigation track record is thin. The disposal burn that apparently didn't happen is exactly the kind of end-of-mission procedure that keeps orbital lanes usable. Whether this was a hardware failure or a planning gap, the result is the same: more junk in a neighborhood that can't afford more junk.","[\"space\",\"debris\",\"starlink\",\"satellites\"]","2026-06-15T18:55:41.000Z","2026-06-15T19:30:04.438Z","2026-06-18T13:27:24.144Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fchinese-rocket-stage-breaks-up-near-starlink-constellation.webp",[516,651,652,518],"debris","starlink",[654],{"name":124,"url":655},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F06\u002Fa-chinese-rocket-breaks-apart-dangerously-close-to-the-starlink-constellation\u002F",{"id":657,"slug":658,"title":659,"dek":660,"body_md":661,"tags_json":662,"published_at":663,"created_at":664,"updated_at":665,"status":91,"review_note":92,"review_notes":666,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":676,"sources":680,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},1211,"federated-learning-estimates-causal-effects-without-pooling-raw-data","Federated Learning Gets Causal Without Sharing Raw Data","A new method estimates causal treatment effects across siloed datasets by sharing only aggregate statistics, not individual records.","Researchers have worked out how to run causal inference across multiple data silos without any site surrendering its raw records.\n\nA new paper proposes estimating the Average Treatment Effect - a standard measure of whether an intervention actually works - from observational data distributed across multiple sites. Instead of pooling individual records, participating sites compute local propensity scores (the likelihood that a given observation received a treatment), then share only those aggregates. A weighting scheme called Membership Weights adjusts for the fact that each site's population differs from the whole, yielding two new estimators - Federated Inverse Propensity Weighting and an augmented variant - that the authors test on simulated and real-world data, outperforming meta-analysis approaches in both settings.\n\nMost data that could answer important questions in medicine, policy, and social science sits locked in silos: hospital networks barred from sharing patient records, agencies that will not export individual data across jurisdictions, or organizations that cannot justify the legal exposure. Standard meta-analysis collapses when any single site has a coverage gap, a condition researchers call positivity violation; this method turns cross-site heterogeneity into an asset rather than a problem by exploiting the fact that different sites may assign treatments differently.\n\nThis is academic research, not a production tool, and federated methods have a long history of promising more than they deliver outside the lab. Still, the specific gap being addressed here - extracting causal claims, not just correlations, from siloed observational data - is one with genuine demand in healthcare and policy research that existing approaches handle badly.","[\"federated-learning\",\"causal-inference\",\"privacy\",\"research\"]","2026-06-15T04:00:00.000Z","2026-06-16T18:34:19.855Z","2026-06-19T13:16:06.192Z",[667,669,672],{"id":95,"reviewer":96,"round":97,"reason":668,"status":99},"Add a concise concluding paragraph that summarizes the news and its implications, ensuring the article ends with a clear wrap‑up.",{"id":670,"reviewer":96,"round":393,"reason":671,"status":99},"editor-r2","Add a concise concluding paragraph that summarises the new method, its significance, and the open questions, providing a clear wrap‑up to the article.",{"id":673,"reviewer":96,"round":674,"reason":675,"status":99},"editor-r3",3,"Add a clear concluding paragraph that summarises the new federated causal inference method, its significance for privacy‑sensitive domains, and the remaining open questions.",[677,678,679,139],"federated-learning","causal-inference","privacy",[681],{"name":161,"url":682},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.17961",{"id":684,"slug":685,"title":686,"dek":687,"body_md":688,"tags_json":689,"published_at":663,"created_at":690,"updated_at":691,"status":91,"review_note":92,"review_notes":692,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":693,"sources":696,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},1208,"diffusion-models-improve-climate-emulator-detail-but-miss-worst-storms","AI Climate Emulators Show Promise, Fall Short on Extremes","A new diffusion model framework replicates regional precipitation patterns well but still misses the most severe weather events.","Researchers introduced a new generative AI framework for mimicking expensive climate simulations, and found it mostly works — except when it matters most.\n\nA team published ParamDiffusion, a two-stage diffusion-based model designed to emulate regional climate models (RCMs), the high-resolution simulations that translate coarse global forecasts into local precipitation data. They benchmarked it against three other deep-learning approaches using a validation framework built around climate-science criteria, not just standard ML metrics. Diffusion models reproduced precipitation statistics accurately, capturing spatial patterns and distributional tails. But when it came to the most extreme simulated events, none of the four models consistently included those within their uncertainty estimates.\n\nThat gap matters more than it sounds. Regional climate models are expensive to run in both compute time and energy, and cheap emulators could let researchers test far more scenarios. But extreme precipitation is precisely where flood planning, infrastructure design, and disaster risk assessment depend on getting the numbers right.\n\nDiffusion models have made a strong case in image generation. Climate science sets a harder standard, because miscalibrated tails in a precipitation forecast have consequences that a blurry photograph does not.","[\"climate\",\"machine-learning\",\"diffusion-models\",\"generative-ai\"]","2026-06-16T18:00:50.617Z","2026-06-19T13:16:06.135Z",[],[243,209,694,695],"diffusion-models","generative-ai",[697],{"name":161,"url":698},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.14570",{"id":700,"slug":701,"title":702,"dek":703,"body_md":704,"tags_json":705,"published_at":706,"created_at":707,"updated_at":708,"status":91,"review_note":92,"review_notes":709,"image_url":710,"persona_id":92,"persona_name":92,"section":13,"tags":711,"sources":716,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},915,"study-detects-hormone-disrupting-chemicals-in-us-breast-milk","Hormone-Disrupting Chemicals Found in US Breast Milk Samples","A new study detected dangerous endocrine disruptors in US breast milk, intensifying pressure on regulators who have long moved slowly on these compounds.","New research has turned up dangerous hormone-disrupting chemicals in breast milk samples collected from mothers in the United States.\n\nScientists analyzed samples of US breast milk and found chemicals classified as endocrine disruptors — compounds that can interfere with the body's hormone-signaling system. Infants are particularly sensitive to these exposures; their developing systems lack the defenses adults have built up over years. The findings add to a years-long body of evidence that everyday chemical contamination is reaching nursing children.\n\nThe context matters because breast milk is widely regarded as the optimal source of infant nutrition, which makes contamination findings politically and practically uncomfortable. US regulatory agencies have historically moved more slowly than their European counterparts to restrict hormone-disrupting compounds, a gap that shifts the burden of exposure from manufacturers onto individuals.\n\nThis is not the first study to flag chemical contamination in breast milk, and it will not be the last. Whether it moves policy depends less on the science — that case has been building for years — and more on whether regulators are willing to treat it as an industry problem rather than a parenting one.","[\"endocrine disruptors\",\"public health\",\"chemicals\",\"regulation\"]","2026-06-14T14:47:12.000Z","2026-06-14T15:30:17.453Z","2026-06-18T12:55:52.569Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fstudy-detects-hormone-disrupting-chemicals-in-us-breast-milk.webp",[712,713,714,715],"endocrine disruptors","public health","chemicals","regulation",[717],{"name":718,"url":719},"Hacker News","https:\u002F\u002Fwww.theguardian.com\u002Fus-news\u002F2026\u002Fjun\u002F14\u002Fbreast-milk-research-chemicals",{"id":721,"slug":722,"title":723,"dek":724,"body_md":725,"tags_json":726,"published_at":727,"created_at":728,"updated_at":729,"status":91,"review_note":92,"review_notes":730,"image_url":731,"persona_id":92,"persona_name":92,"section":13,"tags":732,"sources":735,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},787,"arxiv-hosts-new-paper-titled-maxproof","ArXiv hosts new paper titled Maxproof","A newly posted preprint called Maxproof appears on arXiv, drawing modest attention on Hacker News.","A preprint named Maxproof was uploaded to arXiv on June 12, 2026. The paper’s abstract and authors are listed on the repository, but no summary has been widely circulated yet.\n\nWhat actually happened: the submission shows up under the computer science category and has already earned 50 points and three comments on Hacker News. Beyond the basic bibliographic entry, the discussion thread is thin, with participants asking for clarification about the paper’s scope.\n\nWhy it matters: new arXiv submissions often signal emerging research directions, and early community reaction can hint at relevance. In this case, the limited chatter suggests the work has yet to capture broader interest or clarify its contribution.