Trending news and stories on Mushimu

Israeli tanks, bulldozers, and ground troops pushed west of Gaza’s "Yellow Line" on Wednesday, November 18, 2026, launching the first major ground operation publicly announced by the Israel Defense Forces (IDF) since the ceasefire took effect on October 10, 2025. The incursion struck central Gaza sectors northeast of the Nuseirat refugee camp and north of Al-Bureij, severing traffic along a stretch of Salah al-Din Road—the enclave’s primary north-south highway—and sending displaced Palestinian families fleeing makeshift shelters under artillery fire. Confirmed across reporting from [BBC News](https://www.bbc.co.uk/news/articles/cw8r7510degno), [Al Jazeera](https://www.aljazeera.com/news/2026/11/18/what-does-israels-new-ground-offensive-mean-for-palestinians-in-gaza), [Middle East Eye](https://www.middleeasteye.net/news/israel-launches-assault-targeting-residential-areas-gaza-despite-ceasefire), and [Xinhua](https://english.news.cn/20261119/1b9dbef7fea14d98a377800c1adc5bf9/c.html), the escalation unfolded just hours after the U.S.-backed Board of Peace warned the United Nations Security Council that the 13-month-old truce was [at risk of collapse](https://www.mushimu.com/post/9614). ### Why Central Gaza—and Why Beyond the Yellow Line? Under the first phase of the October 2025 truce, Israeli forces withdrew to a demarcation boundary known as the Yellow Line, initially retaining control over roughly 53% of Gaza before gradually shifting barriers deeper into the Strip to encompass more than 60%—nearly two-thirds—of the territory. More than two million Palestinians, roughly 90% of them displaced and 1.4 million living in tents or improvised shelters, remain crowded inside the remaining western strip. Early Wednesday afternoon, COGAT—the Israeli defense body overseeing civilian affairs in the occupied Palestinian territory—issued urgent Arabic-language evacuation orders covering residential blocks **660, 661, and 666** near Nuseirat and Bureij and closed the adjacent section of Salah al-Din Road until further notice. In a joint statement, the IDF and the Shin Bet domestic security service described the push as a "precise" and targeted operation to locate and dismantle a "terror tunnel" and Hamas military infrastructure west of the Yellow Line that posed an "immediate and direct threat to IDF troops." Israeli Defense Minister Israel Katz reinforced that posture, warning that Israel will not pull out until Hamas is dismantled and remains prepared to "seize additional territory, evacuate local residents and incorporate it into Israel’s permanent security perimeter." ### Displacement Under Fire and Diverging Accounts on the Ground Witnesses and local Palestinian accounts paint a starkly different picture of the incursion’s timing and civilian impact. Residents interviewed by Middle East Eye and BBC Arabic said tanks and bulldozers advanced roughly 1.5 kilometers west of the Yellow Line under heavy artillery shelling and gunfire at the very moment evacuation warnings were broadcast. "We couldn't take any of our belongings with us—not even our blankets, pillows, or clothes," one Nuseirat resident told BBC Arabic. Video verified by BBC Verify captured a civilian ducking amid artillery blasts while running along Salah al-Din Road, alongside audible automatic gunfire. Separate strikes and shelling compounded the toll across the Strip on Wednesday. Palestinian news agency WAFA and al-Awda Hospital officials reported one person killed and several wounded in al-Mughraqa, while Gaza’s Civil Defence agency said three separate Israeli drone strikes killed three men in Nuseirat, Gaza City, and Deir al-Balah. Asked about civilian casualties, the IDF reiterated its general operational statement and told AFP that two of the drone strikes targeted "terrorists," while saying it had no information on the third strike. Hamas condemned the tank advance and forced evacuations as "a dangerous development and a flagrant violation of the ceasefire agreement," urging international mediators to intervene immediately and accusing Prime Minister Benjamin Netanyahu’s caretaker government of escalating militarily to deflect domestic political crises. The Popular Front for the Liberation of Palestine warned that the push served as a prelude to permanent territorial contraction. ### Ceasefire Fragility and the Disarmament Deadlock Wednesday’s incursion exposes the structural deadlock at the heart of the stalled peace framework. According to Gaza’s Health Ministry—whose overall figures the UN treats as reliable—at least 1,658 Palestinians have been killed and 6,000 wounded in near-daily Israeli strikes since the October 2025 ceasefire began, bringing the cumulative toll since October 7, 2023, to more than 74,700, while the Israeli military reports six soldiers killed during the truce period. With an estimated 82% of Gaza’s structures damaged or destroyed and $18 billion in international pledges still stalled, negotiations remain frozen over sequencing: Hamas has signaled willingness to surrender weapons only alongside a phased Israeli withdrawal from the 60% of Gaza currently occupied, whereas Israel insists on full disarmament prior to any pullout. By crossing the Yellow Line with armor rather than relying solely on airstrikes, Israel’s security cabinet has signaled that if diplomatic disarmament stalls, the demarcation line itself remains open to unilateral military revision. ### Sources & Further Reading - [BBC News (UK)](https://www.bbc.co.uk/news/articles/cw8r7510degno) - [Al Jazeera (Qatar)](https://www.aljazeera.com/news/2026/11/18/what-does-israels-new-ground-offensive-mean-for-palestinians-in-gaza) - [Middle East Eye (UK / Middle East)](https://www.middleeasteye.net/news/israel-launches-assault-targeting-residential-areas-gaza-despite-ceasefire) - [Xinhua (China)](https://english.news.cn/20261119/1b9dbef7fea14d98a377800c1adc5bf9/c.html) ~WorldNews ~Gaza ~Israel ~MiddleEast ~Ceasefire
Israeli Forces Launch First Major Ground Operation Beyond Gaza's 'Yellow Line' Near Nuseirat and Bureij
0
Discuss 11
Rockstar Games has poured cold water on hopes that *Grand Theft Auto 6* could receive a post-launch 60 FPS or even 40 FPS performance mode on PlayStation 5 Pro, confirming that Sony’s mid-generation console lacks the processor headroom required to run Leonida’s world simulation any faster than 30 frames per second. In an explicit technical statement provided to [Kotaku](https://kotaku.com/gta-6-wont-get-a-60-fps-mode-even-on-ps5-pro-because-its-not-powerful-enough-2000755979) and reported on November 18, 2026, by [Eurogamer](https://www.eurogamer.net/ps5-pro-not-powerful-enough-gta-6-60fps), [Clarín](https://www.clarin.com/estados-unidos/gta-6-modo-60-40-fps-rockstar-explica-bloqueado-30-fps-ps5-xbox_0_pIqAAoQ5Z7.html), and [Beebom](https://beebom.com/ps5-pro-not-powerful-enough-to-handle-gta-6-at-60-fps/), Rockstar Technology Production Director **Marc Chambers** drew a sharp line between GPU visual enhancements and CPU simulation limits. "While the PS5 Pro’s central processor is slightly better than the PS5 base console, it is not powerful enough to run the game’s complex simulation at higher frame rates," Chambers said. ### Why PS5 Pro’s GPU Upgrade Cannot Buy 40 FPS or 60 FPS For much of the PS5 and Xbox Series X|S generation, players grew accustomed to choosing between a 30 FPS "Quality" preset and a 60 FPS "Performance" mode—or, on 120 Hz televisions, a middle-ground 40 FPS mode. That industry habit fueled speculation that PS5 Pro’s hardware muscle would eventually unlock at least a 40 FPS patch for *GTA 6* after launch. Chambers’ explanation highlights the architectural bottleneck behind Sony’s mid-generation refresh. While the PS5 Pro significantly expands graphics compute units and adds dedicated PlayStation Spectral Super Resolution (PSSR) upscaling hardware, it retains the same core AMD Zen 2 CPU architecture as the 2020 base console, offering only a modest clock boost rather than a generational leap in single-thread or multi-thread throughput. In a dense open-world sandbox like *GTA 6*, every single frame requires the CPU to update NPC crowd routines, multi-lane traffic systems, police response logic, vehicle physics, streaming geometry, and ray-tracing bounding volume hierarchies (BVH). Cutting frame time from 33.3 milliseconds (30 FPS) down to 25 milliseconds (40 FPS) or 16.6 milliseconds (60 FPS) cannot be solved simply by lowering internal rendering resolution—because resolution reductions only relieve GPU pixel shading, not CPU simulation work. As Digital Foundry recently documented in its launch [frame-rate breakdown](https://www.mushimu.com/post/9593), even base PS5 and Xbox Series X dip below 30 FPS during heavy high-speed chases and explosive shootouts, confirming how tight Rockstar’s CPU budget already is. ### How Rockstar Spends the PS5 Pro’s Extra Silicon Rather than stripping back crowd density or AI systems to chase higher frame rates on a single console SKU, Rockstar used the PS5 Pro’s extra graphics and AI-upscaling budget to elevate lighting and image reconstruction. According to Chambers, *GTA 6*’s baseline presentation on current consoles relies on ray-traced reflections and ray-traced global illumination. On PS5 Pro, the studio adds **ray-traced shadows** on top of that stack while deploying **PSSR** for cleaner edge reconstruction and sharper distant detail. That trade-off makes PS5 Pro the cleanest console showcase for Vice City and Leonida’s neon reflections, dynamic weather, and dense foliage—yet it also turns the console into a pure visual-fidelity upgrade rather than a frame-rate rescue for smoothness-first players. ### The Longer Horizon: Waiting for PC and Next-Gen Silicon Rockstar’s blunt admission underscores a broader inflection point for ultra-ambitious AAA development, arriving in the same launch week that veteran designer Warren Spector argued *GTA 6*’s scale may prove [unsustainable as an industry template](https://www.mushimu.com/post/9651). Across the current console generation, *GTA 6* joins a small handful of deeply simulation-heavy titles where CPU complexity—not shader load—dictates the ceiling. For players determined to experience *GTA 6* at 60 FPS, 120 FPS, or unlocked variable refresh rates, Chambers’ remarks leave little ambiguity: no firmware tweak or post-launch PS5 Pro patch is coming to rewrite Zen 2 math. True high-frame-rate play in Leonida will wait for Rockstar’s eventual PC release—and, inevitably, the next generation of console hardware. ### Sources & Further Reading - [Kotaku (US — Original Rockstar Statement)](https://kotaku.com/gta-6-wont-get-a-60-fps-mode-even-on-ps5-pro-because-its-not-powerful-enough-2000755979) - [Eurogamer (UK)](https://www.eurogamer.net/ps5-pro-not-powerful-enough-gta-6-60fps) - [Clarín (Argentina)](https://www.clarin.com/estados-unidos/gta-6-modo-60-40-fps-rockstar-explica-bloqueado-30-fps-ps5-xbox_0_pIqAAoQ5Z7.html) - [Beebom (India)](https://beebom.com/ps5-pro-not-powerful-enough-to-handle-gta-6-at-60-fps/) ~Gaming ~GTA6 ~PS5Pro ~RockstarGames ~ConsoleGaming
Rockstar Says Even PS5 Pro Is 'Not Powerful Enough' to Run GTA 6 at 60 FPS or 40 FPS
0
Discuss 11
After nearly eight years of legal maneuvering and two months of courtroom testimony, the landmark U.S. federal criminal trial against **Huawei Technologies** and three affiliated subsidiaries is moving into jury deliberations in Brooklyn, setting the stage for a verdict with major legal and geopolitical consequences for U.S.–China technology relations. Presided over by U.S. District Judge Ann Donnelly in the Eastern District of New York under docket `1:18-cr-00457`, closing arguments concluded on Tuesday, November 17, 2026, with Judge Donnelly charging the jury ahead of deliberations on Wednesday, November 18, according to [MLQ.ai](https://mlq.ai/news/huawei-jury-to-begin-deliberating-in-brooklyn-after-two-days-of-closing-arguments-on-racketeering-bank-fraud-and-iran-sanctions-charges/), [AsiaOne / Reuters](https://www.asiaone.com/money/huawei-says-us-put-sinister-spin-business-activities-failed-prove-charges-trial), and [Dataconomy](https://dataconomy.com/2026/11/18/huawei-calls-us-criminal-case-a-fictionalised-portrait/). Huawei faces **12 remaining felony counts**, including racketeering conspiracy (RICO), conspiracy to steal trade secrets, bank and wire fraud, violations of U.S. sanctions on Iran under the International Emergency Economic Powers Act (IEEPA), money laundering, and obstruction of justice. ### Inside the Two-Month Trial: Over 60 Witnesses vs. 15 Why is a prosecution born during the first Trump administration reaching a jury only now, five years after former Chief Financial Officer Meng Wanzhou returned to China? The timeline reflects both the complexity of corporate RICO litigation and a pivotal pretrial ruling in June 2026. The case erupted in December 2018 when Meng—the daughter of Huawei founder Ren Zhengfei—was arrested in Vancouver on a U.S. warrant over alleged bank fraud tied to Huawei’s Iranian affiliate Skycom. In 2021, Meng signed a Deferred Prosecution Agreement (DPA) admitting she had misled HSBC executives about Huawei’s Iran dealings; U.S. charges against her personally were later dismissed, leaving the corporate entities as defendants. In June 2026, however, Judge Donnelly ruled that Meng’s DPA admissions could be introduced as evidence against the company itself, paving the way for jury selection and opening testimony in September 2026. Over the ensuing two months, federal prosecutors called more than 60 witnesses, while Huawei’s defense team called roughly 15. ### Prosecution’s RICO Thesis vs. Huawei’s "Fictionalised Portrait" Defense During roughly six hours of closing arguments on Monday, Assistant U.S. Attorney Meredith Arfa urged jurors to view the evidence not as isolated corporate missteps, but as an integrated two-decade criminal enterprise built on "theft, lies, and cover-ups." Prosecutors pointed to alleged intellectual-property thefts targeting five U.S. competitors—including surveillance footage of a Huawei engineer removing part of T-Mobile’s automated smartphone-testing robot "Tappy" inside a secure lab, as well as disputed incidents involving Cisco Systems router code, Quintel antenna concepts, and Fujitsu. To tie those episodes into a corporate-wide racketeering pattern, the government cited a 2013 internal memo and bonus program allegedly rewarding staff for obtaining competitors’ confidential information, alongside Meng’s 2021 admissions as "indisputable evidence" of defrauding HSBC and other lenders over sanctioned Iran business. Delivering a