\n\nFor now, Maxproof remains a placeholder in the literature, awaiting deeper peer review and clearer explanation before its impact can be judged.","[\"arxiv\",\"preprint\",\"research\"]","2026-06-12T12:00:02.000Z","2026-06-12T14:00:11.605Z","2026-06-18T00:16:54.768Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Farxiv-hosts-new-paper-titled-maxproof.webp",[733,734,139],"arxiv","preprint",[736],{"name":718,"url":737},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.13473",{"id":739,"slug":740,"title":741,"dek":742,"body_md":743,"tags_json":744,"published_at":745,"created_at":746,"updated_at":747,"status":91,"review_note":92,"review_notes":748,"image_url":749,"persona_id":92,"persona_name":92,"section":13,"tags":750,"sources":755,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},767,"jwst-data-points-to-black-holes-behind-early-universe-red-dots","JWST Pins 'Little Red Dots' on Gas-Shrouded Black Holes","New spectral analysis of GLIMPSE-17775 suggests these early-universe objects are quasars buried in dense gas, and probably smaller than prior estimates.","New spectral data from the James Webb Space Telescope points to a straightforward explanation for the \"little red dots\" that briefly looked like they might break cosmology.\n\nA team led by Vasily Kokorev at the University of Texas at Austin analyzed GLIMPSE-17775, a compact red object that existed about 1.8 billion years after the Big Bang. By splitting its light into more than 40 distinct chemical signatures, they found that the light had bounced around inside a dense cocoon of gas before reaching us. That only happens when surrounding gas is far denser than anything astronomers see in an ordinary galaxy. At the center likely sits an actively feeding supermassive black hole, a configuration some researchers call a \"black hole star.\" The team published their findings in The Astrophysical Journal.\n\nThe study carries a quieter implication that matters as much as the headline result: prior mass estimates for these objects were probably unreliable. Earlier measurements relied on tracking how fast gas moves around the black hole, but in an environment this dense, the light is already heavily distorted before it reaches any telescope. If these black holes are smaller than they appeared, the puzzle of how something so massive formed so early in cosmic history becomes far less urgent, and one of the main reasons astronomers thought these objects were exotic simply dissolves.\n\nThe \"little red dots\" were briefly candidates for breaking cosmology. They may turn out to be ordinary quasars wearing unusually heavy coats.","[\"cosmology\",\"black holes\",\"jwst\",\"early universe\"]","2026-06-12T09:00:00.000Z","2026-06-12T10:06:02.402Z","2026-06-18T11:43:10.139Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fjwst-data-points-to-black-holes-behind-early-universe-red-dots.webp",[751,752,753,754],"cosmology","black holes","jwst","early universe",[756],{"name":142,"url":757},"https:\u002F\u002Fmashable.com\u002Fscience\u002Fjames-webb-space-telescope-black-hole-star-evidence",{"id":759,"slug":760,"title":761,"dek":762,"body_md":763,"tags_json":764,"published_at":765,"created_at":766,"updated_at":767,"status":91,"review_note":92,"review_notes":768,"image_url":771,"persona_id":92,"persona_name":92,"section":13,"tags":772,"sources":776,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},759,"austin-engineers-unveil-hydrogel-jacket-that-harvests-water-from-air","UT Austin Jacket Pulls 900 ml of Drinking Water from Air Daily","Researchers wove a biomass-derived hydrogel fiber into a wearable that passively harvests up to 900 ml of drinking water from ambient air each day.","Engineers at the University of Texas at Austin have built a wearable jacket that harvests drinkable water from the air around you.\n\nThe garment is woven from a biomass-derived hydrogel fiber engineered to absorb atmospheric moisture and release it as liquid water. In testing, it produced up to 900 ml per day, roughly the volume of a standard water bottle. The biomass origin means the base material is renewable rather than petroleum-derived, which matters for anyone thinking about manufacturing at scale.\n\nAtmospheric water harvesting has been a research theme for years, but most devices to date have been stationary, bulky, or dependent on external power. A passive wearable changes the delivery model: the person wearing it becomes the collection surface, which is relevant for hikers, disaster responders, and populations in arid regions where centralized water infrastructure is absent or unreliable. Nine hundred milliliters will not fully hydrate an adult working in extreme heat, but as a daily supplement it is not nothing.\n\nWhat the announcement leaves open: how the jacket holds up after repeated soaking and wringing, how much water it actually pulls in low-humidity climates where demand would be highest, and what it costs to produce outside a university lab.","[\"atmospheric water\",\"wearables\",\"materials science\",\"sustainability\"]","2026-06-12T03:50:18.000Z","2026-06-12T04:25:26.998Z","2026-06-18T11:36:37.578Z",[769],{"id":95,"reviewer":96,"round":97,"reason":770,"status":99},"The draft is truncated, contains placeholder text, lacks a clear lead, proper context, and any analysis; rewrite with a concise lead, factual details, and sober commentary.","https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Faustin-engineers-unveil-hydrogel-jacket-that-harvests-water-from-air.webp",[773,774,194,775],"atmospheric water","wearables","sustainability",[777],{"name":778,"url":779},"Digital Trends","https:\u002F\u002Fwww.digitaltrends.com\u002Fcool-tech\u002Fthis-jacket-pulls-drinking-water-straight-from-the-air\u002F",{"id":781,"slug":782,"title":783,"dek":784,"body_md":785,"tags_json":786,"published_at":787,"created_at":788,"updated_at":789,"status":91,"review_note":92,"review_notes":790,"image_url":791,"persona_id":92,"persona_name":92,"section":13,"tags":792,"sources":794,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},737,"nasa-deep-space-network-held-up-on-artemis-ii-after-near-overload","NASA's Deep Space Network Held Up for Artemis II, Mostly by Timing","The network that choked on Artemis I survived the crewed mission without throttling science data — helped more by a shorter flight than any capacity fix.","NASA's Deep Space Network made it through the Artemis II crewed mission without repeating its near-meltdown from 2022.\n\nDuring Artemis I, which flew in late 2022, the global array of deep space antennas — responsible for keeping contact with more than 40 robotic science missions simultaneously — couldn't absorb the extra load of NASA's Orion capsule. Downlinks from the James Webb Space Telescope and Mars rovers were delayed or cut as Artemis I took communications priority. For Artemis II, launched April 1 with four astronauts aboard, NASA's appetite for Orion data was even larger. But the mission lasted just over nine days versus Artemis I's 25, and it carried fewer CubeSats requiring their own tracking and telecom services. Shorter mission, fewer satellites: the math was friendlier.\n\nThe DSN is a chokepoint that only makes headlines when something breaks, and the underlying tension — a fixed antenna network serving a growing roster of science missions plus an increasingly demanding crewed program — hasn't been resolved. Artemis III, which involves a lunar surface landing, will almost certainly run higher communications demands and for longer. A nine-day success is not the same as a solved capacity problem.\n\nNASA got a favorable draw this time. The harder question is what happens when a crewed Moon mission doesn't cooperate with the calendar.","[\"space\",\"nasa\",\"artemis\",\"deep-space-network\"]","2026-06-11T18:34:01.000Z","2026-06-11T19:24:15.935Z","2026-06-18T11:09:04.517Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fnasa-deep-space-network-held-up-on-artemis-ii-after-near-overload.webp",[516,555,556,793],"deep-space-network",[795],{"name":124,"url":796},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F06\u002Fafter-nearly-breaking-nasas-deep-space-network-worked-well-on-artemis-ii\u002F",{"id":798,"slug":799,"title":800,"dek":801,"body_md":802,"tags_json":803,"published_at":804,"created_at":805,"updated_at":806,"status":91,"review_note":92,"review_notes":807,"image_url":808,"persona_id":92,"persona_name":92,"section":13,"tags":809,"sources":812,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},732,"us-solar-output-finally-eclipses-coal-for-a-full-month","Solar Tops Coal in US Power Generation for the First Time","A milestone a decade in the making: US solar farms collectively outproduced coal plants for the first time on record.","For the first time on record, solar power generated more electricity across the United States than coal.\n\nThe milestone, crossed in 2026, caps a shift that has been building for years. Coal's share of US generation has declined steadily as aging plants retire and go unrepaired, while solar capacity has surged on the back of falling hardware costs and federal incentives from climate legislation. The crossover was widely expected — the question for the past few years was when, not whether.\n\nCoal dominated US electricity for most of the 20th century, so the symbolic weight here is real even if the direction of travel has been visible for a decade. Grid operators now have to manage solar's intermittency at a scale that was largely theoretical not long ago — midday surpluses, sharp evening demand ramps, curtailment calls that coal-heavy grids never had to make.\n\nWorth keeping in perspective: \"more than coal\" still isn't \"dominant.\" Natural gas remains the largest single source of US electricity by a wide margin, and solar's lead will narrow once summer ends. A June record is a real one, but it is also a convenient one.","[\"energy\",\"solar\",\"coal\",\"climate\"]","2026-06-11T16:10:05.000Z","2026-06-11T18:27:15.264Z","2026-06-18T11:03:44.010Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fus-solar-output-finally-eclipses-coal-for-a-full-month.webp",[482,810,811,243],"solar","coal",[813],{"name":718,"url":814},"https:\u002F\u002Fwww.theguardian.com\u002Fus-news\u002F2026\u002Fjun\u002F11\u002Fsolar-energy-us-coal",{"id":816,"slug":817,"title":818,"dek":819,"body_md":820,"tags_json":821,"published_at":822,"created_at":823,"updated_at":824,"status":91,"review_note":92,"review_notes":825,"image_url":826,"persona_id":92,"persona_name":92,"section":13,"tags":827,"sources":831,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},705,"nsf-shuts-down-alaskas-deep-ocean-monitoring-network","NSF Shuts Down Ocean Network Alaska's Fisheries Depend On","The Ocean Observatories Initiative, a $368M sensor network used by fishers, coastal planners, and the military, is slated for decommissioning.","The NSF is pulling the plug on a nearly $368 million deep-ocean monitoring network that Alaska's fishing industry, coastal planners, and the military depend on.\n\nThe National Science Foundation announced in May it would decommission the Ocean Observatories Initiative, a network of underwater instruments that measures ocean chemistry, wave action, water temperature, and salinity in real time. The system was not a research prototype — it had become operational infrastructure. Fishery managers use its data to set catch limits. Coastal hazard planners rely on it for early warning of dangerous wave conditions. Even the military taps it, though the announcement did not elaborate on how.\n\nAlaska stands to lose the most. It is the country's top fish-producing state, and its waters are warming at roughly twice the global average rate — which makes real-time ocean data more valuable now than when the OOI was built, not less. A misread marine heatwave, or a fish-stock estimate off by a meaningful margin, translates into millions of dollars in economic damage and years of regulatory fallout.\n\nThe case for cutting the network has to be budgetary, because the scientific case for keeping it has only gotten stronger. Walking away from a $368 million investment at the exact moment ocean temperatures are setting records and salmon runs are growing less predictable is the kind of decision that tends to look worse in hindsight than it does in a budget spreadsheet.","[\"ocean\",\"climate\",\"alaska\",\"science-funding\"]","2026-06-11T13:19:00.000Z","2026-06-11T14:25:37.534Z","2026-06-18T10:40:50.286Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fnsf-shuts-down-alaskas-deep-ocean-monitoring-network.webp",[828,243,829,830],"ocean","alaska","science-funding",[832],{"name":124,"url":833},"https:\u002F\u002Farstechnica.com\u002Fscience\u002F2026\u002F06\u002Falaskans-will-be-flying-blind-after-nsf-decommissions-ocean-monitoring-network\u002F",{"id":835,"slug":836,"title":837,"dek":838,"body_md":839,"tags_json":840,"published_at":841,"created_at":842,"updated_at":843,"status":91,"review_note":92,"review_notes":844,"image_url":845,"persona_id":92,"persona_name":92,"section":13,"tags":846,"sources":851,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},661,"ultrasonic-espresso-cuts-brewing-energy-by-up-to-75","UNSW Researchers Brew Espresso With Sound, Not Heat","An ultrasonic method from UNSW Sydney extracts espresso at room temperature, cuts energy use by 75%, and fooled 100 blind tasters.","Researchers at UNSW Sydney built an espresso machine that replaces hot water with ultrasound — and a blind panel of 100 tasters couldn't tell the difference.\n\nThe team developed a process that pulses ultrasonic waves through room-temperature water and coffee grounds. The sound energy drives the same extraction that normally requires heat, pulling out oils, acids, and aromatic compounds without a heating element. Energy consumption drops by up to 75% compared to conventional machines. In blind tests, tasters could not reliably distinguish an ultrasonic shot from a hot-brewed one.\n\nEspresso machines are among the more power-hungry kitchen appliances — a typical home portafilter machine pulls 1,000 to 1,500 watts per brew cycle, and commercial units run harder. A 75% reduction without a detectable drop in cup quality is a real number, not a marketing rounding. Eliminating the boiler also removes warm-up time, which is its own quiet annoyance for anyone who brews before they're fully awake.\n\nThis is peer-reviewed research, not a product launch. No commercial ultrasonic espresso machine exists yet, and the gap between a university prototype and a kitchen appliance shelf is exactly where most promising coffee technology goes to quietly stall.","[\"coffee\",\"ultrasound\",\"energy efficiency\",\"food science\"]","2026-06-11T03:47:13.000Z","2026-06-11T04:22:50.379Z","2026-06-18T10:07:08.012Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fultrasonic-espresso-cuts-brewing-energy-by-up-to-75.webp",[847,848,849,850],"coffee","ultrasound","energy efficiency","food science",[852,854,857],{"name":778,"url":853},"https:\u002F\u002Fwww.digitaltrends.com\u002Fcool-tech\u002Fradical-new-coffee-making-method-uses-sound-skips-hot-water-and-reduces-energy-bills\u002F",{"name":855,"url":856},"TechRadar","https:\u002F\u002Fwww.techradar.com\u002Fhome\u002Fcoffee-machines\u002Fwho-needs-hot-water-researchers-made-an-ultrasonic-espresso-machine-that-brews-coffee-without-heat-and-drinkers-couldnt-notice-any-difference",{"name":858,"url":859},"Wired","https:\u002F\u002Fwww.wired.com\u002Fstory\u002Fscientists-brew-espresso-with-ultrasonic-waves\u002F",{"id":861,"slug":862,"title":863,"dek":864,"body_md":865,"tags_json":866,"published_at":867,"created_at":868,"updated_at":869,"status":91,"review_note":870,"review_notes":871,"image_url":875,"persona_id":92,"persona_name":92,"section":13,"tags":876,"sources":879,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},629,"fda-clears-new-uv-filter-bemotrizinol-for-us-sunscreens","FDA Clears First New Sunscreen Ingredient in 26 Years","The US just approved bemotrizinol, the first new sunscreen ingredient in 26 years - a UV filter that Europe and Asia have used safely since 2000.","The FDA has approved bemotrizinol as an active sunscreen ingredient - the first addition to the US-approved list since 1996.\n\nThe ingredient has been in use in European, Asian, and Australian sunscreens since 2000, where it earned a following for enabling lighter, non-greasy formulations. It filters both UVA and UVB rays, doesn't leave a white cast, and is more photostable than most existing chemical sunscreen ingredients - meaning it degrades more slowly on skin. With the approval, it becomes only the third ingredient to receive GRASE status (generally recognized as safe and effective), joining the two mineral filters, zinc oxide and titanium dioxide, that previously held that designation alone. DSM Nutritional Products will bring it to US shelves first, in a product called Parsol Shield expected later this year; other manufacturers get access in roughly 18 months.\n\nThe approval gap exists because the US classifies sunscreens as drugs rather than cosmetics, demanding clinical evidence that most other countries don't require - even for ingredients with decades of real-world safety data on three continents. That regulatory structure is why consumers here spent 25 years using formulations that dermatologists elsewhere considered outdated. The approval pathway was unlocked by the CARES Act, the 2020 pandemic relief legislation, which created a faster route for over-the-counter drug approvals backed by sufficient existing evidence.\n\nIt took a COVID relief bill to fix a sunscreen backlog that had been building since the Clinton administration - the FDA's OTC review process may want to sit with that.","[\"fda\",\"sunscreen\",\"regulation\",\"science\"]","2026-06-10T18:00:00.000Z","2026-06-10T19:32:20.936Z","2026-06-18T09:39:44.073Z","Publisher review could not be read (unparseable response); needs human review.",[872],{"id":873,"reviewer":392,"round":97,"reason":870,"status":874},"publisher-r1","open","https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Ffda-clears-new-uv-filter-bemotrizinol-for-us-sunscreens.webp",[877,878,715,13],"fda","sunscreen",[880],{"name":881,"url":882},"Lifehacker","https:\u002F\u002Flifehacker.com\u002Fhealth\u002Fnew-sunscreen-ingredient-fda-approved?utm_medium=RSS",{"id":884,"slug":885,"title":886,"dek":887,"body_md":888,"tags_json":889,"published_at":890,"created_at":891,"updated_at":892,"status":91,"review_note":870,"review_notes":893,"image_url":895,"persona_id":92,"persona_name":92,"section":13,"tags":896,"sources":899,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},626,"nasa-names-artemis-iii-crew-and-hints-at-dual-lander-docking","Artemis III Has a Crew and a Two-Lander Docking Plan","NASA named its lunar landing crew and confirmed astronauts will dock with both a Blue Origin and SpaceX lander in low-Earth orbit before any Moon descent.","NASA has named the Artemis III crew and confirmed a mission plan that includes dockings with two separate commercial landers in low-Earth orbit, targeting launch no earlier than summer 2027.\n\nThe agency announced the crew alongside details of a flight architecture that calls for rendezvous operations with both a Blue Origin lander and a SpaceX Starship lander while still in LEO. Artemis program manager Jeremy Parsons discussed additional details after the public announcement. Key questions remain open, including how ready each vehicle actually is and what the final form of each lander will actually look like.\n\nThe dual-lander architecture is the notable detail. NASA awarded lunar landing contracts to both SpaceX and Blue Origin, and building both vehicles into the same mission profile puts each company on a shared operational deadline. For the crew, it means additional docking operations before any lunar descent begins — and for NASA, it means two separate commercial programs both need to be ready on roughly the same schedule.\n\nArtemis III's summer 2027 window carries a familiar asterisk. \"No earlier than\" is NASA's language for a floor, not a forecast, and the program has a long record of moving dates to the right. Artemis I flew uncrewed in late 2022; Artemis II, a crewed lunar flyby, has itself slipped past multiple targets. Both commercial landers still have to demonstrate they can operate at the scale this mission demands — and that question, notably, remains open.","[\"nasa\",\"artemis\",\"space exploration\",\"moon\"]","2026-06-10T17:31:13.000Z","2026-06-10T19:22:56.137Z","2026-06-18T09:35:36.232Z",[894],{"id":873,"reviewer":392,"round":97,"reason":870,"status":874},"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fnasa-names-artemis-iii-crew-and-hints-at-dual-lander-docking.webp",[555,556,897,898],"space exploration","moon",[900],{"name":124,"url":901},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F06\u002Fwe-managed-to-glean-some-interesting-details-about-the-artemis-iii-mission\u002F",{"id":903,"slug":904,"title":905,"dek":906,"body_md":907,"tags_json":908,"published_at":909,"created_at":910,"updated_at":911,"status":91,"review_note":912,"review_notes":913,"image_url":923,"persona_id":92,"persona_name":92,"section":13,"tags":924,"sources":929,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},597,"china-launches-24mw-windpowered-underwater-data-centre","China's 24 MW Seabed Data Center Uses the Ocean to Stay Cool","A new facility on the seabed combines offshore wind power with seawater cooling, skipping both the grid and the chillers.","China has opened what it calls the world's first offshore wind-powered underwater data center, with an initial capacity of 24 megawatts.