five-hour closing argument for Steptoe on Tuesday, lead defense counsel **Brian Heberlig** attacked the government’s narrative as a "fictionalised portrait" that "began with guilt and worked backwards." Highlighting Huawei’s massive scale—$126 billion in 2025 revenue and roughly 213,000 employees worldwide—Heberlig argued that prosecutors cherry-picked a handful of historical infractions committed by low-level managers or rogue staff who were subsequently fired or disciplined. On Iran, he contended that global banks knew Huawei operated there and competed for its business anyway while accepting the compliance risk, and he maintained that employee departures from the United States reflected routine corporate downsizing rather than obstruction of justice. Courtroom friction flared sharply over Meng’s 2021 statements. When Heberlig sought to frame Meng’s words as made merely "to the best of her knowledge and belief" rather than deliberate deception, prosecutors objected twice—and Judge Donnelly sustained both objections, while Assistant U.S. Attorney Robert Pollack told jurors in rebuttal that defense counsel was shouting because the facts and law weighed against Huawei. ### Geopolitical Timing Amid the Shenzhen APEC Summit The Brooklyn jury begins its work at a delicate diplomatic juncture. Even as U.S. and Chinese leaders engage in high-stakes talks during the [Shenzhen APEC summit](https://www.mushimu.com/post/9515) alongside [escalating U.S. scrutiny over AI chip export controls](https://www.mushimu.com/post/9653), Beijing has consistently denounced the Brooklyn prosecution as politicized "economic bullying" and technological containment rather than neutral law enforcement. Because the defendants are corporate entities rather than individual executives in U.S. custody, any guilty verdict would translate into massive potential fines, criminal forfeiture, probation monitorships, and reinforced legal justification for Washington’s telecommunications and semiconductor blacklists—rather than prison sentences. Conversely, an acquittal—or a hung jury on the flagship RICO count—would deal a high-profile courtroom setback to the Justice Department’s China Initiative-era corporate enforcement doctrine just as trade and technology negotiations resume at the highest level. ### Sources & Further Reading - [MLQ.ai — Brooklyn Trial Closing Arguments & Docket Summary](https://mlq.ai/news/huawei-jury-to-begin-deliberating-in-brooklyn-after-two-days-of-closing-arguments-on-racketeering-bank-fraud-and-iran-sanctions-charges/) - [AsiaOne / Reuters (Singapore)](https://www.asiaone.com/money/huawei-says-us-put-sinister-spin-business-activities-failed-prove-charges-trial) - [Dataconomy — Huawei Closing Defense Details](https://dataconomy.com/2026/11/18/huawei-calls-us-criminal-case-a-fictionalised-portrait/) ~WorldNews ~Huawei ~UnitedStates ~China ~Cybersecurity
Huawei Federal Criminal Trial Enters Jury Deliberations in Brooklyn After Two-Month Courtroom Battle
0
Discuss 11
Al menos **80 centros educativos** de las cuatro provincias de Galicia —con especial virulencia en A Coruña, donde la Policía Nacional contabilizó hasta 31 colegios e institutos evacuados— sufrieron este miércoles, 18 de noviembre de 2026, una oleada coordinada de falsas amenazas de bomba recibidas de forma casi simultánea por correo electrónico. La alerta obligó a desalojar a miles de alumnos, cortar calles céntricas y desplegar unidades de la Policía Nacional, Guardia Civil, Policía Local, guías caninos y especialistas en desactivación de explosivos (**TEDAX**) antes de confirmarse que no existía artefacto alguno. Según el balance difundido en torno a las 20:00 horas por la Consellería de Educación de la Xunta de Galicia y recogido por [La Voz de Galicia](https://www.lavozdegalicia.es/noticia/educacion/2026/11/18/amenazas-bomba-80colegios-gallegoshabia-gente-asustada-llorando/00031795016357695515868.htm), [El Español](https://www.elespanol.com/quincemil/actualidad/galicia/20261118/falsas-amenazas-bomba-colegios-galicia-directo-gran-despliegue-policial-veintena-centros-coruna/1003744463043_10.html) y [20minutos / Europa Press](https://www.20minutos.es/galicia/desalojos-revisiones-unos-80-colegios-galicia-por-falsos-avisos-bomba_7062881_0.html), los mensajes empezaron a entrar en las bandejas de secretaría y dirección entre las 11:50 y las 13:00 horas con una redacción intimidatoria prácticamente calcada: *«Minamos el edificio... Si desea salvar la vida de los estudiantes, abandone el edificio inmediatamente»*. ### El mapa del despliegue y la grieta del filtro de correo no deseado La mayor presión policial se concentró en A Coruña y su área metropolitana. Centros históricos y de gran tamaño como Zalaeta, Compañía de María, Franciscanos, Ramón de la Sagra, Otero Pedrayo, Rosalía Mera, Liceo La Paz, Santa María del Mar, Esclavas, Dominicos, Peñarredonda o Labaca, además del CEIP Vila de Rutis en Culleredo, activaron sus protocolos internos tras recibir el correo. En el colegio Compañía de María —donde cursan unos 1.200 estudiantes y trabaja alrededor de un centenar de docentes— la evacuación se completó en pocos minutos por grupos de edad: los mayores fueron conducidos a la explanada y el pabellón de Riazor, mientras los más pequeños quedaron resguardados en el Fogar de Santa Margarida y Aspronaga hasta que los TEDAX levantaron el precinto a las 14:30 horas. En el CEIP Zalaeta, los niños de menor edad aguardaron bajo soportales cercanos entre escenas de nerviosismo y llantos de familias que acudían a recogerlos. Aunque la Jefatura Superior de Policía confirmó ya a las 13:45 horas que las inspecciones habían descartado explosivos en los edificios revisados, el episodio dejó al descubierto una vulnerabilidad operativa inesperada: **los filtros antispam**. En varios colegios, entre ellos el Eusebio da Guarda de la plaza de Pontevedra, el correo amenazante quedó retenido en la carpeta de correo no deseado. Los responsables del centro no lo detectaron hasta cerca de las tres de la tarde, momento en el que ordenaron un desalojo preventivo y lamentaron no haber recibido un aviso directo previo de la Policía ni de la propia Consellería, de modo que fue la AMPA quien alertó de urgencia a las familias para recoger a los menores. Fuera de la capital coruñesa, la Jefatura Superior de Policía confirmó avisos en al menos seis centros de Santiago de Compostela, tres en Lugo, dos en Viveiro, dos en Pontevedra —incluido el IES Sánchez Cantón— y uno en Monforte de Lemos. A esa relación se sumaron las intervenciones de la Guardia Civil en un centro de Ribadeo y en otros cuatro de la provincia de A Coruña, además de incidentes en As Pontes y A Baña. La Xunta advirtió de que el recuento cercano a los 80 centros podría variar si algún equipo directivo revisa buzones secundarios o carpetas filtradas durante la mañana del jueves, aunque confirmó que las clases se reanudarán con total normalidad. ### Un patrón europeo bajo investigación policial La Delegación del Gobierno en Galicia y la Policía Nacional investigan ahora el origen técnico de los envíos masivos, mientras la Consellería de Educación expresó su preocupación y reclamó «que las autoridades las investiguen para que se apliquen las medidas correspondientes a los autores de estas amenazas». El detalle clave para los investigadores reside en la simultaneidad de los correos y en una sintaxis que replica casi palabra por palabra campañas de falsas amenazas registradas en fechas recientes contra centros escolares de **Italia, Irlanda y el Reino Unido**, además de otros puntos de España. Este tipo de envíos automatizados suele apoyarse en listas públicas de correos institucionales extraídas de portales educativos autonómicos y en servidores de correo temporales o comprometidos, buscando maximizar el colapso de las fuerzas de seguridad y el pánico social con un coste técnico mínimo para el autor. La coincidencia temporal con otras incidencias de seguridad urbana en España —como la octava alerta en menos de una semana en Córdoba por un paquete sospechoso en la avenida de Guerrita que movilizó igualmente a los TEDAX este miércoles— ha multiplicado la sensibilidad social en torno a los avisos por explosivos. ### Consecuencias legales y el reto de la comunicación escolar Más allá de la investigación policial para identificar al remitente —que se enfrentaría a posibles delitos de desórdenes públicos y simulación de peligro con movilización de servicios de emergencia bajo el Código Penal español—, la jornada deja dos lecciones prácticas para la administración educativa gallega. Por un lado, los protocolos de evacuación ejecutados por los claustros evitaron aglomeraciones descontroladas y permitieron inspecciones rápidas en menos de una hora en la mayoría de los recintos. Por otro, la brecha vivida en centros donde el aviso acabó oculto en la carpeta de *spam* evidencia la necesidad de contar con un canal de alerta inmediata y bidireccional entre las fuerzas de seguridad, la Consellería de Educación y las direcciones escolares cuando se detecte una campaña masiva dirigida contra un mismo dominio o provincia. ### Fuentes consultadas - [La Voz de Galicia (Galicia / España)](https://www.lavozdegalicia.es/noticia/educacion/2026/11/18/amenazas-bomba-80colegios-gallegoshabia-gente-asustada-llorando/00031795016357695515868.htm) - [El Español — Quincemil (España)](https://www.elespanol.com/quincemil/actualidad/galicia/20261118/falsas-amenazas-bomba-colegios-galicia-directo-gran-despliegue-policial-veintena-centros-coruna/1003744463043_10.html) - [20minutos / Europa Press (España)](https://www.20minutos.es/galicia/desalojos-revisiones-unos-80-colegios-galicia-por-falsos-avisos-bomba_7062881_0.html) ~Spain ~Galicia ~Educacion ~Seguridad ~ACoruna
Al menos 80 colegios de Galicia reciben falsos avisos de bomba simultáneos y obligan a desplegar a los TEDAX
0
Discuss 4
Only **4.1%** of games released on Steam over the past five years also have a console version—yet that tiny multiplatform slice accounts for **69.8% of all estimated Steam gross revenue**, according to a sweeping quantitative study published by developer Venn Studios and highlighted by Japanese trade outlet [Automaton West](https://automaton-media.com/en/news/survey-suggests-only-4-1-of-steam-games-have-console-versions-but-they-account-for-nearly-70-of-revenue/) on November 18, 2026. Examining **84,040** Steam games first released between November 5, 2021, and October 4, 2026, via Gamalytic revenue snapshots, Wikidata platform mappings, and a stratified 60-game manual audit, the [Venn Studios market intelligence report](https://venn-studios.com/research/steam-cross-platform/) exposes just how sharply Steam’s commercial center of gravity diverges from its sheer catalog volume. When read alongside fresh 2026 SteamDB release telemetry reported by [GamingOnLinux](https://www.gamingonlinux.com/2026/11/steam-is-officially-drowning-in-game-releases/), the numbers explain why Valve’s storefront is simultaneously breaking revenue records at the top while buckling under build-review backlogs at the bottom. ### The 4.1% Funnel: Why Console Ports Track Commercial Scale Venn Studios cross-referenced 202,787 Wikidata app-to-platform links against its 84,040-title Steam cohort as of November 11, 2026, matching 64.8% of games to Wikidata records and keeping the remaining 29,552 unmatched entries in the denominator rather than assuming them away. Even after adjusting for Wikidata’s incomplete coverage using a hand-checked 60-game pilot, Venn estimates that roughly 4% to 5% of Steam titles reach any given console family—about **4.7%** on PlayStation 4/5, **3.8%** on Xbox One/Series, and **3.7%** on Nintendo Switch and Switch 2—while mobile crossovers sit near or below 1% (0.54% measured on iOS and 0.49% on Android). Yet revenue concentrates dramatically inside that narrow multiplatform band. Broken down by estimated Steam-only gross earnings, console availability climbs cleanly at every tier: - **Under $10,000:** 1.0% have a confirmed console version. - **$10,000 to $100,000:** 8.5% appear on consoles. - **$100,000 to $1 million:** 21.0% have console releases. - **$1 million to $10 million:** 45.0% ship on consoles. - **$10 million and above:** 71.6% are multiplatform console releases. As Venn Studios summarizes the causal direction, ports generally follow commercial traction or major multiplatform publisher budgets rather than rescuing struggling PC launches on their own. Platform preferences also split along budget lines: among breakout hits above $10 million in the hand-checked sample, PlayStation (7 of 10) and Xbox (8 of 10) dominate—accounting for **65.6%** and **60.6%** of revenue-weighted Steam share respectively—whereas Nintendo Switch narrowly leads in the $10,000–$100,000 indie bracket (6.3% on Wikidata versus 5.6% for PlayStation and 4.5% for Xbox, while capturing 29.8% of overall revenue share). ### The Other Half of the Ratio: 27,286 Releases and Steam’s AI Flood Why has the percentage of multiplatform Steam releases stayed so low even as PC-console simultaneous launches have become standard for AA and AAA publishers? New 2026 SteamDB data points directly to the denominator: Steam’s open-door Steam Direct pipeline is absorbing an unprecedented wave of PC-only releases. According to [GamingOnLinux’s analysis of SteamDB telemetry](https://www.gamingonlinux.com/2026/11/steam-is-officially-drowning-in-game-releases/), Steam recorded **27,286 game releases so far in 2026**—already shattering 2025’s full-year record of 21,185 with more than a month still left on the calendar, and peaking at **3,738 releases in a single month** (up from 2,415 last year). Crucially, games carrying mandatory generative-AI disclosures jumped nearly threefold year-over-year, from **4,682 in 2025 to 13,445 in 2026**. That flood is now straining Valve’s own human operations. In a recent developer notice, Valve acknowledged that Steam build reviews—normally completed within three to five business days—are currently taking up to **12 business days**, citing repeated re-reviews. Meanwhile, console ecosystems still enforce certification fees, hardware devkits, age-rating compliance, and platform-holder curation—barriers that filter out almost all low-budget, hobbyist, and AI-assisted asset-flip launches before they ever reach PlayStation, Xbox, or Nintendo storefronts. ### Methodological Caveats and What It Means for Developers Independent analysts should read both datasets with clear eyes. Venn Studios explicitly notes that Gamalytic revenue numbers are third-party estimates of Steam-only gross sales—not official Valve accounting—and exclude console and mobile earnings. Moreover, counting older 2021–2023 cohorts alongside 2026 releases naturally gives earlier hits more years to ship post-launch console ports, while leaving 35.2% of obscure titles unmatched in Wikidata makes the 4.1% headline a strict lower bound. Even with those caveats, the structural divergence is unmistakable. Steam’s openness remains unmatched for discoverability experiments, Steam Deck indies, and PC-native genres like strategy or simulation. Yet for studios chasing sustainable mid-tier or breakout revenue—and for players trying to navigate an algorithmic storefront swamped by zero-review shovelware—the market reality of 2026 is stark: Valve’s catalog volume is overwhelmingly PC-exclusive, while nearly 70 cents of every dollar spent on Steam still flows to the small fraction of games built to clear console standards. ### Sources & Further Reading - [Automaton West (Japan)](https://automaton-media.com/en/news/survey-suggests-only-4-1-of-steam-games-have-console-versions-but-they-account-for-nearly-70-of-revenue/) - [Venn Studios — Steam Cross-Platform Market Intelligence Report](https://venn-studios.com/research/steam-cross-platform/) - [GamingOnLinux — SteamDB 2026 Release Volume Analysis](https://www.gamingonlinux.com/2026/11/steam-is-officially-drowning-in-game-releases/) ~Gaming ~Steam ~Valve ~GameDev ~PCGaming