\n\nThe facility sits on the seabed and draws its power from offshore wind turbines rather than the national grid. It uses surrounding seawater as a passive cooling system, cutting out the chillers and cooling towers that consume a large share of energy at conventional facilities. At 24 MW the installation is modest by hyperscale standards, but it is positioned as a proof-of-concept combining two engineering bets: that submerged hardware can operate reliably over time, and that dedicated offshore wind can supply steady enough power to run it.\n\nBoth bets matter beyond this single site. Cooling is one of the hardest cost problems in data center design, and seawater sidesteps it almost entirely. Pairing the facility with its own wind farm rather than buying renewable-energy credits after the fact also means the sustainability claim is actual rather than accounting.\n\nMicrosoft ran a smaller seabed experiment off Scotland's coast and chose not to commercialize it. China's project is larger, wind-connected, and positioned as infrastructure rather than research. Whether it holds up under sustained load will tell the industry more than Microsoft's cautious trial did.","[\"data centers\",\"offshore wind\",\"renewable energy\",\"cooling\"]","2026-06-10T13:39:32.000Z","2026-06-10T14:54:23.930Z","2026-06-18T09:13:51.820Z","Add concrete details: exact offshore location, operating company, launch date, full Wired citation (author, date, URL), and compare typical data‑center power\u002Fcooling needs to show why 24 MW wind‑powered, seawater‑cooled setup is noteworthy.",[914,916,918,920],{"id":95,"reviewer":96,"round":97,"reason":915,"status":874},"Add concrete details (location, operator, launch date, capacity description), cite the source more precisely, and include brief context on why this matters compared to typical data‑center power and cooling.",{"id":670,"reviewer":96,"round":393,"reason":917,"status":874},"Add missing concrete details such as the exact offshore location, the company operating the centre, a precise citation to the Wired article (date, author, URL), and a brief comparison of typical data‑center power and cooling requirements to show why this 24 MW wind‑powered, seawater‑cooled setup is noteworthy.",{"id":673,"reviewer":96,"round":674,"reason":919,"status":874},"Add the missing specifics – exact offshore location, launch date, the operating company (with verification), and a full Wired citation (author, date, URL); also include a brief comparison of typical data‑center power and cooling needs to show why 24 MW wind‑powered, seawater‑cooled is notable.",{"id":921,"reviewer":96,"round":922,"reason":912,"status":874},"editor-r4",4,"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fchina-launches-24mw-windpowered-underwater-data-centre.webp",[925,926,927,928],"data centers","offshore wind","renewable energy","cooling",[930],{"name":858,"url":931},"https:\u002F\u002Fwww.wired.com\u002Fstory\u002Fchina-opens-worlds-first-wind-powered-underwater-data-center\u002F",{"id":933,"slug":934,"title":935,"dek":936,"body_md":937,"tags_json":938,"published_at":939,"created_at":940,"updated_at":941,"status":91,"review_note":92,"review_notes":942,"image_url":943,"persona_id":92,"persona_name":92,"section":13,"tags":944,"sources":947,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},570,"ai-designed-dna-vaccine-shows-coronavirus-protection-in-human-trial","Cambridge's AI-Designed Vaccine Passes Its First Human Trial","Cambridge researchers used AI to find conserved viral features, then built a needle-free DNA vaccine that just cleared its first human safety trial.","AI helped design a vaccine intended to work against entire coronavirus families, not just the strain it was built for, and its first human trial just came back safe.\n\nResearchers at the University of Cambridge used AI to scan thousands of viruses in the sarbecovirus family - which includes SARS, COVID-19, and related bat coronaviruses that could jump to humans - and identify genetic features that evolution has left largely unchanged across strains. Those conserved features became the blueprint for a DNA vaccine, which is more shelf-stable than the mRNA shots used during the pandemic and can be delivered without needles via a high-pressure liquid injection system. In its first human trial, the vaccine stimulated antibodies capable of recognizing multiple types of sarbecoviruses and was deemed safe and well-tolerated. The caveat: the researchers themselves describe the immune responses as modest, and larger trials are still needed to determine whether the vaccine can actually prevent infection.\n\nThe real advance here is the design method, not the result. Traditional vaccines target one known strain; AI's value is finding what a virus can't easily mutate away from, across thousands of variants at once. If that targeting logic holds up in larger trials, it could shift pandemic preparedness from reactive - chase the new variant after the outbreak starts - to anticipatory.\n\nA universal vaccine is still years away, this trial was small, and \"modest immune response\" is a considerable distance from the efficacy threshold regulators require. But for the first time, the framework for getting there is becoming concrete.","[\"ai\",\"vaccines\",\"biotech\",\"pandemic-preparedness\"]","2026-06-10T10:09:14.000Z","2026-06-10T11:24:56.562Z","2026-06-18T08:52:34.669Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fai-designed-dna-vaccine-shows-coronavirus-protection-in-human-trial.webp",[19,945,396,946],"vaccines","pandemic-preparedness",[948],{"name":142,"url":949},"https:\u002F\u002Fmashable.com\u002Fscience\u002Fworlds-first-ai-designed-vaccine-explained",{"id":951,"slug":952,"title":953,"dek":954,"body_md":955,"tags_json":956,"published_at":957,"created_at":958,"updated_at":959,"status":91,"review_note":92,"review_notes":960,"image_url":961,"persona_id":92,"persona_name":92,"section":13,"tags":962,"sources":966,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},553,"cornell-process-restores-used-ev-batteries-to-95-capacity","Cornell's Chemical Bath Gives Dead EV Batteries a Second Life","A Cornell University process restores spent lithium-ion battery cells to 95% of original capacity and cuts recycling costs by 56%, challenging the industry's shred-and-smelt default.","Cornell researchers say a chemical treatment can bring a worn-out EV battery back to near-original capacity — and make recycling cheaper in the same step.\n\nA team at Cornell University developed a process that soaks degraded lithium-ion battery cathodes in a chemical solution, repairing the crystal structure that breaks down over repeated charge cycles. In lab tests, treated cells recovered up to 95% of their original capacity. The process also cut recycling costs by 56% compared to conventional methods, which typically shred and smelt batteries to extract raw lithium, cobalt, and nickel — destroying the cell to recover the ingredients.\n\nThat distinction matters more than the headline number. The dominant recycling model is inherently destructive: you dismantle to reclaim, then pay again to refine and rebuild. A repair-first approach, if it scales, could extend useful battery life before recycling is necessary at all — relieving pressure on the mineral supply chain and addressing one of the quieter friction points in the used-EV market, where degraded capacity is a persistent drag on resale value. Unlike swappable battery schemes that require hardware standardization, a chemical treatment could work on cells already in the field.\n\nThe research is still in the lab, and the distance between a compelling chemical process and a cost-effective industrial operation has swallowed more than a few promising ideas.","[\"electric vehicles\",\"battery recycling\",\"lithium-ion\",\"research\"]","2026-06-10T06:02:03.000Z","2026-06-10T08:27:39.022Z","2026-06-18T08:38:15.297Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fcornell-process-restores-used-ev-batteries-to-95-capacity.webp",[963,964,965,139],"electric vehicles","battery recycling","lithium-ion",[967],{"name":778,"url":968},"https:\u002F\u002Fwww.digitaltrends.com\u002Fcool-tech\u002Fa-chemical-bath-could-bring-your-old-ev-battery-back-to-near-full-strength\u002F",{"id":970,"slug":971,"title":972,"dek":973,"body_md":974,"tags_json":975,"published_at":976,"created_at":977,"updated_at":978,"status":91,"review_note":92,"review_notes":979,"image_url":980,"persona_id":92,"persona_name":92,"section":13,"tags":981,"sources":985,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},504,"commonwealth-fusions-400-mw-design-gains-peerreview-backing","Five Papers Make the Physics Case for Commonwealth Fusion's Reactor","Commonwealth Fusion submitted five peer-reviewed papers modeling SPARC at 400 MW - more rigorous than the press-release promises that usually fill this field.","Commonwealth Fusion Systems published five peer-reviewed papers this week laying out the physics behind its planned 400 MW SPARC tokamak.\n\nThe papers, appearing together in a single journal issue, update the reactor's design and model its expected fusion output. CFS, an MIT spinout that has raised more than $2 billion in private capital, is building SPARC as a demonstration device before a larger commercial plant follows. Publishing the design in peer-reviewed journals invites independent physicists to check the math — a different kind of accountability than the press releases that have historically powered the fusion startup sector.\n\nIf the models hold, SPARC would be the first tokamak to produce substantially more fusion energy than it consumes, a milestone the field has chased for decades. The 400 MW figure is thermal fusion output, not net electricity delivered to the grid; how much the plant actually gains will depend on what it draws to sustain the plasma.\n\nPeer-reviewed models are not a working reactor. Fusion has a long history of confident projections that outlived the projects behind them, and CFS still has to build SPARC and make it run.","[\"fusion\",\"energy\",\"nuclear\",\"commonwealth fusion\"]","2026-06-09T20:25:56.000Z","2026-06-09T23:18:22.569Z","2026-06-18T08:11:22.816Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fcommonwealth-fusions-400-mw-design-gains-peerreview-backing.webp",[982,482,983,984],"fusion","nuclear","commonwealth fusion",[986],{"name":124,"url":987},"https:\u002F\u002Farstechnica.com\u002Fscience\u002F2026\u002F06\u002F__trashed-19\u002F",{"id":989,"slug":990,"title":991,"dek":992,"body_md":993,"tags_json":994,"published_at":995,"created_at":996,"updated_at":997,"status":91,"review_note":92,"review_notes":998,"image_url":999,"persona_id":92,"persona_name":92,"section":13,"tags":1000,"sources":1001,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},489,"ultrasound-wristband-reads-finger-motions-drives-robot-hand-live","MIT Ultrasound Wristband Maps 22 Hand Movements to Drive a Robot","A ring of wrist-worn ultrasound transducers can track every finger movement with enough fidelity to control a robot hand in real time.","MIT researchers have built an ultrasound wristband that maps 22 distinct hand movements and uses them to pilot a robotic hand in real time.\n\nPublished in Nature Electronics in March 2026, the device is a ring of small ultrasound transducers worn around the wrist. It captures what the team describes as 22 degrees of freedom — covering every major way a human hand can bend, rotate, and reposition its fingers. That data gets relayed to a robotic hand, which mirrors the movements without a physical connection to the fingers themselves. No instrumented glove required.\n\nThe wrist is a logical place to instrument a hand. The tendons and muscles that move the fingers all pass through the wrist on their way out, so a sensor worn there can infer a surprising amount about finger position without touching the hand directly. If that signal is reliable and low-latency enough to drive a robot in real time — and the Nature Electronics paper suggests it is — the approach could matter for prosthetics, remote surgery, and industrial teleoperation, all of which currently depend on systems that are either bulky, camera-dependent, or both.\n\nPeer review at a serious journal is a meaningful bar to clear. What the source doesn't address is how the system holds up across different hand sizes, or whether performance degrades outside controlled lab conditions — two questions that tend to separate interesting research from anything that ships.","[\"robotics\",\"wearables\",\"research\",\"ultrasound\"]","2026-06-09T19:17:53.000Z","2026-06-09T20:35:19.809Z","2026-06-18T07:59:52.031Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fultrasound-wristband-reads-finger-motions-drives-robot-hand-live.webp",[101,774,139,848],[1002],{"name":106,"url":1003},"https:\u002F\u002Fthenextweb.com\u002Fnews\u002Fmit-ultrasound-wristband-hand-tracking-robot-control",{"id":1005,"slug":1006,"title":1007,"dek":1008,"body_md":1009,"tags_json":1010,"published_at":1011,"created_at":1012,"updated_at":1013,"status":91,"review_note":92,"review_notes":1014,"image_url":1015,"persona_id":92,"persona_name":92,"section":13,"tags":1016,"sources":1020,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},439,"donut-labs-solidstate-battery-claim-proved-to-be-plain-lithiumion","Donut Lab's Solid-State Battery Was Just Lithium-Ion","A YouTube investigation backed by a whistleblower and over 20 battery experts says Donut Lab's production-ready solid-state cell is just lithium-ion.","A popular YouTube channel just took apart one of the battery industry's bigger claims.\n\nRyan Inis Hughes, who runs the Ziroth channel, published an investigation concluding that Donut Lab's solid-state battery, marketed as ready for mass production, is a conventional lithium-ion cell. Hughes described the gap as deliberate deception, not an honest mistake. His case drew on testimony from Lauri Peltola, the former Chief Commercial Officer of Nordic Nano Group, the manufacturer Donut Lab had named as its production partner. Hughes also cited more than 20 independent battery experts, including Julian Zahnow.\n\nSolid-state batteries have been the \"almost there\" promise of the EV and consumer-electronics industries for years. When a startup claims to have cracked mass production, capital moves. If the allegations hold, the damage runs two ways: investors who backed Donut Lab, and the broader field of researchers doing real work on solid-state chemistry who now share a headline with a fraud.\n\nDonut Lab is not the first energy startup to discover that bold claims outlast the chemistry. It probably will not be the last.","[\"batteries\",\"solid-state\",\"startups\",\"fraud\"]","2026-06-09T08:40:51.000Z","2026-06-09T09:14:55.903Z","2026-06-18T07:03:21.852Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fdonut-labs-solidstate-battery-claim-proved-to-be-plain-lithiumion.webp",[1017,1018,64,1019],"batteries","solid-state","fraud",[1021],{"name":1022,"url":1023},"The Verge","https:\u002F\u002Fwww.theverge.com\u002Fscience\u002F946608\u002Fdonut-labs-debunk-solid-state-battery",{"id":1025,"slug":1026,"title":1027,"dek":1028,"body_md":1029,"tags_json":1030,"published_at":1031,"created_at":1032,"updated_at":1033,"status":91,"review_note":92,"review_notes":1034,"image_url":1035,"persona_id":92,"persona_name":92,"section":13,"tags":1036,"sources":1041,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},417,"china-approves-first-commercial-brain-computer-implant","China Issues the First Commercial Approval for a Brain Implant","China's drug regulator approved NeuraMatrix's NEO implant for spinal cord patients, putting it ahead of Neuralink on the commercialization timeline.","China has become the first country to grant commercial regulatory approval to a brain-computer interface.\n\nThe National Medical Products Administration cleared NEO, a coin-sized implant developed by Shanghai-based NeuraMatrix alongside researchers at Tsinghua University, for use in patients with spinal cord injuries. The device translates neural signals into machine commands. The clearance is a full commercial approval, not a compassionate-use exemption or a limited research carve-out.\n\nNeuralink, the most-watched Western competitor, has conducted human trials under FDA oversight but has not received commercial clearance in the United States. That gap matters beyond bragging rights: commercial approval lets manufacturers build supply chains, train clinical teams, and price the device for real patients. Whoever reaches that milestone first on the regulatory track tends to set the clinical standards other markets reference.\n\nA narrow indication for spinal cord injuries is a measured start, and the obvious question regulators elsewhere will ask is whether NEO's safety and efficacy data would survive independent scrutiny. Commercial approval in China does not automatically transfer. But it establishes a precedent, and precedents in medical devices are hard to reverse.","[\"brain-computer interface\",\"neuralink\",\"china\",\"medical devices\"]","2026-06-08T18:46:18.000Z","2026-06-08T20:29:25.218Z","2026-06-18T06:38:15.168Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fchina-approves-first-commercial-brain-computer-implant.webp",[1037,1038,1039,1040],"brain-computer interface","neuralink","china","medical devices",[1042],{"name":106,"url":1043},"https:\u002F\u002Fthenextweb.com\u002Fnews\u002Fbrain-implants-bci-china-neuralink-commercial-race",{"id":1045,"slug":1046,"title":1047,"dek":1048,"body_md":1049,"tags_json":1050,"published_at":1051,"created_at":1052,"updated_at":1053,"status":91,"review_note":92,"review_notes":1054,"image_url":1055,"persona_id":92,"persona_name":92,"section":13,"tags":1056,"sources":1059,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},392,"russias-rassvet-broadband-network-loses-its-first-satellite","Russia's Rassvet Constellation Drops Its First Satellite","One of Rassvet's 16 satellites re-entered Earth's atmosphere within weeks, a rough start for Russia's answer to Starlink.","One satellite from Russia's fledgling Rassvet broadband constellation has already fallen back to Earth, barely weeks after its first operational batch launched.\n\nObject 4, one of 16 satellites in Rassvet's initial deployment, re-entered the atmosphere around June 6th. The other 15 remain in orbit. Russia has pitched Rassvet as its domestic answer to Starlink, a low-Earth-orbit broadband network it would control outright rather than sourcing from foreign providers it can no longer count on.\n\nLosing one in sixteen satellites this early is a different kind of problem than an established constellation would face. Starlink can absorb individual failures across thousands of spacecraft; Rassvet is working from a much thinner margin. A 6 percent attrition rate before the network is even operational is a harder number to dismiss, and there is no indication Russia has the launch cadence to backfill losses quickly.\n\nRussia has announced ambitious targets for Rassvet, including eventual nationwide broadband coverage. Fifteen satellites and one early re-entry in is a modest foundation for that promise.","[\"russia\",\"satellites\",\"broadband\",\"space\"]","2026-06-07T12:00:00.000Z","2026-06-08T14:32:08.358Z","2026-06-18T06:13:09.987Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Frussias-rassvet-broadband-network-loses-its-first-satellite.webp",[1057,518,1058,516],"russia","broadband",[1060],{"name":400,"url":1061},"https:\u002F\u002Fwww.tomshardware.com\u002Ftech-industry\u002Frussias-rassvet-constellation-loses-its-first-satellite-to-orbital-decay",{"id":1063,"slug":1064,"title":1065,"dek":1066,"body_md":1067,"tags_json":1068,"published_at":1069,"created_at":1070,"updated_at":1071,"status":91,"review_note":92,"review_notes":1072,"image_url":1073,"persona_id":92,"persona_name":92,"section":13,"tags":1074,"sources":1078,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},325,"peruvian-amazon-station-tests-low-impact-ecotourism-model","A Peruvian Field Station Puts Ecotourism's Claims to the Test","A citizen science station in the Peruvian Amazon is building a model where tourists collect real data — and the research funds conservation directly.","An Amazon research station is making the case that ecotourism doesn't have to be a contradiction in terms.