Only 4.1% of Steam Games Have Console Versions—Yet They Drive 69.8% of Platform Revenue
0
Discuss 8
House Democrats have opened a formal congressional investigation into whether U.S. Commerce Secretary Howard Lutnick used his cabinet authority over semiconductor export controls to shield Cantor Fitzgerald—the Wall Street firm he led for decades before transferring ownership to trusts for his adult children—from hundreds of millions of dollars in potential Nvidia-related losses amid an expanding China chip-smuggling scandal. In a five-page letter dated November 17 and made public on November 18, 2026, three senior committee ranking members—Rep. Gregory Meeks (Foreign Affairs), Rep. Jamie Raskin (Judiciary), and Rep. Sydney Kamlager-Dove (Foreign Affairs Subcommittee on South and Central Asia)—demanded sweeping trading records, Southeast Asia client files, and internal communications from Cantor Fitzgerald Chairman Brandon Lutnick, the secretary’s son. According to [Al Jazeera](https://www.aljazeera.com/economy/2026/11/18/us-lawmakers-launch-probe-into-lutnicks-actions-amid-nvidia-scandal) and [The Guardian](https://www.theguardian.com/us-news/2026/nov/18/howard-lutnick-nvidia-cantor-fitzgerald), the lawmakers asked directly whether Secretary Lutnick "seriously and perhaps irreversibly" compromised U.S. national security to protect his family firm’s financial exposure. ### Following the Money Behind the Megaspeed Delay The inquiry marks an escalation from earlier [Senate oversight summonses](https://www.mushimu.com/post/9345) and [Justice Department scrutiny of Singapore cloud operator Megaspeed](https://www.mushimu.com/post/9484) by targeting the financial plumbing around the secretary’s family firm. Before Howard Lutnick’s Senate confirmation in February 2025, Cantor Fitzgerald had disclosed holding more than 3.5 million Nvidia shares, valued at over $1.3 billion. Upon taking office, Lutnick divested personal stakes and transferred ownership into trusts benefiting his adult children. According to the [House Foreign Affairs Committee release](https://democrats-foreignaffairs.house.gov/2026/11/ranking-members-meeks-raskin-and-kamlager-dove-probe-whether-lutnicks-family-finances-shaped-his-response-to-the-nvidia-spy-scandal/), Cantor Fitzgerald cut its reported Nvidia shareholdings by nearly half between September 2025 and March 2026—precisely the window during which White House and U.S. intelligence officials were examining allegations that China’s Ministry of State Security recruited former Megaspeed chief executive Alice Huang in an operation involving former Nvidia CEO Jensen Huang. Even after that reduction, Cantor Fitzgerald still held roughly $310 million in Nvidia stock at the end of March 2026. Lawmakers are also scrutinizing a separate policy front: efforts by Lutnick to narrow proposed U.S. restrictions on overseas data centers using advanced Nvidia processors to serve Chinese customers, including a Malaysian data center project tied to a Cantor Fitzgerald client. With the [119th Congress letter](https://democrats-foreignaffairs.house.gov/2026/11/ranking-members-meeks-raskin-and-kamlager-dove-probe-whether-lutnicks-family-finances-shaped-his-response-to-the-nvidia-spy-scandal/) demanding a complete log of Cantor Fitzgerald’s Nvidia trades since January 2024, records of any Nvidia investments linked to President Donald Trump, and preservation of encrypted and personal-device messages by December 1, the committee is probing whether financial timing overlapped with regulatory restraint. ### The Commerce Department’s Gatekeeper Power and Cabinet Friction At the center of the policy dispute sits the structural power of the Commerce Department’s Bureau of Industry and Security (BIS). While intelligence agencies, the Treasury Department, and the Justice Department can investigate illicit chip diversion, BIS holds the legal pen on adding foreign firms to the Entity List—cutting them off from U.S.-origin semiconductors and software. Earlier this month, Treasury Secretary Scott Bessent publicly broke ranks by stating that four federal agencies had favored blacklisting Megaspeed for months while Lutnick’s Commerce Department held off. That rare public cabinet rift, combined with Reuters reporting that senior White House aides have discussed whether Lutnick should step aside, underscores how export-control enforcement has shifted from routine trade compliance into high-stakes national security politics. The controversy has unfolded even as Nvidia’s board ruled out reinstating Jensen Huang this week while halting China H200 shipments in its [blockbuster Q3 earnings report](https://www.mushimu.com/post/9508). ### Cantor Fitzgerald’s Swap-Hedge Defense and Congressional Leverage Both the Commerce Department and Cantor Fitzgerald forcefully reject any suggestion of impropriety. A Commerce spokesperson insisted Lutnick has complied fully with ethics rules, divestitures, and recusal mandates across Cantor Fitzgerald, Newmark, and BGC. Cantor Fitzgerald told The Guardian and Reuters that the Democrats’ allegations are "inaccurate," arguing the firm holds no material proprietary bet on Nvidia. Instead, the firm says its reported holdings stem from client-driven equity swap hedges, and that the early-2025-to-2026 decline reflected clients unwinding swap positions rather than an executive decision to dump shares ahead of a regulatory crackdown. Whether that institutional defense closes the matter hinges on two factors: verifiable trade-level timestamps and political control on Capitol Hill. In prime brokerage and derivatives desks, delta-hedging client equity swaps routinely places large stock blocks on a dealer’s books without giving the dealer directional upside—yet proving that distinction requires inspecting the counterparty ledgers and client communications Democrats are now demanding. Because House Democrats remain in the minority during the lame-duck 119th Congress, they currently lack unilateral subpoena power unless Republicans join the inquiry or oversight carries into the next congressional term following the November 2026 midterms. Until those records are produced, the collision between Wall Street family trusts and U.S. AI export-control credibility remains unresolved. ### Sources & Further Reading - [House Foreign Affairs Committee — Ranking Members' Official Release](https://democrats-foreignaffairs.house.gov/2026/11/ranking-members-meeks-raskin-and-kamlager-dove-probe-whether-lutnicks-family-finances-shaped-his-response-to-the-nvidia-spy-scandal/) - [Al Jazeera / Reuters (Qatar / International)](https://www.aljazeera.com/economy/2026/11/18/us-lawmakers-launch-probe-into-lutnicks-actions-amid-nvidia-scandal) - [The Guardian (UK)](https://www.theguardian.com/us-news/2026/nov/18/howard-lutnick-nvidia-cantor-fitzgerald) ~WorldNews ~UnitedStates ~Nvidia ~ExportControls ~Politics
House Democrats Probe Commerce Secretary Howard Lutnick and Cantor Fitzgerald Over Nvidia Export-Control Decisions
0
Discuss 9
miHoYo’s in-house Honkai AI R&D Team, working alongside researchers from Tsinghua University, the National University of Singapore, the University of Hong Kong, and the University of Chinese Academy of Sciences, has unveiled **Gamer**, an open-weight Vision-Language-Action foundation model capable of playing complex 3D and 2D games—including *Honkai: Star Rail*—from start to finish using only raw screen pixels and standard mouse and keyboard inputs. Published on arXiv on November 13, 2026, and reported across Japanese specialist press by [Automaton West](https://automaton-media.com/en/news/mihoyo-develops-ai-model-that-can-play-complex-games-like-honkai-star-rail-just-like-a-human-would-with-mouse-and-keyboard-inputs/) on November 18, the project stands out because it touches zero lines of backend game code. Unlike conventional game bots that read internal memory states, object coordinates, or studio-exposed APIs, Gamer operates strictly through the same visual and physical loop available to human players: it ingests 5 Hz screen captures and emits mouse displacements, scroll events, and keyboard state blocks. ### Breaking the Real-Time Latency Wall Why have general-purpose multimodal models struggled so badly in live action games? On paper, frontier vision-language models already score well on static screenshots and turn-based reasoning. In practice, however, benchmarks published on the [Gamer project page](https://mihoyo-gamer.github.io/) show that switching from a "time-paused" evaluation—where the game engine freezes while the model thinks—to real-time execution causes general-purpose models to lose 17 to 35 percentage points on average across suites like MCU, ViZDoom, OmniGameArena, and GameWorld. By the time a standard cloud LLM finishes decoding a multi-token response, the enemy has moved or the timing window has closed. Gamer resolves that bottleneck across both its action representation and its serving stack. Rather than emitting one coarse action per screenshot or relying on a separate continuous control head, the model takes one screenshot every 200 milliseconds (5 Hz) and autoregressively predicts a mouse displacement vector alongside six consecutive 33-millisecond key-state blocks. That design yields an effective 30 Hz control cadence directly inside the language model’s token stream. To keep inference inside that 200 ms budget, the team built **Streaming-vLLM**, an optimized serving engine hosted alongside the [official GitHub repository](https://github.com/RainBowLuoCS/Gamer). By pinning each client’s KV cache across turns, deploying architecture-specific kernels, and dispatching verified action blocks early before the full response finishes decoding, Streaming-vLLM slashes end-to-end request latency on an NVIDIA H200 running the `Qwen3.5-35B-A3B` Mixture-of-Experts backbone from 1,306 ms down to 65.12 ms—a 20.1x speedup that lifts theoretical control capacity from 4.62 Hz to 92.16 Hz. Under real-time evaluation, `Gamer-35B-A3B-CPT` actually improves by 3.4 percentage points over time-paused scoring rather than collapsing. ### How a Pixel-Only Agent Survives Long-Horizon RPG Quests Playing *Honkai: Star Rail* requires far more than reflex timing. Players must follow shifting quest objectives, talk to NPCs, track failed attempts, navigate multi-room zones, and clear logic or rhythm mini-games over long sessions that easily blow past a single context window. According to the [arXiv technical report](https://arxiv.org/abs/2611.18397) and its companion [Hugging Face Paper page](https://huggingface.co/papers/2611.18397), the 27-author team tackled this through a three-stage training pipeline built on `Qwen3-VL-8B` and `Qwen3.5-35B-A3B` backbones: - **Stage I — Continued Pre-Training (CPT):** The team trained on more than 10,000 hours of gameplay video paired with synchronized device-level keyboard and mouse logs. Two subtle training tricks prevent naive imitation collapse: a repeat-aware loss decay ($\gamma = 0.90$) that keeps long key-holds (like holding `W` to sprint) from drowning out crucial action transitions, and random key remapping per trajectory so the policy learns spatial intent rather than memorizing hardcoded bindings. - **Stage II — Sparse Reasoning and Cross-Context Memory (SFT):** Using roughly 1,000 hours of curated *Honkai: Star Rail* footage, the researchers added around 1.2 million "Think" annotations and 223,000 structured memory summaries across 63,000 activity segments. Crucially, Gamer does not burn tokens thinking on every frame; it triggers sparse reasoning only at route shifts, target switches, new prompts, or task transitions, while folding completed objectives and failed attempts into persistent memory summaries as older frames age out of context. - **Stage III — Online Reinforcement Learning (TA-GRPO):** Finally, Temporal-Aware Group Relative Policy Optimization refines memory-linked rollout segments against environment-verified rewards without requiring a separate value critic or learned reward model. On 40 held-out *Honkai: Star Rail* mini-game tasks spanning seven distinct puzzle types, `Gamer-RL` achieved a 49.7% macro-average success rate—solving tasks across all seven categories and improving by 10.5 percentage points over supervised fine-tuning alone (39.2%), while nearly tripling the strongest general-purpose baseline (`GPT-6.1-Sol`) at 17.1%. ### Automated QA vs. Live-Service Anti-Cheat Reality Open-sourcing an agent that plays a live-service gacha hit strictly through mouse and keyboard inputs naturally raises eyebrows among players worried about farming bots and automated account progression. Both the researchers and independent observers point to two constraints that separate Gamer from consumer cheat tools. First, the latency numbers that make real-time play viable were measured on datacenter NVIDIA H200 GPUs running MoE models, well beyond typical desktop hardware. Second, the team explicitly stated that it "does not endorse using Gamer for cheating, unauthorized access, exploiting security vulnerabilities, or any behavior that harms users or undermines fairness." For studios, the strategic upside sits elsewhere—a contrast to Western industry debates currently centered on generative asset pipelines (as recently examined in [Mushimu’s coverage of Ben Feder’s remarks](https://www.mushimu.com/post/9552)). By training an agent that perceives only rendered pixels and acts through standard OS inputs, miHoYo’s R&D division gains a general-purpose automated QA tester capable of verifying quest progression, catching geometry soft-locks, and benchmarking accessibility barriers across engine updates without writing bespoke test hooks for every game mode. Whether that pixel-only paradigm spreads from Chinese live-service powerhouses to broader studio QA pipelines will depend on how quickly inference costs fall from datacenter H200 clusters down to local workstation cards. ### Sources & Further Reading - [Automaton West (Japan)](https://automaton-media.com/en/news/mihoyo-develops-ai-model-that-can-play-complex-games-like-honkai-star-rail-just-like-a-human-would-with-mouse-and-keyboard-inputs/) - [arXiv Technical Report (2611.18397)](https://arxiv.org/abs/2611.18397) - [Gamer GitHub Repository](https://github.com/RainBowLuoCS/Gamer) - [Hugging Face Papers — Gamer (THUSI-Lab / Honkai AI R&D)](https://huggingface.co/papers/2611.18397) ~Gaming ~AI ~HonkaiStarRail ~miHoYo
miHoYo Unveils 'Gamer,' an Open-Weight AI Agent That Plays Honkai: Star Rail via Screen Pixels and Mouse Inputs
0
Discuss 11
On the eve of *Grand Theft Auto 6*'s global launch on November 19, 2026, *Deus Ex* and *System Shock* creator Warren Spector has offered one of the sharpest industry diagnoses of what Rockstar's milestone means—and why no other studio should try to repeat it. Writing on LinkedIn just days after his own 44-year development career concluded with his August 2026 retirement, Spector congratulated Rockstar's developers on what "looks like it will be a high point of sorts for our medium and our business," while warning plainly: **"I just hope GTA 6 is the last of its kind."** ### Why a Billion-Dollar Hit Still Worries Veteran Builders Spector's critique, covered across PC Gamer, French gaming outlet *Big-N*, and Brazilian outlet *GameVicio*, focuses on capital concentration during an industry-wide employment contraction. Pointing to *GTA 6*'s estimated billion-dollar development price tag—even before counting what he termed "an undoubtedly crazy marketing budget"—Spector argued that Take-Two and Rockstar have built an outlier business model that the wider industry neither can nor should emulate. > "That's especially true when thousands of developers are out of work. A billion dollars pays for a lot of games and a lot of salaries. Personally I'd rather have hundreds of games and full employment than one game no matter how amazing it is." Neither Rockstar nor Take-Two has officially disclosed *GTA 6*'s final production budget, though analysts project the title will cross $2 billion in opening-weekend revenue alone after surpassing $1.4 billion in pre-orders. Yet Spector's arithmetic reflects a widening structural split in 2026: only a handful of mega-publishers with multi-year cash reserves can sustain near-decade development cycles across more than 4,700 staff, while mid-sized and AAA studios elsewhere have endured repeated rounds of layoffs and cancellations—including recent cuts at Firaxis Games and Crystal Dynamics. ### The Paradox of Underfunded Failures Crucially, Spector did not dismiss the craft or investment behind *GTA 6*; in fact, he argued Rockstar's success validates a principle publishers keep ignoring at the opposite end of the market. "Give a game the time and money to prove its potential (key word, potential) and you have a chance of making something great," he wrote. "Underfund a game or force it out the door before it's ready [guarantees] failure no matter how much potential it has." That observation carries personal weight. Prior to retiring in August 2026, Spector served as chief creative officer and creative director at OtherSide Entertainment on the stealth project *Thick as Thieves*, which shifted away from its original PvPvE scope before hitting early access and running short on time and funding. Across the broader sector, studios caught between those two poles—neither backed by a *GTA*-scale blank check nor insulated by lean indie budgets—are finding little room to survive mistakes. ### What Comes After the "High Point"? Spector's remarks also arrive alongside growing unease inside AAA publishing over how few companies can afford another bet of this scale. While some studio founders hope *GTA 6* opens the door to higher baseline software pricing, and former Take-Two executives acknowledge AAA blockbusters face mounting economic strain, even Rockstar's closest peers are pivoting toward safer franchise pipelines or heavier automation. If *GTA 6* becomes the industry's undisputed commercial peak of the decade—drawing hundreds of millions of hours away from competing titles on launch week while setting technical expectations few publishers can finance—Spector's warning lands as a timely reminder: a medium cannot thrive on a single billion-dollar monument if the rest of its creative ecosystem is hollowed out to keep up. ### Sources & Further Reading - [PC Gamer — 'I Just Hope GTA 6 Is the Last of Its Kind': Warren Spector Congratulates Rockstar While Warning Industry Can't Keep This Up](https://www.pcgamer.com/games/grand-theft-auto/i-just-hope-gta-6-is-the-last-of-its-kind-warren-spector-congratulates-rockstar-on-its-landmark-accomplishment-but-says-the-industry-cant-keep-this-up/) - [GameVicio (Brazil) — Criador de Deus Ex espera que GTA 6 seja "o último de seu tipo"](https://www.gamevicio.com/noticias/2026/11/warren-spector-gta-6-ultimo-seu-tipo/) - [Big-N (France) — Warren Spector espère que GTA 6 sera le dernier jeu de cette envergure](https://big-n.fr/actualites/64676-warren-spector-gta-6-dernier-jeu-envergure/) - [Eurogamer — Warren Spector Retirement Context and Career Overview](https://www.eurogamer.net/warren-spector-announces-retirement) ~Gaming ~GTA6 ~WarrenSpector ~Industry ~AAA
Warren Spector Praises GTA 6 as a Landmark Achievement—and Hopes It Is 'The Last of Its Kind' for AAA Budgets
0
Discuss 5
NVIDIA has shipped **Linux Graphics Driver 620.73.04**, the first maintenance update in its new **R620 Production Branch (Long-Lived Branch)** series for 64-bit `x86_64` and `aarch64` Linux systems as well as FreeBSD `x64`. Arriving exactly two weeks after the initial R620 public debut (`620.70.06`) on November 4, 2026, the point release carries an intentionally terse changelog—"Minor bug fixes and improvements"—signaling a rapid stabilization pass before distributions and Linux gaming setups roll forward onto the R620 tree. While the `.73.04` bump itself introduces no headline features, understanding why the R620 branch matters—and who should move to it versus wait for distro packaging—requires looking at what NVIDIA landed in `620.70.06` and how Production Branch releases slot into modern Wayland and Proton stacks. ### What Changed Between 620.70.06 and 620.73.04 As documented by Phoronix and Russian Linux portal *Comss.ru*, NVIDIA published `620.73.04` on November 18, 2026, across both its official `.run` installer packages and its updated **R620 README**. Unlike the `615.78.08` New Feature Branch or experimental developer drops, R620 sits in NVIDIA's Production Branch track—the branch targeted at sustained stability across consumer GeForce RTX cards (from RTX 50-series Blackwell down through RTX 20-series and GTX 16-series desktop and laptop GPUs), TITAN RTX, and workstation/datacenter accelerators, alongside Windows' matching NVIDIA RTX Enterprise R620 U2 (`621.24`). Two weeks earlier, `620.70.06` introduced the structural work that `620.73.04` now refines: - **Wayland Explicit Sync Throughput:** Optimized explicit synchronization overhead in high-framerate Wayland compositor benchmarks—an essential bottleneck on modern KDE Plasma, GNOME, and Gamescope sessions. - **Vulkan & Game Fixes:** Resolved Vulkan crashes and device-lost / corruption regressions in demanding Proton and native workloads, explicitly targeting *Assassin's Creed Shadows*, *Final Fantasy VII Rebirth*, and multi-GPU device-group rendering. - **OSD Telemetry & Kernel Stability:** Expanded `__GL_SHOW_GRAPHICS_OSD` visual indicator support to cover Vulkan backends alongside OpenGL, while addressing Linux kernel issues on specific NUMA topologies, installer edge cases, and virtual-terminal (`VT`) switching glitches. ### Why a Minimal Changelog Follows a Major Branch Launch Rapid, quiet `.04` point releases immediately after a fresh Production Branch `.06` drop typically reflect post-launch regression triage caught as early adopters move onto the branch across newer 6.x/7.x kernels and Wayland compositors. With Linux gaming increasingly running through Vulkan translation layers (DXVK, VKD3D-Proton) and Wayland explicit sync rather than legacy X11 paths, small synchronization or VT-switch defects surface quickly once a branch hits public mirrors. Noticeably, NVIDIA's official `620.73.04` README continues to ship both X11 GLX modules (`libglxserver_nvidia.so.620.73.04`) and Wayland external platform libraries (`libnvidia-egl-wayland` and `libnvidia-egl-gbm`), ensuring compatibility across both legacy Xorg desktops and Wayland-first gaming distributions. ### Should Linux Gamers and Desktop Users Install It Manually? For most desktop Linux users and Steam/Lutris players—especially those running immutable or rolling gaming distributions such as Bazzite, Nobara, Pop!_OS, openSUSE, or Arch—waiting for your distribution's signed NVIDIA repository packages remains strongly preferable to executing `sh ./NVIDIA-Linux-x86_64-620.73.04.run` directly. Distribution packages handle DKMS/open kernel module rebuilds, 32-bit multilib Vulkan/OpenGL libraries required by Steam and Wine, and Optimus/PRIME laptop power management cleanly without overwriting system libraries. Users who rely on standalone handheld-style eGPU docks, dual-GPU laptops, or bleeding-edge Wayland compositors and already migrated manually to `620.70.06` are the primary candidates for an immediate bump to `620.73.04`, picking up NVIDIA's latest stability sweep across the R620 Production Branch. ### Sources & Further Reading - [Phoronix — NVIDIA 620.73.04 Linux Driver Released With Bug Fixes](https://www.phoronix.com/news/NVIDIA-620.73.04-Linux) - [NVIDIA Official Documentation — Linux Graphics Driver 620.73.04 README & Installed Components](https://us.download.nvidia.com/XFree86/Linux-x86_64/620.73.04/README/index.html) - [Comss.ru — NVIDIA Graphics Driver 620.73.04 Linux: R620 Production Branch](https://www.comss.ru/page.php?id=22952) - [LinuxAdictos — NVIDIA 620.70.06 R620 Wayland and Vulkan Changes Overview](https://www.linuxadictos.com/nvidia-620-70-06-llega-a-linux-con-mejoras-para-wayland-y-vulkan.html) ~Linux ~NVIDIA ~LinuxGaming ~Wayland ~Drivers
NVIDIA Releases Linux Driver 620.73.04 in R620 Production Branch for Stability Following Wayland and Vulkan Overhaul
0
Discuss 9
Just hours before *Grand Theft Auto 6* unlocks globally on November 19, 2026, evidential hearings in Rockstar Games' landmark unfair-dismissal employment tribunal concluded in Glasgow on Wednesday—while a fresh cross-check of the game's launch credits confirmed that **at least 28 of the 34 employees fired in October 2025** are named in the final credits. Judge Lucy Wiseman wrapped up the Glasgow hearings after closing arguments from Rockstar's legal team on November 18, declining to set a fixed publication date for her written ruling. According to *People Make Games* reporter Chris Bratt (cited by VGC), the verdict could take anywhere from a few weeks to a few months—leaving the fired developers in limbo even as *GTA 6* arrives alongside a separate [1,600-worker UK union recognition ballot](https://www.mushimu.com/post/9599) running through November 26. ### Inside Rockstar's Closing Defense—and the Union Counter-Narrative Twenty-three of the 34 developers dismissed across the UK and Canada in October 2025 are pursuing claims through the **Independent Workers' Union of Great Britain (IWGB)**, alleging they were unlawfully targeted for trade-union organizing to deprive the union of workplace leverage. Rockstar—whose flagship Rockstar North studio sits on Holyrood Road in Edinburgh—has consistently denied union-busting, insisting the terminations were strictly for gross misconduct. In closing submissions reported by *The Independent*, Rockstar counsel **Thomas Cordrey** argued that the claimants were "caught red-handed" breaching strict confidentiality rules on a semi-public Discord server with hundreds of members, where moderators allegedly declined to "over police" discussions and "let the taps of disparagement and confidential information flow." Cordrey pointed to testimony from Rockstar head of licensing **Jennifer Kolbe**, who told the tribunal she was "shocked and appalled" by the server logs uncovered in October 2025, including messages allegedly referencing unreleased *GTA 6* features and launch timing while Rockstar stood on the "precipice of announcing further delays" that weighed heavily on parent company Take-Two's share price. Crucially, Cordrey also had to defend *how* Rockstar fired the workers: issuing termination letters directly without holding individual disciplinary hearings. He urged Judge Wiseman to rule that skipping "34 extended disciplinary hearings with multiple rounds across multiple jurisdictions" in the face of "inevitable dismissals for gross misconduct" fell "within the band of reasonable responses" under UK employment law, portraying the claimants as unapologetic and "blameworthy." The IWGB, by contrast, maintains that communications on the server were protected union activity and has accused Rockstar of infiltrating and surveilling a union-linked Discord to break organizing momentum. ### Why 28+ Fired Workers Still Made the GTA 6 Credits The human and structural paradox of the dispute came into sharper focus on November 18 when *Eurogamer* cross-referenced the IWGB's list of 34 dismissed employees against *Grand Theft Auto 6*'s sprawling end credits—a roster Rockstar recently disclosed encompasses **more than 4,711 developers across 12 global offices**, more than double the headcount of *Red Dead Redemption 2*. *Eurogamer* verified **28 direct name matches** in the launch credits, alongside several verified name discrepancies, close matches still being clarified, and a handful of omissions—meaning most, though not quite all, of the sacked staff received formal credit for years of work on a project analysts expect to generate $2 billion in opening-weekend revenue alone. Notably, the fired workers themselves had previously urged fans not to boycott *GTA 6* because of the years they invested in building it. ### What Happens Next After Launch Day For the 23 claimants awaiting Judge Wiseman's ruling, *GTA 6*'s commercial explosion offers little immediate closure. As one fired developer told *People Make Games* outside the hearings, the year-long wait since October 2025 has blurred into a "new depressing reality," with some former staffers having moved on to other jobs while others still hope a favorable tribunal judgment will enable reinstatement. Whatever Judge Wiseman decides on whether Rockstar's summary dismissals were legally "reasonable" or unlawful retaliation, the outcome lands directly atop a broader labor shift inside British AAA development. With 1,600 UK Rockstar staff actively voting on formal union recognition through November 26, the Glasgow verdict will serve as a decisive legal precedent for where UK employment tribunals draw the line between corporate NDA enforcement and protected collective organizing. ### Sources & Further Reading - [The Independent (UK) — Fired Rockstar Workers 'Caught Red-Handed' Flouting Rules, Tribunal Told](https://www.independent.co.uk/news/uk/home-news/grand-theft-auto-discord-iwgb-rockstar-canada-b3087430.html) - [Eurogamer — Most, Though Not Quite All, of the 34 Fired Rockstar Employees Are Named in the GTA 6 Credits](https://www.eurogamer.net/fired-rockstar-employees-named-gta-6-credits-tribunal) - [VGC — Rockstar's 'Union Busting' Tribunal Comes to an End, Pending Verdict](https://www.videogameschronicle.com/news/rockstars-union-busting-tribunal-comes-to-an-end-pending-verdict/) - [The Scotsman (Scotland) — Rockstar Games 'Did Not Want Union to Have Leverage or Power', Tribunal Told](https://www.scotsman.com/news/scottish-news/rockstar-games-did-not-want-union-to-have-leverage-or-power-tribunal-told-9356834) ~Gaming ~GTA6 ~RockstarGames ~Labor ~UKNews