\n\nA citizen science organization operating in the Peruvian Amazon has structured its model around an uncomfortable question: does ecotourism actually protect the ecosystems tourists come to see, or just charge admission to them? The station brings paying visitors in to conduct field research — species surveys, biodiversity monitoring, data collection that professional scientists need but rarely have enough hands to complete. Revenue from those visits funds the station's conservation operations directly. The goal is to close the loop between tourist dollars and measurable environmental outcomes, rather than leaving it as an article of faith.\n\nMost ecotourism works on the honor system. The lodge claims the proceeds go toward conservation, the brochure features a jaguar, and guests leave feeling good about the carbon offset they didn't calculate. Tying visitor activity to verifiable scientific output changes that dynamic, at least in principle: if the research is happening, the claim is checkable. In a region where deforestation and extractive industries continue to erode habitat, models that generate both data and funding may prove more durable than the solar-panel-and-composting variety.\n\nThe honest caveat is that this is one station, in one corner of the Amazon. Whether the model survives at scale — without diluting the science, the experience, or both — is a harder test than any single field season.","[\"ecotourism\",\"conservation\",\"citizen science\",\"amazon\"]","2026-06-05T16:00:00.000Z","2026-06-05T16:16:33.041Z","2026-06-18T05:06:13.497Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fperuvian-amazon-station-tests-low-impact-ecotourism-model.webp",[1075,1076,1077,611],"ecotourism","conservation","citizen science",[1079],{"name":858,"url":1080},"https:\u002F\u002Fwww.wired.com\u002Fstory\u002Fpreserving-the-places-tourists-go-to-study\u002F",{"id":1082,"slug":1083,"title":1084,"dek":1085,"body_md":1086,"tags_json":1087,"published_at":1088,"created_at":1089,"updated_at":1090,"status":91,"review_note":92,"review_notes":1091,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":1092,"sources":1097,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},268,"waymo-robotaxi-batteries-find-second-life-in-grid-storage","Waymo's Retired Taxi Batteries Head to the Power Grid","Used batteries from Waymo's robotaxi fleet will serve as stationary grid backup storage in California and Texas energy projects.","Waymo's used robotaxi batteries are getting a second life as stationary grid backup storage in California and Texas.\n\nBatteries pulled from retired Waymo vehicles will be repurposed for energy storage projects in two of the country's largest electricity markets. The arrangement applies a well-established concept, second-life battery programs, to a fleet operating at a scale most automakers haven't matched. Waymo runs one of the largest autonomous vehicle fleets in the US, which means it generates used battery packs on a predictable cycle and at meaningful volume. Details on the specific project operators and total storage capacity were not disclosed.\n\nGrid storage is expensive to build from scratch, and sourcing it from retired vehicle batteries rather than manufacturing new cells cuts costs while diverting material from landfills. California and Texas face different but pressing grid challenges: California manages renewable intermittency, while Texas runs an isolated grid whose fragility was badly exposed by the 2021 winter storm. Cheap, redeployed batteries address both problems without requiring a new factory.\n\nSecond-life battery programs have been discussed since at least the Nissan Leaf era, but most pilots stayed small. Waymo's fleet-scale supply of uniform packs could finally provide the volume advantage that makes the economics work where earlier efforts stalled.","[\"energy-storage\",\"waymo\",\"electric-vehicles\",\"grid\"]","2026-06-04T11:00:45.000Z","2026-06-04T21:19:51.694Z","2026-06-18T04:06:22.292Z",[],[1093,1094,1095,1096],"energy-storage","waymo","electric-vehicles","grid",[1098],{"name":124,"url":1099},"https:\u002F\u002Farstechnica.com\u002Fscience\u002F2026\u002F06\u002Fused-waymo-robotaxi-batteries-become-backup-storage-for-power-grids\u002F",{"id":1101,"slug":1102,"title":1103,"dek":1104,"body_md":1105,"tags_json":1106,"published_at":1107,"created_at":1108,"updated_at":1109,"status":91,"review_note":92,"review_notes":1110,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":1111,"sources":1113,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},247,"microsoft-claims-majorana-2-chip-boosts-qubit-reliability-1000fold","Microsoft's Majorana 2 Chip Cuts Quantum Deadline to 2029","A 1,000-fold qubit reliability leap, driven partly by agentic AI, pushes Microsoft's scalable quantum computer target forward by four years.","Microsoft's second topological quantum chip delivers qubits 1,000 times more reliable than last year's model, and the company has moved its scalable quantum computer deadline forward from 2033 to 2029.\n\nMicrosoft unveiled Majorana 2, the follow-up to the topological chip it introduced in 2025. The company credits agentic AI with driving the materials science discoveries behind the reliability leap. That improvement proved significant enough to revise the public roadmap, shaving four years off the timeline to a scalable quantum computer.\n\nMicrosoft has been the contrarian in the quantum race, betting on topological qubits while IBM and Google have spent years scaling conventional architectures. Topological qubits are theorized to be inherently more stable and error-resistant, but they have been stubbornly difficult to build reliably in hardware. If the 1,000x figure survives peer review, it would validate a decade-long wager and reshape the competitive calculus in Microsoft's favor.\n\nMicrosoft has retracted quantum claims before; a Nature paper asserting Majorana fermion evidence was pulled after internal review found data analysis problems. That history makes 2029 a deadline worth bookmarking, not banking on.","[\"quantum computing\",\"microsoft\",\"ai\",\"hardware\"]","2026-06-02T19:09:02.000Z","2026-06-02T21:49:44.606Z","2026-06-18T03:41:34.576Z",[],[156,1112,19,39],"microsoft",[1114],{"name":106,"url":1115},"https:\u002F\u002Fthenextweb.com\u002Fnews\u002Fmicrosoft-majorana-2-quantum-chip-agentic-ai-discovery",{"id":1117,"slug":1118,"title":1119,"dek":1120,"body_md":1121,"tags_json":1122,"published_at":1123,"created_at":1124,"updated_at":1125,"status":91,"review_note":92,"review_notes":1126,"image_url":1127,"persona_id":92,"persona_name":92,"section":13,"tags":1128,"sources":1130,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},142,"spacex-lands-416b-golden-dome-satellite-deal","SpaceX Wins $4.16B Pentagon Deal for Golden Dome Satellites","A new Space Force contract makes SpaceX the primary builder of Trump's missile-tracking satellite layer, just as the company eyes a record IPO.","SpaceX has secured a $4.16 billion Pentagon contract to build the satellite backbone of Trump's Golden Dome missile defense system.\n\nThe US Space Force announced the award on Friday, following Bloomberg's earlier report. The contract covers sensor-equipped satellites designed to detect and track airborne threats from orbit. SpaceX already holds separate contracts with the Space Force for other parts of the Golden Dome program, cementing its role as the system's primary infrastructure builder. The company is also reportedly preparing what could be the largest initial public offering in history.\n\nThe timing is worth noting. A multibillion-dollar defense contract is exactly what IPO investors want to see on the books before a market debut. More broadly, SpaceX is now positioned as a defense prime — the kind of role historically held by Lockheed Martin and Northrop Grumman — not just a commercial launch provider. Golden Dome is modeled on Israel's Iron Dome but at far greater scale: a space-based intercept layer designed to cover the entire United States.\n\nIron Dome's real-world effectiveness has been contested even in the relatively contained theater it was built for — a space-based missile intercept system is a significantly harder engineering problem, and that gap has not been addressed publicly.","[\"spacex\",\"defense\",\"satellites\",\"missile-defense\"]","2026-05-29T21:48:52.000Z","2026-05-30T09:01:57.036Z","2026-06-18T02:47:17.081Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fspacex-lands-416b-golden-dome-satellite-deal.webp",[573,575,518,1129],"missile-defense",[1131],{"name":1022,"url":1132},"https:\u002F\u002Fwww.theverge.com\u002Fscience\u002F940207\u002Fspacex-golden-dome-satellite-contract",{"id":1134,"slug":1135,"title":1136,"dek":1137,"body_md":1138,"tags_json":1139,"published_at":1140,"created_at":1141,"updated_at":1142,"status":91,"review_note":92,"review_notes":1143,"image_url":1144,"persona_id":92,"persona_name":92,"section":13,"tags":1145,"sources":1149,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},42,"investors-shift-toward-clean-energy-amid-geopolitical-tensions-security-concerns","Investors shift toward clean energy amid geopolitical tensions, security concerns","Money is flowing into renewables as nations seek energy independence during ongoing Middle East instability.","Investors are increasingly directing capital toward clean energy projects, driven by growing concerns about energy security amid geopolitical conflicts.\n\nThe shift comes as the ongoing conflict in the Middle East has disrupted traditional energy supply chains, prompting nations to reassess their reliance on fossil fuel imports. Institutional investors and sovereign wealth funds are viewing renewables not just as an environmental play, but as a strategic imperative for economic stability.\n\nClean energy sectors including solar, wind, and battery storage have seen a notable uptick in venture funding and infrastructure investment over the past quarter. This represents a departure from the sector's recent funding struggles, when high interest rates and supply chain uncertainties made capital more expensive.\n\nThe trend reflects a broader recalibration: energy security is now being discussed in boardrooms and government ministries as a matter of national strategy, not just climate policy. For investors, this creates a rare alignment between geopolitical risk mitigation and the long-term fundamentals of the energy transition.\n\nWhether this momentum holds depends on how the geopolitical situation evolves — and whether project costs continue to moderate.","[\"energy\",\"investing\",\"clean-tech\",\"geopolitics\"]","2026-05-03T09:26:48.000Z","2026-05-03T11:47:17.619Z","2026-06-17T22:56:16.302Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Finvestors-shift-toward-clean-energy-amid-geopolitical-tensions-security-concerns.webp",[482,1146,1147,1148],"investing","clean-tech","geopolitics",[1150],{"name":718,"url":1151},"https:\u002F\u002Fwww.ft.com\u002Fcontent\u002F9921f2b5-c910-4cec-a50f-cad453935a1a",{"id":1153,"slug":1154,"title":1155,"dek":1156,"body_md":1157,"tags_json":1158,"published_at":1159,"created_at":1160,"updated_at":1161,"status":91,"review_note":92,"review_notes":1162,"image_url":1163,"persona_id":92,"persona_name":92,"section":13,"tags":1164,"sources":1167,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},56,"neanderthals-operated-bone-marrow-fat-factories-125000-years-ago","Neanderthals Operated Bone-Marrow 'Fat Factories' 125,000 Years Ago","Archaeologists using chemical analysis found Neanderthals deliberately processed animal bones for fat, challenging assumptions about their cognitive abilities.","Researchers at Leiden University analyzed 125,000-year-old animal bones from a site in France and found chemical signatures indicating deliberate fat extraction — essentially running small-scale rendering operations.\n\nThe team examined bone fragments for specific lipid patterns that only appear when marrow is heated and processed, rather than simply consumed raw. This suggests Neanderthals understood that boiling bones yielded more calories than eating marrow directly, and they planned around this.\n\nWhy it matters: This adds to a growing pile of evidence that Neanderthals weren't brutish scavengers but capable of planning complex food-processing operations. The ability to conceptualize, plan, and execute a multi-step transformation process — bones to fat — requires cognitive abilities we haven't traditionally credited them with. It also raises questions about how we define \"technology\" — if rendering fat counts, Neanderthals had it thousands of years before Homo sapiens.\n\nThe findings join a recent trend of using biochemical analysis to read Neanderthal behavior in ways traditional archaeology couldn't. The tools have gotten better at reading what people actually did, not just what they left behind.","[\"archaeology\",\"research\",\"anthropology\",\"science\"]","2026-05-02T20:42:20.000Z","2026-05-03T11:48:02.471Z","2026-06-17T22:58:11.632Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fneanderthals-operated-bone-marrow-fat-factories-125000-years-ago.webp",[1165,139,1166,13],"archaeology","anthropology",[1168],{"name":718,"url":1169},"https:\u002F\u002Fwww.universiteitleiden.nl\u002Fen\u002Fnews\u002F2025\u002F07\u002Fneanderthals-ran-fat-factories-125000-years-ago",{"id":97,"slug":1171,"title":1172,"dek":1173,"body_md":1174,"tags_json":1175,"published_at":1176,"created_at":1177,"updated_at":1178,"status":91,"review_note":92,"review_notes":1179,"image_url":1180,"persona_id":92,"persona_name":92,"section":13,"tags":1181,"sources":1183,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},"six-under-the-radar-science-stories-worth-knowing","Six under-the-radar science stories worth knowing","From why dolphins swim so fast to how mushrooms use urine to talk to each other, a batch of research that deserves more attention.","Six scientific studies that didn't get the coverage they deserved.\n\nOne teamcrushed soda cans to understand how thin materials deform under pressure — useful for anyone designing packaging or studying structural failures. Researchers also figured out why dolphins can swim so fast: their skin produces a mucus-like slime that reduces drag. Meanwhile, scientists discovered that mushrooms communicate using urine, releasing nitrogen-rich compounds that act as chemical signals to neighboring fungi.\n\nThese studies represent the kind of basic research that gets overshadowed by flashier announcements but often leads to practical applications down the line. The soda can work could inform better packaging design. The dolphin finding adds to our understanding of marine biomechanics. And the mushroom communication research adds to a growing body of evidence that fungi are more socially complex than previously thought.\n\nThe broader lesson: the most interesting science doesn't always come with a press release.","[\"science\",\"research\",\"biology\"]","2026-05-02T14:23:31.000Z","2026-05-03T11:44:46.194Z","2026-06-17T22:50:34.152Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Fsix-under-the-radar-science-stories-worth-knowing.webp",[13,139,1182],"biology",[1184],{"name":124,"url":1185},"https:\u002F\u002Farstechnica.com\u002Fscience\u002F2026\u002F05\u002Fresearch-roundup-6-cool-science-stories-we-almost-missed-4\u002F",{"id":1187,"slug":1188,"title":1189,"dek":1190,"body_md":1191,"tags_json":1192,"published_at":1193,"created_at":1194,"updated_at":1195,"status":91,"review_note":92,"review_notes":1196,"image_url":1197,"persona_id":92,"persona_name":92,"section":13,"tags":1198,"sources":1202,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},5,"rare-amoeba-infection-killed-man-after-months-of-symptoms","Rare Amoeba Infection Killed Man After Months of Symptoms","Doctors finally diagnosed a man's mysterious worsening condition as a parasitic infection, but treatment came too late.","A man died after doctors spent months trying to understand why his skin was covered in necrotic lesions.\n\nThe 55-year-old patient's condition baffled physicians. Tests eventually confirmed the lesions were caused by free-living amoebas — microscopic organisms that can invade the body through breaks in the skin. By the time doctors identified the parasite and began treatment, the infection had spread too far. He died despite aggressive antifungal and antiparasitic therapy. Doctors suspect three factors — each unremarkable on its own — combined to create a fatal outcome.\n\nThis case highlights how difficult it can be to diagnose rare parasitic infections. The amoebas involved are common in soil and water worldwide, yet severe infections remain exceptionally uncommon. Most doctors will go their entire careers without seeing a case like this.\n\nThe lesson, if there is one, is less about avoiding a specific threat and more about the limits of modern medicine when faced with something outside the usual playbook.","[\"health\",\"medicine\",\"parasites\"]","2026-05-01T21:05:37.000Z","2026-05-03T11:45:10.215Z","2026-06-17T22:51:06.816Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Frare-amoeba-infection-killed-man-after-months-of-symptoms.webp",[1199,1200,1201],"health","medicine","parasites",[1203],{"name":124,"url":1204},"https:\u002F\u002Farstechnica.com\u002Fhealth\u002F2026\u002F05\u002Famoebas-eat-man-alive-over-months-in-puzzling-ultra-rare-cautionary-tale\u002F",{"id":1206,"slug":1207,"title":1208,"dek":1209,"body_md":1210,"tags_json":1211,"published_at":1212,"created_at":1213,"updated_at":1214,"status":91,"review_note":92,"review_notes":1215,"image_url":1216,"persona_id":92,"persona_name":92,"section":13,"tags":1217,"sources":1218,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},114,"alzheimers-scientists-say-money-and-policy-matter-as-much-as-lab-work","Alzheimer's Scientists Say Money and Policy Matter as Much as Lab Work","A leading researcher argues that funding gaps and fragmented efforts—not just biology—are slowing progress on dementia treatments.","At a recent WIRED Health event, Alzheimer's researcher John Hardy laid out the challenges facing the field. Decades of research and billions in funding have produced few effective treatments for the disease. Hardy argued that the next breakthrough will require more than scientific discovery—it will need better coordination, more money, and smarter policy.\n\nThe reason this matters is straightforward: the traditional model of individual labs chasing individual discoveries hasn't delivered for patients. Hardy suggested the field needs to treat funding and coordination as seriously as it treats protein folding or amyloid hypotheses. Without that shift, another decade could pass with little to show.\n\nIt's a familiar argument in medical research, but it carries weight coming from someone who's spent decades in the trenches. Whether the money and coordination will materialize is another question—the field has historically struggled with both.","[\"healthcare\",\"research\",\"science\",\"policy\"]","2026-05-01T18:55:56.000Z","2026-05-03T11:51:14.779Z","2026-06-17T23:06:10.321Z",[],"https:\u002F\u002Fcdn.xyz.onl\u002Farticle-images\u002Falzheimers-scientists-say-money-and-policy-matter-as-much-as-lab-work.webp",[138,139,13,34],[1219],{"name":858,"url":1220},"https:\u002F\u002Fwww.wired.com\u002Fstory\u002Fjohn-hardy-dementia-wired-health\u002F",{"id":1222,"slug":1223,"title":1224,"dek":1225,"body_md":1226,"tags_json":1227,"published_at":1228,"created_at":1229,"updated_at":1230,"status":91,"review_note":92,"review_notes":1231,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":1232,"sources":1235,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},10,"some-scorpions-evolved-metal-reinforced-stingers-and-scientists-want-to-know-why","Some scorpions evolved metal-reinforced stingers and scientists want to know why","Researchers found zinc and manganese in scorpion stingers, but the evolutionary driver remains unclear.","Some scorpions are essentially metal machines.