Rockstar's Glasgow Employment Tribunal Wraps Hearings Pending Verdict as 28+ Fired Staff Appear in GTA 6 Credits
0
Discuss 3
With *Grand Theft Auto 6* unlocking globally at midnight on November 19, 2026, indie publisher Devolver Digital is taking the most direct—and absurd—route possible to counter-program Rockstar's 212 GB juggernaut: offering free Steam keys for its own November 19 release, ***Volvy's Adventure: Reslimed***, to players who record themselves deleting *GTA 6* from their consoles. Revealed via Devolver Digital's social campaign under the hashtag `#MakeRoomForVolvy` and covered across Eurogamer, Beebom, and GamerBraves alongside the game's live Steam storefront listing, the promotion turns a year-and-a-half-old running joke into a pointed indie-versus-AAA launch stunt. ### The Catch Inside the "#MakeRoomForVolvy" Trade The rules of the giveaway expose an immediate platform mismatch that Devolver itself cheerfully acknowledges. To enter, players are asked to repost the publisher's announcement alongside a video clip showing them deleting their *Grand Theft Auto 6* pre-load off their PlayStation 5 or Xbox Series X|S storage—or, for anyone who did not pre-order Rockstar's game, a video of themselves biting into a raw onion. Even if a console player sacrifices all 212 GB of their *GTA 6* installation—forcing a massive re-download right before midnight unlocks—they still cannot play the prize on that console. Developed by *Gato Roboto* and *Gunbrella* studio **doinksoft**, *Volvy's Adventure: Reslimed* is launching **exclusively on PC via Steam** on November 19, meaning Devolver is handing out Steam keys in exchange for console SSD space. Devolver paired the giveaway with a parody trailer mimicking Rockstar's reveal footage, starring its oversized-hand mouse mascot Volvy and "Miss Volvy" fleeing police and sharing a steamy shower scene—leaning into the publisher's fictionalized lore that *Volvy's Adventure: Reslimed* is a "much-anticipated remaster of the original remake of the unreleased action-platformer," despite actually being Volvy's first standalone video game. ### Why Indies Are Counter-Programming GTA 6 Instead of Fleeing Beneath the slapstick marketing lies a real economic calculation that has hung over November 2026 for the entire games industry. *Grand Theft Auto 6* has already generated more than $1.4 billion in pre-orders, prompted market analysts like Newzoo to warn that console software spending would shrink in 2026 without it, and pushed multiple competing publishers to clear out of late November entirely. Devolver Digital took the opposite bet back in May 2025, publicly vowing to ship a title on the exact day *GTA 6* launched—and delaying its own counter-programming slot alongside each of Rockstar's schedule shifts until landing on November 19, 2026. Rather than spending tens of millions of dollars on mainstream ad buys that would be drowned out by Leonida hype, a zero-budget stunt tied directly to *GTA 6*'s enormous 212 GB footprint guarantees visibility for a 2D retro precision platformer that would otherwise vanish in launch-day feeds. ### What doinksoft Actually Built for November 19 Strip away the parody and *Volvy's Adventure: Reslimed* is a compact, throwback 2D side-scrolling platformer built around physical slapstick mechanics: Volvy uses an oversized hand to helicopter across gaps, smash through roadblocks, and punch underground. According to its Steam store page, the single-player release supports 15 interface and subtitle languages, Steam Family Sharing, and launches with a 25% introductory discount on November 19. For most players who already spent weeks waiting on pre-loads, deleting 212 GB on launch eve is obviously a non-starter—which is precisely why the raw-onion backup clause exists. Still, by making *GTA 6*'s storage footprint the punchline of its launch trailer, Devolver managed to hijack a sliver of the biggest release week in gaming history without competing on Rockstar's budget at all. ### Sources & Further Reading - [Eurogamer — Delete GTA 6 From Your Hard Drive and Indie Publisher Devolver Digital Might Give You a Free Game](https://www.eurogamer.net/delete-gta-6-free-game-devolver-digital) - [Beebom — Devolver Digital Wants You to Delete GTA 6 for a Chance to Win a Free Game](https://beebom.com/devolver-digital-says-delete-gta-6-for-free-game/) - [Steam Store — Volvy's Adventure: Reslimed (Release Date: November 19, 2026)](https://store.steampowered.com/app/4523120/Volvys_Adventure_Reslimed/) - [GamerBraves — Devolver Digital Offers Free Copy of Volvy's Adventure to Players Who Delete Their GTA VI Pre-Load](https://www.gamerbraves.com/devolver-digital-offers-free-copy-of-volvys-adventure-to-players-who-delete-their-gta-vi-pre-load/) ~Gaming ~DevolverDigital ~GTA6 ~IndieGames ~Steam
Devolver Digital Challenges Players to Delete Their 212 GB GTA 6 Pre-Load for a Free PC Copy of Volvy's Adventure: Reslimed
0
Discuss 9
A once días de las elecciones generales del 29 de noviembre, la candidatura del Partido Popular por Barcelona que encabeza el alcalde de Badalona, Xavier García Albiol, ha colocado una serie de vallas publicitarias de gran formato en el área metropolitana barcelonesa que dirigen sus advertencias sobre la multirreincidencia delictiva tanto en castellano como en árabe. La iniciativa ha reabierto de inmediato el choque político y mediático en Cataluña sobre el uso electoral de la seguridad ciudadana y la estigmatización de la población inmigrante en plena pugna por el voto conservador. La campaña consta de tres formatos distribuidos en arterias clave de salida y acceso a la capital catalana. La valla más polémica, ubicada en la autopista C-33 a la altura de la Meridiana en dirección a Sabadell y Granollers, muestra el retrato de Albiol bajo el lema **«Hola, delincuentes. Se acabó entrar y salir. Nueva ley contra la multirreincidencia. Vota PP»**, traducido al árabe en la misma lona. Otras dos vallas escritas únicamente en castellano se centran en la vivienda: en la C-31 a su paso por Badalona se lee *«Hola, okupa, ¡48 horas y fuera! Nueva ley contra la okupación ilegal»*, mientras que en el acceso al aeropuerto de El Prat desde Sant Boi de Llobregat otro cartel apela a los viajeros con el eslogan *«Vuela tranquilo. Tu casa te espera a la vuelta»*. ### Qué busca el PP catalán y cómo defiende el uso del árabe Desde la dirección de campaña de los populares en Cataluña, su responsable Juan Fernández ha defendido abiertamente la estética y el tono de las lonas, calificándolas de mensajes «directos, provocadores y fácilmente reconocibles». Según el comunicado oficial recogido por *La Vanguardia*, *Cadena SER* y *La Voz de Galicia*, el objetivo del partido es situar «la defensa de la propiedad privada y la protección de los propietarios» entre las prioridades legislativas de la próxima legislatura, con el compromiso de aprobar en el Congreso reformas penales que permitan desalojar viviendas ocupadas en 48 horas y endurecer el código frente a delincuentes que acumulan detenciones sucesivas sin ingresar en prisión. La apuesta también persigue reforzar el perfil personal de Albiol como candidato por Barcelona, haciendo de la seguridad y la lucha contra la okupación sus «principales señas de identidad» en un territorio donde el PP compite directamente con Vox —y en el plano catalán con el espacio posconvergente e identitario— por capitalizar el malestar vecinal en el cinturón metropolitano. El propio Albiol ha insistido durante la campaña en que «no toda» la inmigración llega a delinquir, pero sostiene que existe «un problema de seguridad con una determinada inmigración», en sintonía con el giro del PP nacional a favor de los centros de retorno europeos en terceros países como Ruanda. ### Del «Limpiando Badalona» de 2015 al marco penal de 2026 El contraste editorial entre cabeceras nacionales y catalanas evidencia la carga simbólica del gesto. Mientras medios como *La Voz de Galicia* y *La Vanguardia* subrayan el eje programático de la propuesta contra la multirreincidencia y la ocupación ilegal —sin dejar de señalar que la inclusión del árabe refuerza el sesgo que asocia delincuencia y origen magrebí—, *elDiario.es* y *Cadena SER* enmarcan la valla en una estrategia continuada que remonta a hace más de una década en Badalona. En 2015, Albiol ya centró su campaña municipal en el eslogan «Limpiando Badalona», después de haber sido llevado a los tribunales por el reparto de panfletos que vinculaban a vecinos rumanos con la inseguridad —causa de la que terminó absuelto al no apreciarse delito de odio penalmente punible—. Quince años después de conquistar por primera vez la alcaldía badalonesa con aquel discurso de orden público, el salto de un eslogan ambiguo a una interpelación bilingüe en castellano y árabe dirigida a «los delincuentes» marca un escalón más explícito en la comunicación electoral del PP en Cataluña. ### Riesgo político y cálculo electoral en Barcelona La maniobra ilustra el dilema que afronta el Partido Popular en la recta final hacia el 29-N en circunscripciones urbanas como Barcelona. Por un lado, polarizar la agenda alrededor de la multirreincidencia y la okupación permite a Albiol disputar a Vox el electorado más sensible al discurso securitario y evitar fugas hacia la derecha radical en el área metropolitana. Por otro, la identificación directa entre idioma árabe y delincuencia expone a los populares a críticas frontales de izquierda, partidos nacionalistas y entidades sociales por xenofobia, complicando la imagen de centralidad moderada que Alberto Núñez Feijóo intenta proyectar en el debate cara a cara y en el resto de España. Con las encuestas otorgando en Barcelona nueve escaños aún disputados entre bloques, la campaña de las vallas sitúa la seguridad y la inmigración en el centro del pulso catalán de la última semana antes de las urnas. ### Fuentes consultadas - [elDiario.es — Albiol desempolva el manual xenófobo e instala carteles en los que se dirige en árabe a los «delincuentes»](https://www.eldiario.es/catalunya/albiol-desempolva-manual-xenofobo-e-instala-carteles-dirige-arabe-delincuentes_1_13673524.html) - [La Vanguardia — El PP lanza una campaña contra okupas y delincuentes también en árabe: «¡48 horas y fuera!»](https://www.lavanguardia.com/politica/20261118/11687377/pp-lanza-campana-publicitaria-okupas-delincuentes-48-horas-fuera.html) - [Cadena SER Catalunya — Albiol cuelga un cartel dedicado a los «delincuentes» en español y en árabe](https://cadenaser.com/cataluna/2026/11/18/albiol-cuelga-un-cartel-dedicado-a-los-delincuentes-en-espanol-y-en-arabe-sercat/) - [La Voz de Galicia — Albiol y el PP despliegan vallas en castellano y árabe para dar un «aviso a los delincuentes»](https://www.lavozdegalicia.es/noticia/espana/2026/11/18/albiol-pp-despliega-vallas-castellano-arabe-dar-aviso-delincuentes/00031795023090509630253.htm) ~España ~Cataluña ~PP ~Elecciones29N ~Inmigración
Albiol instala vallas en castellano y árabe contra «los delincuentes» en los accesos a Barcelona y reabre el choque por la inmigración ante el 29-N
0
Discuss 4
A once días de las elecciones generales del **29 de noviembre de 2026 (29-N)**, la batalla por la mayoría parlamentaria no se libra tanto en los grandes porcentajes nacionales como en unas pocas decenas de últimos escaños provinciales que bailan por décimas bajo la regla D'Hondt. El contraste entre los modelos probabilísticos publicados este miércoles por *El País*, *DatosRTVE* y *ElectoralLab* (recogido por *El Debate*) y los cálculos internos del Frente Amplio adelantados por *elDiario.es* dibuja una aritmética paradójica: mientras los sondeos otorgan al bloque de PP y Vox una probabilidad superior al **90 %** de superar los 176 diputados, el mapa provincial revela decenas de actas fronterizas capaces de alterar tanto la composición interna de cada bloque como el peso real de cada papeleta. ### Cuánto pesa realmente un voto según la provincia El modelo electoral de *El País*, actualizado con encuestas hasta el 16 de noviembre, clasifica 168 escaños como seguros, 140 como probables y **89 en disputa**, con **42 asientos directamente en el aire**. Aunque Madrid (10 escaños en juego) y Barcelona (9) concentran el mayor volumen absoluto de diputados disputados, las circunscripciones medianas y pequeñas disparan el valor marginal de cada papeleta debido al menor número de electores por acta: en **Soria** corresponden unos **25.000 votantes por diputado**, frente a unos **94.000** en Madrid. Según las simulaciones de *El País*, un voto cualquiera tiene de media **una probabilidad entre 116.000** de decidir el último escaño de su provincia. Sin embargo, esa probabilidad se multiplica casi por seis en **Lleida** —donde cinco listas (Junts, PP, PSOE, ERC y Vox) se disputan dos de sus cuatro escaños y un voto al PP alcanza una probabilidad entre **20.000** de mover acta— y por cinco para el PSOE en **Ávila** o **La Rioja**. En el extremo opuesto, una papeleta de PP o Vox en Madrid pesa aproximadamente la mitad de la media nacional. ### El mapa del voto estratégico en derecha e izquierda Al cruzar las simulaciones provincia a provincia, la utilidad táctica dentro de cada espectro ideológico resulta muy asimétrica: - **En la derecha (38 provincias con voto táctico relevante):** según *El País*, votar al **PP** maximiza las opciones del bloque conservador en **21 circunscripciones**, mientras que optar por **Vox** resulta más rentable en **17**. En **Álava** y **Melilla**, una papeleta popular vale más de cinco veces lo que una de Vox para capturar el último escaño; en **Gipuzkoa, Lugo y Ourense**, Vox apenas tiene opciones matemáticas de entrar y el PP se juega el resto frente a EH Bildu o el BNG. En cambio, en territorios como **La Rioja**, los restos favorecen a Vox como mejor baza del bloque. - **En la izquierda (hegemonía táctica del PSOE en 23 provincias):** la debilidad demoscópica de las listas a su izquierda convierte al **PSOE** en la única fuerza capaz de disputar el último resto en **23 circunscripciones** según los modelos de *El País*. Las excepciones nacen de marcas territoriales arraigadas: el **BNG** en Lugo y Ourense (y en liza abierta con los socialistas en A Coruña, donde el PSOE retiene el tercer escaño en el 38 % de las simulaciones de RTVE frente al 27 % del Bloque), **Adelante Andalucía** en Córdoba o **Unidad del Pueblo Canario** en Santa Cruz de Tenerife. ### Las cinco provincias a menos de dos puntos del Frente Amplio Precisamente para combatir la fuga hacia el «voto útil» socialista, la coalición **Frente Amplio** encabezada por Ada Colau ha puesto sobre la mesa sus propios cálculos de campaña, apoyados en una reestimación propia del CIS preelectoral y publicados por *elDiario.es*. Su tesis se apoya en la no linealidad de D'Hondt: en sus modelos internos, pasar del **7 % al 9 %** nacional aporta solo 6 escaños, pero saltar del **9 % al 11 %** desbloquea entre **12 y 13 diputados adicionales**, situando el objetivo del **10 %** en torno a 21 actas. Demoscopistas independientes consultados por el mismo medio rebajan ese optimismo y elevan el listón hasta el **11,5 %** para alcanzar la veintena de escaños ante la pujanza de Vox y la fragmentación territorial. Con todo, la dirección del Frente Amplio identifica **cinco circunscripciones** a menos de dos puntos porcentuales de arrebatar un escaño directamente a PP o Vox: - **Asturias:** a unas **0,2 décimas** porcentuales de disputar el último escaño al PP. - **Las Palmas:** **0,4 puntos** por delante de Vox en la última acta en juego, lo que convierte cualquier trasvase al PSOE en riesgo de ceder el escaño a la derecha. - **Sevilla:** a **1 punto** de arrebatar representación al PP. - **Madrid:** a **1,5 puntos** de lograr su cuarto diputado a costa de los populares. - **Castellón:** Compromís-Front Valencià se halla a **1,9 puntos** de restar un escaño a Vox (mientras que en **Alicante** y **Córdoba** la brecha frente a Vox sube a 2,9 y 3,7 puntos). ### Lo que descuentan los promedios nacionales Frente a esa pelea de décimas en los márgenes, los agregadores nacionales siguen reflejando una ventaja estructural del bloque conservador. Las **20.000 simulaciones** del promedio de **DatosRTVE** (actualizado este 18 de noviembre) fijan un escenario central de **129 escaños para el PP** (103 asegurados), **112 para el PSOE** (90 asegurados) y **64 para Vox** (42 asegurados), lo que entrega la mayoría absoluta de **176 diputados** a PP y Vox en el **94 %** de las simulaciones y reduce al **3 %** la reedición de una investidura progresista como la de 2023. En línea similar, la proyección de **ElectoralLab** sobre 100 encuestas de 22 empresas difundida por *El Debate* sitúa al PP en **127 escaños**, al PSOE en **104** y a Vox en **69** (mayoría absoluta de derecha en el **92 %** de simulaciones), ilustrando el giro en feudos clave como la Comunidad Valenciana: en **Valencia** (16 actas) y **Alicante** (12), tanto PP como PSOE cederían un diputado respecto a 2023 en favor del avance de Vox y la resistencia de Compromís-Front Valencià. Con entre **42 y 50 escaños interbloque** aún abiertos según el método empleado, el último tramo de campaña —y el cara a cara del 19-N— decidirá si la derecha amplía su colchón o si la movilización en las provincias medianas estrecha la aritmética del Congreso. ### Fuentes consultadas - [El País — El mapa del voto táctico para que el último escaño por provincia se lo lleve la derecha o la izquierda](https://elpais.com/espana/elecciones-generales/2026-11-18/el-mapa-del-voto-tactico-para-que-el-ultimo-escano-por-provincia-se-lo-lleve-la-derecha-o-la-izquierda-que-dicen-las-encuestas.html) - [elDiario.es — Los cálculos del Frente Amplio para la campaña: a menos de dos puntos de arrebatar cinco escaños a PP o Vox](https://www.eldiario.es/politica/cuentas-frente-amplio-campana-puntos-arrebatar-cinco-escanos-pp-vox_1_13673928.html) - [DatosRTVE — Encuestas elecciones generales 29N: la carrera por el escaño en cada provincia](https://www.rtve.es/noticias/20261113/encuestas-elecciones-generales-29n-carrera-escanos-provincia/17309293.shtml) - [El Debate — Encuestas electorales para el 29N: último escaño en juego en Valencia, Alicante y Castellón](https://www.eldebate.com/espana/comunidad-valenciana/20261118/encuestas-electorales-29n-resultados-pp-psoe-vox-compromis-valencia-alicante-castellon_481947.html) ~España ~Elecciones29N ~Encuestas ~Política ~FrenteAmplio
El mapa del último escaño ante el 29-N: 42 actas en el aire, voto táctico desigual y cinco provincias a dos puntos para el Frente Amplio
0
Discuss 4
Snap has released **Easy Lens CLI** (`@snap/easylens-cli`), a command-line interface designed to let developers build, preview, test, and publish Snapchat augmented-reality (AR) Lenses directly inside terminal and IDE AI coding assistants such as Anthropic's **Claude Code**, **Cursor**, and **Codex CLI**. Announced on November 18, 2026, via Snap's official newsroom and detailed across TechCrunch and XR technical reporting from *VR.org*, the launch extends Snap's existing web- and mobile-prompted Easy Lens builder into everyday developer agent loops. Under the hood, however, inspecting the actual npm package reveals important caveats around CPU architectures, closed-source licensing, and hardware compatibility that Snap's announcement glosses over. ### How Agents Drive the AR Pipeline Rather than asking engineers to switch into Snap's desktop Lens Studio GUI or paste prompts into a browser tab, Easy Lens CLI acts as an agent-native workflow bridge. On Snap's setup portal (`easylens.snapchat.com/cli`), developers pick Claude Code, Cursor, Codex CLI, or a generic agent to receive bootstrap instructions: install the global package (`npm install -g @snap/easylens-cli@latest`), authenticate with `easylens auth login`, run `easylens agent print` to load the bundled specification, or invoke `easylens agent init` (targeting `claude`, `cursor`, or `codex`) to inject project rules without overwriting existing config files. Once initialized, the coding assistant can assemble Lenses from **more than 100 AR building blocks** and **over 50 AI asset generators**, render preview videos, push test builds straight to a linked smartphone, iterate on errors, and submit finished experiences through Snapchat's standard content review. Supported outputs span camera filters, beauty effects, brand activations, and full **Web Lens** browser-style games complete with leaderboards—all eligible for Snap's **Lens+ Payouts** creator program, which Snap says distributed **$1.8 million** to community creators last month. Developers can also query Lens+ analytics directly from the terminal. Notably, Easy Lens CLI does not replace **CLAD**, Snap's existing Model Context Protocol (MCP) server documented under Lens Studio for driving the desktop editor via Claude Code, Cursor, VS Code, or Codex. Instead, Easy Lens CLI operates as a standalone, prompt-to-publish pipeline with no Lens Studio documentation entry at launch. ### What the npm Package Actually Installs Snap's announcement states simply that Easy Lens CLI "runs on Mac, Windows and Linux." A teardown of the npm package by *VR.org* shows a narrower technical reality. Although distributed via npm (requiring Node 18 or newer), the top-level package contains only four files; its entry script (`bin/easylens.js`, roughly 1 KB) contains zero AR logic and instead dispatches arguments to a compiled native binary pulled via npm `optionalDependencies`. Only **three platform-architecture binaries** exist in the `0.2.99` release (published November 17 after initial `0.0.0-stage` staging builds on November 6): - **macOS Apple Silicon (`arm64`):** roughly 75 MB unpacked across 36 files (Intel `darwin-x64` Macs are unsupported). - **Linux (`x64`):** roughly 103 MB unpacked (no `linux-arm64` binary exists). - **Windows (`x64`):** roughly 70 MB unpacked (excluding Qualcomm Snapdragon `win32-arm64` laptops). Developers on ARM Linux workstations, Snapdragon Windows PCs, or older Intel Macs hit an unsupported architecture message at launch. ### Closed-Source Trade-Offs and Missing Glasses Support Two broader strategic constraints stand out as well. First, free to install is not open source: `@snap/easylens-cli` points to Snap's proprietary Lens Studio License Agreement and Generative Lens Tools Terms with no public source repository. That contrasts sharply with Google's decision two weeks earlier to release **Three-Live** on GitHub under the Apache 2.0 license so developers can inspect how AI-generated 3D scenes are sanitized before rendering. Second, Snap's own Easy Lens documentation explicitly restricts the pipeline to the **Snapchat mobile app**. Lenses generated through Easy Lens CLI are **not** supported on Snap's **SPECS** AR glasses, Camera Kit Web, or Camera Kit Mobile—reinforcing that Snap's immediate priority is feeding fresh Web Lens games and monetizable filters into mobile Snapchat engagement rather than opening an agentic pipeline for standalone XR hardware. ### Sources & Further Reading - [Snap Newsroom — Introducing Easy Lens CLI: Create Snapchat Lenses with the AI Coding Assistant You Already Use](https://newsroom.snap.com/introducing-easy-lens-cli) - [VR.org — Snap's Easy Lens CLI Puts a Lens Build on npm, With Native Binaries for Only Three Platforms](https://vr.org/articles/snap-easy-lens-cli-npm-native-binaries-three-platforms-2026) - [TechCrunch — Snapchat Now Lets Developers Build AR Lenses With AI Assistants Like Claude Code and Cursor](https://techcrunch.com/2026/11/18/snapchat-now-lets-developers-build-ar-lenses-with-ai-coding-assistants-like-claude-code-and-cursor/) - [Snap Easy Lens CLI Portal](https://easylens.snapchat.com/cli) ~AI ~AugmentedReality ~Snapchat ~DeveloperTools ~ClaudeCode
Snap Launches Easy Lens CLI for Claude Code, Cursor, and Codex—With Native Binaries Limited to Three Architectures
0
Discuss 8
Telegram is testing a passphrase-gated **"Hidden Chats"** vault in the beta build of its native macOS client (**version 13.2.0, build 283562**), allowing users to scrub selected conversations and channels from their visible interface until a secret phrase is typed directly into the app's search bar. First documented by Russian tech outlet *Kod.ru* and corroborated across independent reporting by Meduza, Belarusian portal *Onlíner*, and *Rozetked*, the feature introduces a local "double-bottom" compartmentalization layer rather than a new cryptographic protocol. For users navigating workplace privacy, border device checks, or politically sensitive messaging across post-Soviet jurisdictions where Telegram faces sustained blocking and pressure, the distinction between local UI concealment and cross-device encryption matters enormously. ### How the Search-Bar Vault Works Unlike locked folders or passcode-protected apps that leave an obvious padlock icon on display, Hidden Chats exposes no permanent button in the main Telegram window. Inside **Settings → Privacy & Security**, users open the Hidden Chats menu and enter an eight-character-minimum passphrase twice (currently restricted to Latin characters). Once configured, typing that phrase into Telegram's ordinary top search bar unlocks an isolated compartment where selected direct chats and channels can be moved. Once moved inside, hidden conversations are scrubbed across five visible surfaces on that Mac: - **Main list and custom folders:** Hidden chats vanish from both the primary chat feed and user-created folder tabs. - **Search and notifications:** Messages inside hidden threads no longer surface in global or local search results, push banners, or the macOS dock's unread counter badge. - **Voice and video calls:** Incoming calls from hidden contacts are silently declined without ringing unless the user is actively inside the unlocked Hidden Chats view. - **Multiple compartments:** Users can register several distinct passphrases, each unlocking a separate subset of hidden conversations—enabling compartmentalized plausible deniability if someone is forced to unlock one decoy space. - **Automatic relocking:** The hidden section locks itself immediately when the Mac screen locks, the computer enters sleep mode, or the app sits idle for five minutes, and can also be closed manually. ### Local Concealment Is Not End-to-End Secret Chats A critical point underscored across Russian and Belarusian technical coverage is what Hidden Chats is *not*. Despite the similar naming, Hidden Chats does **not** turn cloud-synced chats or channels into Telegram's device-bound, end-to-end encrypted **Secret Chats**. Instead, it acts as a local, passphrase-keyed visibility filter over regular conversations and channels. Two major operational limitations flow directly from that design in the current macOS beta: First, **there is no cross-device sync**. Hiding a channel or conversation on a MacBook does nothing on a linked iPhone, Android handset, or secondary desktop client, where those exact threads remain plainly visible in the main inbox. Anyone relying on Hidden Chats to pass an inspection of multiple devices would still expose those chats on a phone until Telegram expands and synchronizes the feature across platforms. Second, **there is no passphrase recovery**. Telegram does not store the secret phrase on its servers. As *Kod.ru* and *Rozetked* both noted, forgetting the passphrase locks the user out of those hidden conversations on that Mac permanently unless they log out of their Telegram account on the machine and sign back in—resetting local hidden-state configuration. ### Why "Double-Bottom" Privacy Is Arriving Now The macOS beta experiment arrives shortly after Telegram added a targeted-attack protection mode to its macOS test builds restricting automatic media downloads, and amid escalating digital pressure in Russia and neighboring states. Russian authorities have restricted Telegram for months while promoting the state-backed messaging app Max—which security researchers at InterSecLab recently accused of selectively altering user surveillance settings—and physical phone and laptop inspections at Russian border crossings have made visible chat lists and channel subscriptions a recurring legal hazard. Whether Telegram rolls Hidden Chats out to iOS, Android, Windows, and cross-device sync before version 13.2 exits beta remains unconfirmed. Even in its single-device macOS preview, however, the feature signals that Telegram's client engineering is increasingly focused not just on server-side encryption, but on shielding what appears on screen when someone else is looking over a user's shoulder. ### Sources & Further Reading - [Meduza — Telegram Tests Password-Protected 'Hidden Chats' on macOS](https://meduza.io/en/news/2026/11/18/telegram-tests-password-protected-hidden-chats-on-macos) - [Kod.ru (Russia) — В бете Telegram для macOS появились скрытые чаты по секретной фразе](https://kod.ru/telegram-macos-hidden-chats) - [Onlíner Tech (Belarus) — В Telegram можно будет вызвать скрытые чаты по секретной фразе](https://tech.onliner.by/2026/11/18/v-telegram-mozhno-budet-vyzvat-skrytye-chaty-po-sekretnoj-fraze) - [Rozetked (Russia) — В Telegram тестируют скрытые чаты](https://rozetked.me/news/49541-v-telegram-testiruyut-skrytye-chaty-oni-ischezayut-iz-spiska-papok-poiska-i-uvedomleniy) ~WorldNews ~Telegram ~Privacy ~macOS ~Cybersecurity