\n\nResearchers examining scorpion stingers discovered that certain species embed zinc and manganese into the tips of their weapons, creating a biometal composite. The metal makes the stinger harder and more resistant to wear. Different hunting strategies appear to correlate with different metal distributions.\n\nThe finding is notable because most animals use metals for display or structural purposes, not weapon reinforcement. Scorpions that ambush prey showed higher metal concentrations than those that actively hunt, suggesting the reinforcement helps their stinger survive repeated forceful use against tough exoskeletons.\n\nWhy this matters: It's a rare example of a predator evolving internal metal reinforcement purely for mechanical function. The adaptation seems to have nothing to do with warning coloration or defense displays—just making a better puncture tool. Other arachnids and insects haven't independently evolved this trick, which makes it an unusual evolutionary outlier.\n\nThe researchers still don't know exactly when this trait evolved or what evolutionary pressure triggered it.","[\"biology\",\"science\",\"animals\",\"evolution\"]","2026-05-01T16:24:45.000Z","2026-05-03T11:45:26.954Z","2026-06-17T22:51:48.907Z",[],[1182,13,1233,1234],"animals","evolution",[1236],{"name":124,"url":1237},"https:\u002F\u002Farstechnica.com\u002Fscience\u002F2026\u002F05\u002Fscorpions-go-terminator-mode-and-reinforce-their-weapons-with-metal\u002F",{"id":1239,"slug":1240,"title":1241,"dek":1242,"body_md":1243,"tags_json":1244,"published_at":1245,"created_at":1246,"updated_at":1247,"status":91,"review_note":92,"review_notes":1248,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":1249,"sources":1253,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},13,"virgin-galactics-new-ship-arrives-as-money-runs-short","Virgin Galactic's new ship arrives as money runs short","The suborbital tourism business was supposed to take off years ago. Now one of its pioneers is burning the last of its cash.","Virgin Galactic showed off a new spacecraft this week. It may also be the last one it can afford to build.\n\nThe company unveiled its next-generation suborbital vehicle at an event, its first new ship design since 2020. But the celebration was muted by a financial reality that's grown harder to ignore. Virgin Galactic's latest filings show it held $1.1 billion in cash reserves at the end of last year—a figure that sounds large until you consider the company has burned through roughly $3 billion since 2018 with little to show for it. The new ship will require an extended testing program before carrying tourists, and the company hasn't said how it plans to fund that timeline.\n\nThis isn't just Virgin Galactic's problem. The entire suborbital space tourism sector is looking increasingly like a business that promised the future and delivered a handful of wealthy joyrides. Blue Origin flew its first tourists in 2021 and has barely flown since. The model—brief hops to the edge of space at ticket prices north of $250,000—has never made economic sense, and it still doesn't. What looked like a nascent industry in the 2010s now reads more like an expensive experiment that ran out of steam.\n\nThe new ship is technically impressive. Whether anyone will be able to afford to fly on it is a separate question entirely.","[\"space\",\"space tourism\",\"virgin galactic\",\"business\"]","2026-05-01T14:42:45.000Z","2026-05-03T11:45:36.277Z","2026-06-17T22:52:13.779Z",[],[516,1250,1251,1252],"space tourism","virgin galactic","business",[1254],{"name":124,"url":1255},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F05\u002Fthe-suborbital-space-tourism-industry-is-on-life-support\u002F",{"id":1257,"slug":1258,"title":1259,"dek":1260,"body_md":1261,"tags_json":1262,"published_at":1263,"created_at":1264,"updated_at":1265,"status":91,"review_note":92,"review_notes":1266,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":1267,"sources":1268,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},16,"amazons-broadband-constellation-grows-to-61-satellites-in-a-week","Amazon's broadband constellation grows to 61 satellites in a week","The company that once promised to launch 3,200 satellites but has been slow to launch just reached another milestone in its race to compete with Starlink.","Amazon's Project Kuiper added 61 satellites to orbit across two launches this week. The company is playing catch-up in the low-Earth-orbit broadband race, where SpaceX's Starlink already operates thousands of satellites.\n\nThe launches bring Amazon's total constellation to a fraction of what competitors have deployed. After years of delays and regulatory battles, the company is now attempting to scale up deployment rapidly. The FCC required Amazon to launch half its planned constellation by 2026.\n\nAmazon is building its satellites in-house at a facility in Kirkland, Washington, a bet that vertical integration will reduce costs. Whether that strategy works against SpaceX's reusable rockets remains to be seen. The company has contracts for dozens of launches across multiple providers, but SpaceX's launch cadence remains unmatched.\n\nThe broadband market is getting crowded. Amazon is joining Starlink, OneWeb, and others fighting for market share in a sector where profitability has proven elusive.","[\"space\",\"satellites\",\"broadband\",\"amazon\"]","2026-05-01T12:39:03.000Z","2026-05-03T11:45:46.108Z","2026-06-17T22:52:38.194Z",[],[516,518,1058,611],[1269],{"name":124,"url":1270},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F05\u002Frocket-report-falcon-heavy-is-back-russias-soyuz-5-finally-debuts\u002F",{"id":1272,"slug":1273,"title":1274,"dek":1275,"body_md":1276,"tags_json":1277,"published_at":1278,"created_at":1279,"updated_at":1280,"status":91,"review_note":92,"review_notes":1281,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":1282,"sources":1283,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},121,"one-injection-helped-joints-repair-themselves-in-weeks","One Injection Helped Joints Repair Themselves in Weeks","Researchers developed a treatment that may reverse osteoarthritis damage with a single injection, though it's years from availability.","Researchers say they've created a single injection that helps damaged joints repair themselves within weeks — a potential first for osteoarthritis, a condition with no cure.\n\nThe treatment works by stimulating the joint's own cells to regenerate, rather than just managing pain like current therapies. In early tests, damaged tissue showed visible repair within weeks. Osteoarthritis affects tens of millions of people worldwide, causing pain and stiffness as cartilage breaks down over time.\n\nThis matters because every existing treatment manages symptoms: pain relievers, anti-inflammatories, or eventually joint replacement surgery. A therapy that actually reverses damage would be a fundamental shift in how we treat a condition that gets worse with age. The research is still early — years of trials remain before any approval — but the approach represents a genuine departure from the symptom-management status quo.\n\nThe catch: promising results in research settings don't always hold up in larger human trials. Don't cancel your orthopedic appointment yet.","[\"health\",\"research\",\"biotech\",\"medicine\"]","2026-05-01T09:30:00.000Z","2026-05-03T11:51:38.016Z","2026-06-17T23:07:06.214Z",[],[1199,139,396,1200],[1284],{"name":858,"url":1285},"https:\u002F\u002Fwww.wired.com\u002Fstory\u002Fosteoarthritis-joint-damage-single-injection-treatment\u002F",{"id":74,"slug":1287,"title":1288,"dek":1289,"body_md":1290,"tags_json":1291,"published_at":1292,"created_at":1293,"updated_at":1294,"status":91,"review_note":92,"review_notes":1295,"image_url":92,"persona_id":92,"persona_name":92,"section":13,"tags":1296,"sources":1299,"feedback":108,"feedback_at":92,"cost_usd":108,"total_tokens":108},"russia-hides-launch-schedule-after-ukraine-hits-cosmodrome","Russia hides launch schedule after Ukraine hits cosmodrome","Ukraine's strike on Russia's primary spaceport forces a secrecy shift in orbital operations.","Ukraine struck Russia's Plesetsk Cosmodrome, the country's main orbital launch facility, prompting Moscow to stop publishing its launch schedule. Russian space officials acknowledged \"serious inbound attempts\" to the cosmodrome but provided few details about the damage or casualties. The attack marks a significant escalation in Ukraine's ability to project force deep into Russian territory.\n\nThe incident raises questions about Russia's space infrastructure resilience and its ability to maintain regular orbital operations. Plesetsk has been the workhorse of Russia's space program for decades, handling both civilian and military payloads. This is the first time a major spaceport has been directly targeted in an active conflict, setting a precedent that will likely influence how nations protect critical space infrastructure going forward.\n\nRussia's decision to cloak its launch schedule suggests concerns about exposing future launch windows to potential follow-up strikes.","[\"conflict\",\"space\",\"russia\",\"ukraine\"]","2026-04-30T21:35:02.000Z","2026-05-03T11:46:01.445Z","2026-06-17T22:53:11.892Z",[],[1297,516,1057,1298],"conflict","ukraine",[1300],{"name":124,"url":1301},"https:\u002F\u002Farstechnica.com\u002Fspace\u002F2026\u002F04\u002Frussian-cloaks-launch-schedule-after-spaceport-falls-in-ukraines-sights\u002F",{"sections":1303},[1304,1305,1306,1307,1308,1309,1310,1311,1312,1313,1314,1315,1316,1317],{"name":18,"slug":19,"count":20,"latest_published_at":21},{"name":23,"slug":24,"count":25,"latest_published_at":26},{"name":28,"slug":29,"count":30,"latest_published_at":31},{"name":33,"slug":34,"count":35,"latest_published_at":36},{"name":38,"slug":39,"count":40,"latest_published_at":41},{"name":43,"slug":44,"count":45,"latest_published_at":46},{"name":48,"slug":49,"count":50,"latest_published_at":51},{"name":12,"slug":13,"count":14,"latest_published_at":15},{"name":53,"slug":54,"count":55,"latest_published_at":56},{"name":58,"slug":59,"count":60,"latest_published_at":61},{"name":63,"slug":64,"count":60,"latest_published_at":65},{"name":67,"slug":68,"count":69,"latest_published_at":70},{"name":72,"slug":73,"count":74,"latest_published_at":75},{"name":77,"slug":78,"count":79,"latest_published_at":80}]