Telegram Tests Passphrase-Gated "Hidden Chats" in macOS Beta 13.2.0, Hiding Threads Behind the Search Bar
0
Discuss 5
Russian state prosecutors told Moscow's Basmanny District Court on November 18 that Krestina, Angelina, and Maria Khachaturyan acted in "necessary self-defense" when they killed their abusive father in July 2018, demanding that the eight-year murder prosecution be terminated and all pre-trial restrictions lifted. The intervention—reported by state wire services RIA Novosti and TASS alongside independent outlets Meduza, Echo FM, and Russian domestic portal *Pravmir*—marks the sharpest institutional rift yet between Russia's supervisory prosecution service and the Investigative Committee over one of the country's most consequential domestic-violence cases. Yet because Russian criminal procedure restricts courts from directly overriding an investigator's pre-trial measures, the sisters remain barred from speaking to one another for now. ### Why Prosecutors and Investigators Are Deadlocked The procedural confrontation exposed in Basmanny District Court on November 18 centers on a rare public collision between two pillars of Russia's law-enforcement apparatus. Hearing a defense complaint filed on behalf of the youngest sister, Maria Khachaturyan—represented by attorney Yaroslav Pakulin—the prosecutor declared the case investigator's actions "unlawful" and accused investigators of causing "egregious delays" by repeatedly extending the inquiry without a final ruling. Speaking to *Pravmir*, defense lawyer Mari Davtyan, who represents Angelina Khachaturyan and has worked on domestic-violence cases for over a decade, revealed that the Prosecutor General's Office has actually been demanding the termination of the sisters' prosecution since **April 2026** on the grounds that no crime was committed (*otsutstviye sostava prestupleniya*). Despite that supervisory directive, the Investigative Committee investigator maintained in a formal written submission to the court that evidence still tied the sisters to their father's death and refused to lift preventive restrictions. That refusal was handed back by the bench. On November 17, Basmanny District Court upheld a defense delay complaint and took the unusual step of issuing a formal **private ruling (*chastnoye postanovleniye*)** addressed directly to the leadership of the investigative body over procrastination and failure to execute prosecutorial orders. On November 18, the court ruled the investigator's refusal to reconsider Maria's restrictions unlawful as well. ### Eight Years After 2018—and a 2025 Posthumous Verdict The deadlock looks even starker against the legal record established in 2025. Krestina, Angelina, and Maria—then aged 19, 18, and 17—were arrested in July 2018 after admitting they killed their 57-year-old father, Mikhail Khachaturyan, following years of documented physical, psychological, and sexual abuse. After initial detention, they were placed under strict preventive restrictions that still prohibit the three sisters from contacting each other, while Maria has remained barred from using a phone or the internet for more than eight years (contacts with their mother are permitted). In **April 2025**, following a posthumous trial consented to by Mikhail Khachaturyan's relatives, a Moscow court found the deceased father guilty of violent sexual acts against his daughters, causing serious bodily harm, and distributing pornography—a verdict Moscow City Court upheld on appeal in **November 2025**. Rather than closing the daughters' file afterward, however, the Investigative Committee reopened the murder inquiry in **January 2026** to examine whether the sisters had exceeded the legal limits of permissible self-defense. For Russian legal analysts and women's rights advocates, that sequence highlights a structural gap in Russian criminal law: Russia still lacks a dedicated domestic-violence statute, leaving victims who fight back during prolonged abuse exposed to prosecution under standard homicide rules unless investigators recognize imminent threat at the moment of the act. ### What Happens Next for the Three Sisters Despite two consecutive courtroom victories on November 17 and 18, defense lawyers caution against expecting immediate relief. As Mari Davtyan explained to *Pravmir*, Russian courts lack the statutory authority to directly annul an investigator's pre-trial measure during an active preliminary probe; the Basmanny court can only declare the refusal unlawful and compel the investigator to reconsider the defense motion promptly. Still, the dual pressure of a formal court rebuke to Investigative Committee leadership and an explicit on-the-record self-defense finding by state prosecutors sharply narrows the investigator's room to stall. Maria's attorney, Yaroslav Pakulin, told TASS (via Echo FM) that the court found no persuasive grounds for dragging the investigation out further, raising defense expectations that the criminal case could finally be closed on rehabilitative grounds—recognizing necessary self-defense—before the end of 2026. Whether the Investigative Committee yields within the coming month will test how much authority Russia's courts and prosecutors can still exert over protracted criminal probes. ### Sources & Further Reading - [Meduza — Russian Prosecutors Call for an End to the Khachaturyan Sisters' Murder Case, Saying They Acted in Self-Defense](https://meduza.io/en/news/2026/11/18/russian-prosecutors-call-for-an-end-to-the-khachaturyan-sisters-murder-case-saying-they-acted-in-self-defense-when-they-killed-their-abusive-father) - [The Moscow Times — Prosecution Seeks to Drop Khachaturyan Sisters Murder Case](https://www.themoscowtimes.com/2026/11/18/prosecution-seeks-to-drop-khachaturyan-sisters-murder-case-a94304) - [Pravmir (Russia) — Генпрокуратура признала действия сестер Хачатурян самообороной: интервью с Мари Давтян](https://www.pravmir.ru/genprokuratura-predlozhila-prekratit-delo-sester-hachaturyan-priznav-ih-dejstviya-samooboronoj/) - [Echo FM (Russia) — Адвокат Марии Хачатурян: генпрокуратура сочла необходимой самообороной действия сестёр](https://echofm.online/news/advokat-marii-hachaturyan-genprokuratura-sochla-neobhodimoj-samooboronoj-dejstviya-sestyor-hachaturyan) ~WorldNews ~Russia ~Justice ~HumanRights ~RuleOfLaw
Russian Prosecutors Urge Dismissal of Khachaturyan Sisters Murder Case, Citing Self-Defense After Eight-Year Probe
0
Discuss 8
Embark Studios has pulled down an official esports recap video for *The Finals* and canceled two planned regional follow-ups after replacing the tournament's human commentators with the shooter's fictional AI announcers, Scotty and June. The backlash—sparked by commentators who called four qualifier cycles for the Asia-Pacific (APAC) Closed Qualifiers ahead of **The Grand Major 2026**—has turned a single promotional highlight reel into a broader flashpoint over generative AI and broadcast labor across competitive gaming. In a formal statement provided to IGN and echoed in an apology on X by Embark's Competitive Experience and Community Lead Oscar Lundberg, the Stockholm-based studio conceded that omitting the live broadcast talent "wasn't the right call." Beyond deleting the APAC recap video, Embark confirmed that matching Americas (AMER) and Europe, Middle East, and Africa (EMEA) highlight packages built in the same AI-narrated format will not be released as planned. ### How the Broadcast Was Overwritten The controversy centered not on whether Embark had permission to use its own characters' voices, but on how human labor was erased—and echoed—after the fact. Scotty and June are voiced via text-to-speech models trained on performances by paid human actors, and in October Embark expanded their in-game presence with a "Dynamic Commentary" system capable of reading player usernames and reacting to live match state. For the APAC Closed Qualifiers, however, real commentators—including ProtoANZ (Kai), TheeHyperactive, and Bluepanda—called the matches live across four cycles of open and closed competition. When Embark published its official highlight reel on November 16 to drum up casual interest ahead of the **$200,000 Grand Major** at DreamHack Stockholm (November 27–29, 2026), their voices were gone. Speaking to IGN, Proto explained the production pipeline as organizers described it to him: Embark's video team did not feed the commentators' raw broadcast audio into an AI voice cloner. Instead, producers combed through the tournament VOD to pick out narrative moments, handed a rough script to a writer, and generated fresh Scotty and June TTS lines—yet the resulting script still lifted phrasing and lines that the human casters had delivered live on air. ### Why Paid Talent Still Felt Betrayed To outsiders, Embark's initial rationale—using recognizable in-universe hosts to bridge casual shooter fans into competitive play—might appear distinct from replacing live match coverage altogether. To the casters involved, the sequence felt like a structural dismissal. Bluepanda, who noted he has spent nearly seven years in esports hearing that his work was valued only to be sidelined or told talent budgets were shifting, told IGN that Embark's initial reaction read like damage control rather than genuine remorse, prompting him to step away from the game and esports altogether. Proto pointed out the basic economic contradiction: Embark had already paid the casters decent rates to call the live qualifiers, making the decision to pay again for scripted AI dubbing over their highlights both baffling and demoralizing. The reaction rapidly spilled beyond *The Finals* into the wider esports ecosystem. Commentators from *League of Legends*, *Marvel Rivals*, and fighting-game circuits condemned the clip—not just on principle, but over the quality of the synthetic delivery itself—warning that studios normalizing AI play-by-play in official shoulder content risk eroding the grassroots commentary pipelines that sustain competitive scenes between majors. ### Embark's AI Pattern and the Push for Contract Protections The incident did not land in a vacuum. Embark has repeatedly defended synthetic voice pipelines across its portfolio, including defending AI-generated item callouts in *ARC Raiders* earlier in November 2026 by arguing it was impractical for voice actors to record every permutation. While players have debated that trade-off inside PVE/extraction gameplay, esports broadcasts rely directly on human pacing, improvisation, and personality—qualities commentators argue cannot be treated as disposable scaffolding. For Proto, the episode represents a warning sign extending well beyond microphones into back-end broadcast production. Just as automatic AI camera observers in shooters can already switch perspectives without human operators, unchecked generative commentary opens what he called a "Pandora's Box" unless broadcast workers organize. He urged commentators, observers, producers, and camera operators across esports to unionize or withhold labor until event contracts explicitly prohibit using their broadcasts or voices in generative AI workflows—alongside concrete local-scene support and clearer talent communication from Embark ahead of Stockholm. Whether Embark's decision to scrap the AMER and EMEA recaps repairs trust before The Grand Major kicks off on November 27 remains to be seen. What the episode already proves is that even a studio built around AI-assisted tooling cannot substitute synthetic mascots for real tournament broadcasters without triggering a cross-scene revolt. ### Sources & Further Reading - [PC Gamer — Embark Studios Apologizes for Replacing Esports Casters With AI Voices in Now-Deleted The Finals Video](https://www.pcgamer.com/software/ai/embark-studios-apologizes-for-replacing-esports-casters-with-ai-voices-in-now-deleted-the-finals-video/) - [IGN — Esports Casters Call for Change After Embark Studios Replaces Talent for The Finals Highlight Reel With AI Commentators](https://www.ign.com/articles/its-been-a-shtshow-esports-casters-call-for-change-after-embark-studios-replaces-talent-for-the-finals-grand-major-qualifier-highlight-reel-with-games-ai-commentators) - [Oscar Lundberg (Embark Studios) Official Apology Statement on X](https://x.com/Schnobben/status/2122641480485666816) - [The Finals Wiki — The Grand Major 2026 Format, Dates, and Prize Pool](https://www.thefinals.wiki/wiki/The_Grand_Major_2026) ~Gaming ~TheFinals ~Esports ~GenerativeAI ~Industry
Embark Studios Pulls The Finals Recap and Scraps Two Regional Videos After Replacing Esports Casters With AI
0
Discuss 11
RadixArk, the AI infrastructure company behind the open-source SGLang serving engine and Miles reinforcement-learning stack, has released its first open-weight model: **DaisyRose-27B**. Built directly on top of Alibaba's [Qwen3.8-27B](https://www.mushimu.com/post/8976) and published under a permissive Apache 2.0 license in BF16, FP8, and NVFP4 checkpoints, the 27-billion-parameter multimodal reasoning model is an explicit bet that pure reinforcement learning—without a single step of supervised fine-tuning or distillation—can extract frontier-class agentic coding performance from a dense model small enough to run on a single GPU. The release arrives at an inflection point for local and mid-sized open-weight builders. While Alibaba recently introduced its cloud-only [Qwen4-Max](https://www.mushimu.com/post/9581) without open weights for a Qwen 4 27B tier yet, `Qwen3.8-27B` remains the workhorse dense foundation for single-node deployments. Rather than altering the base model's tokenizer, chat template, image-understanding pipeline, or 256K-token context window, RadixArk kept DaisyRose-27B as a strict drop-in replacement—concentrating its entire compute budget on post-training efficiency and long-horizon agentic execution. ### Why Pure RL—and Why Only 1,200 Tasks? Conventional post-training pipelines lean heavily on supervised fine-tuning (SFT) over synthetic trajectories or distillation from a larger proprietary teacher before applying light RL. RadixArk bypassed both stages. Working alongside Eigent AI, the team started with thousands of candidate tasks across software engineering, machine learning, and scientific problem-solving, then winnowed them down to roughly **1,200 curated environments**. Every surviving task had to clear three filters. First, it was redesigned specifically around failures observed when running `Qwen3.8-27B` directly, calibrated so the base model's pass rate landed inside an informative gradient window rather than at zero or saturation. Second, tasks demanded genuine multi-step planning: each sandbox attempt ran for up to **1.5 hours**, spanning more than **100 tool calls** and over **256K output tokens**. Third, tasks with exploitable grading shortcuts were audited and repaired or discarded. To stretch learning across sessions reaching **1 million tokens** without exceeding the 256K-token native window, RadixArk used harness summarization across chained trajectories. ### Asynchronous RL and the 80% Sandbox Cost Cut Training agents that think and execute shell commands for 90 minutes per rollout normally starves training GPUs while trainers wait on slow environments. RadixArk solved that bottleneck using **Miles** in fully asynchronous mode on **144 NVIDIA H200 GPUs** over 18 days (63,000 GPU-hours total): 128 GPUs continuously generated agent rollouts while 16 GPUs ran weight updates on an internal training backend, pushing fresh weights back out after every step. According to RadixArk's telemetry, that backend achieved **37% model FLOPs utilization (MFU)**, compared with **16% MFU** on a comparable open-source Megatron trainer—a **57% reduction in GPU time** and roughly double the end-to-end throughput (**16,800 tokens/second** versus 8,300). On the environment side, RadixArk built its execution layer on the open-source **AgentENV** distributed sandbox platform, running roughly **944,000 sandboxes** between mid-September and mid-November at up to **1,566 concurrent instances**, dynamically sizing sandbox memory to cut cloud sandbox spending by roughly **80%** and lifting creation reliability above **99.95%** from late October onward. ### Reading the Benchmarks—and Their Caveats Across RadixArk's evaluations (using `temperature=1.0`, `top_p=0.95`, `top_k=20`, and `xhigh` reasoning effort), DaisyRose-27B improves on its `Qwen3.8-27B` base across all seven reported suites: - **Terminal-Bench 4.1:** 12.0% pass@1 (up from 5.6% on Qwen3.8-27B; averaged over 8 repetitions with `mini-swe-agent` summarization) - **Terminal-Bench 2.1:** 75.5% (up from 73.0%) - **SWE-Bench Pro v2:** 94.7% (up from 91.3%) - **DeepSWE v1.1:** 46.2% (up from 42.2%) - **IFBench:** 82.4% (up from 79.5%, beating every larger open-weight model listed in RadixArk's comparison) - **GPQA Diamond:** 90.7% (up from 89.2%, edging out GPT-6 Luna's 90.5%) - **Humanity's Last Exam (text-only):** 34.3% (up from 33.9%) Two methodological details merit scrutiny before treating the scoreboard as an apples-to-apples sweep. First, RadixArk reports the higher of its BF16 and FP8 runs per benchmark (noting the two stayed within one percentage point where both ran). Second, because several reference baselines lacked Terminal-Bench 4.1 entries, the model card substitutes Terminal-Bench 4.0 scores for `Qwen3.8-27B`, `Muse Glimmer-30B`, `Qwen3.6-35B-A3B`, and `Muse Spark Open`, while `Inkling 2 Small` (594B total / 17B active) and `MiMo V2.6 Flash` (309B total / 15B active) still lead on raw coding benchmarks like DeepSWE v1.1. Even with those caveats, matching or outscoring 500B–800B MoE models on IFBench and GPQA Diamond at 27B dense parameters demonstrates how much headroom remains in post-training curation alone. ### Single-GPU Deployment and Hybrid GDN Tuning For practitioners deploying locally, the [SGLang Cookbook](https://docs.sglang.io/cookbook/autoregressive/RadixArk/DaisyRose-27B) lists single-GPU configurations on NVIDIA H200, GB300, RTX PRO 6000, RTX 5090, and DGX Spark across BF16, FP8, and NVFP4 checkpoints. Combining DaisyRose-27B with RadixArk's **DSpark** speculative draft model (`RadixArk/Qwen3.8-27B-DSpark`) yields **3.4 to 3.5 accepted tokens per step** on GB300 in long-context thinking workloads. One non-obvious operational pitfall highlighted in SGLang's documentation involves the `--mamba-full-memory-ratio` flag. Because the Qwen3.8-27B architecture uses hybrid Gated DeltaNet (GDN) layers alongside paged attention, SGLang's default `0.9` ratio over-provisions the KV cache pool and quietly throttles concurrency; calculating the ratio from expected average request length and GDN state slots unlocks full multi-request throughput on a single card. ### Sources & Further Reading - [Hugging Face Blog — DaisyRose-27B: Pushing the Frontier of Post-Training Efficiency](https://huggingface.co/blog/RadixArk/daisyrose-27b-pushing-the-frontier-of-post-trainin) - [RadixArk Official Launch Article — DaisyRose-27B](https://www.radixark.com/blog/daisyrose-27b) - [SGLang Documentation — Deploying DaisyRose-27B With DSpark Speculative Decoding](https://docs.sglang.io/cookbook/autoregressive/RadixArk/DaisyRose-27B) - [GitHub — kvcache-ai/AgentENV Distributed Agent Sandbox](https://github.com/kvcache-ai/AgentENV) - [Hugging Face Model Card — RadixArk/DaisyRose-27B](https://huggingface.co/RadixArk/DaisyRose-27B) ~AI ~OpenSourceAI ~Qwen ~ReinforcementLearning ~SGLang
RadixArk Releases DaisyRose-27B: Pure Reinforcement Learning Doubles Terminal-Bench 4.1 Score on Qwen3.8-27B
0
Discuss 9
El desembarco de **Alberto Núñez Feijóo** en Guipúzcoa y Vizcaya ha convertido la pugna por el último escaño guipuzcoano en uno de los puntos de mayor fricción de la recta final hacia el 29-N. Tras el lanzamiento de piedras, huevos y petardos contra el cabeza de lista del PP por Guipúzcoa, **Jon Echeverría**, el pasado sábado 14 de noviembre en Hernani —y una agresión previa a dos jóvenes de Nuevas Generaciones en la sede de Barakaldo—, el líder popular alteró su agenda para arropar a su candidato en Eibar bajo un fuerte dispositivo de la Ertzaintza, acusando a Pedro Sánchez de *«blanquear a Bildu»* y asegurando que *«la violencia es consecuencia directa»* de esa estrategia parlamentaria. ### De Hernani a Eibar: seguridad blindada y campaña sin previo aviso La tensión que rodeó el acto de Eibar retrata el clima excepcional con el que los populares afrontan la campaña en determinados municipios de Guipúzcoa. Según la crónica sobre el terreno publicada por [El Mundo](https://www.elmundo.es/elecciones/elecciones-generales/2026/11/17/6afca35ce4d4d8b81f8b45ab.html), dos docenas de agentes de la Ertzaintza —incluidos efectivos apostados en azoteas— acordonaron la plaza frente al Centro Social para contener a grupos de jóvenes que coreaban consignas de *«¡Kanpora!»* y descalificaciones contra la comitiva popular, obligando a situar el autobús electoral del PP —recién limpiado tras sufrir pintadas en Lleida— como barrera visual. El episodio prolonga una secuencia iniciada en la Parte Vieja de Hernani, donde la presidenta del PP guipuzcoano, Muriel Larrea, denunció que los agresores no solo arrojaron objetos pirotécnicos y piedras sino que intimidaron al equipo con gestos simulando disparos. Como documentó la prensa vasca en [El Diario Vasco](https://www.diariovasco.com/politica/jon-echeverria-prosigue-campana-actos-sin-anunciar-tras-agresion-hernani-20261116132029-nt.html), la dirección territorial optó tras aquel incidente por mantener su ruta provincial mediante actos sin convocatoria pública previa y solicitó mayor protección al Ministerio del Interior. Para [RTVE](https://www.rtve.es/noticias/20261115/jon-echevarria-positivo-pp-increpado-hernani-perfil/17315021.shtml), la crisis ha proyectado inesperadamente al ámbito nacional la figura de Echeverría, nacido en San Sebastián en 2004, afiliado a los 17 años y exaspirante a la alcaldía de Zumárraga en 2023, convertido ahora en símbolo del partido frente a la hostilidad callejera. ### La batalla estratégica: 7.000 votos, el PNV y los «socialistas vascos de verdad» Detrás de la respuesta simbólica —reforzada por la propuesta de Feijóo de reformar la LOREG para que los [exiliados por ETA recuperen el voto](https://www.mushimu.com/post/9391) en sus municipios de origen— subyace un cálculo electoral muy concreto. El PP calcula que se encuentra a unos **7.000 votos** de arrebatar un escaño clave en Guipúzcoa, disputa directa con EH Bildu en la que el partido busca capitalizar tanto el rechazo social a la intimidación como el desgaste de los partidos tradicionales vascos. Horas después de acompañar a Echeverría en Eibar, Feijóo trasladó ese mensaje al Palacio Euskalduna de Bilbao. Tal como recoge [Infobae / Agencia EFE](https://www.infobae.com/espana/agencias/2026/11/17/feijoo-pide-el-voto-a-los-socialistas-vascos-de-verdad-y-votantes-del-pnv-desencantados/), el dirigente popular tendió la mano a los *«votantes del PNV desencantados»* y apeló expresamente a los *«socialistas vascos de verdad»*, sugiriendo que a parte de las bases del PSE les avergüenza que la dirección federal de Pedro Sánchez exhiba mayor sintonía con Arnaldo Otegi que con referentes históricos como Eduardo Madina o Nicolás Redondo. *«Las piedras que le lanzaron a Jon son las mismas con las que Sánchez levantó el muro»*, sentenció Feijóo. ### El dilema electoral vasco ante el 29-N La estrategia popular busca convertir cada incidente en Euskadi en munición nacional contra los pactos del Ejecutivo central, presentando el voto al PP vasco como un «voto doble» para frenar a la izquierda abertzale en Madrid. Sin embargo, el escenario vasco plantea un equilibrio complejo: aunque la condena a los ataques en Hernani y Barakaldo refuerza la cohesión del centro-derecha en el resto de España, en Euskadi el PP compite con un PNV celoso de su espacio moderado y con un electorado socialista refractario al discurso de polarización nacional. La respuesta de las urnas el 29 de noviembre determinará si la confrontación directa en Guipúzcoa logra movilizar esos 7.000 sufragios decisivos o consolida los bloques existentes. ### Fuentes consultadas - [El Mundo (España — Crónica sobre el terreno en Eibar y Guipúzcoa)](https://www.elmundo.es/elecciones/elecciones-generales/2026/11/17/6afca35ce4d4d8b81f8b45ab.html) - [RTVE Noticias (España — Perfil de Jon Echeverría e incidentes en Hernani)](https://www.rtve.es/noticias/20261115/jon-echevarria-positivo-pp-increpado-hernani-perfil/17315021.shtml) - [Infobae / Agencia EFE (España — Mitin de Bilbao y llamamiento a votantes de PSE y PNV)](https://www.infobae.com/espana/agencias/2026/11/17/feijoo-pide-el-voto-a-los-socialistas-vascos-de-verdad-y-votantes-del-pnv-desencantados/) - [El Diario Vasco (País Vasco — Campaña con actos sin anunciar tras Hernani)](https://www.diariovasco.com/politica/jon-echeverria-prosigue-campana-actos-sin-anunciar-tras-agresion-hernani-20261116132029-nt.html) ~España ~Euskadi ~Elecciones29N ~PP ~Política
Tensión en la campaña vasca: las agresiones en Hernani y Eibar convierten el escaño de Guipúzcoa en eje del 29-N
0
Discuss 6
Britain’s Prince William and Catherine, the Princess of Wales, wrapped up a three-day tour of India on November 18, 2026, with talks alongside Indian Prime Minister Narendra Modi in New Delhi—capping the royal couple’s first official overseas journey together in four years and their first return to India in a decade. While Kensington Palace framed the meeting around bilateral warmth and environmental innovation following Tuesday’s **Earthshot Prize** ceremony in Mumbai, both London and New Delhi explicitly used the visit to reinforce strategic alignment around **Vision 2035** and the newly implemented **India–UK Comprehensive Economic and Trade Agreement (CETA)**. ### Soft-Power Diplomacy Meets Post-CETA Strategic Alignment The New Delhi meeting was added late to the royal itinerary after Modi confirmed he would return from a scheduled trip to the Philippines in time to host the future British monarch and his wife at his official residence complex. As reported by [BBC News](https://www.bbc.co.uk/news/articles/cq62ped2ypy4o) and India’s [Hindustan Times](https://www.hindustantimes.com/india-news/prince-william-kate-meet-pm-modi-in-delhi-on-final-day-of-india-visit-101795013178375.html), the scheduling flexibility signaled priority on both sides. For Downing Street, deploying William and Catherine—particularly as the Princess of Wales resumes high-profile international duties following her cancer treatment—serves as a calibrated diplomatic asset at a moment when the United Kingdom is seeking deeper non-EU trade and technology anchors. For New Delhi, the optics dovetailed neatly with India’s economic messaging. In a statement posted after Wednesday’s talks, Modi praised the Earthshot Prize for *"championing innovative climate solutions worldwide and also showcasing Indian ingenuity in addressing global environmental challenges,"* before tying the encounter directly to hard commercial policy: *"Look forward to further strengthening the India-UK Comprehensive Strategic Partnership and building on the momentum of the recently implemented Comprehensive Economic and Trade Agreement, for the shared prosperity of the people of our nations and the wider world."* Acting British High Commissioner to India Ben Mellor similarly pointed to **Vision 2035** as the framework translating prime ministerial pledges made in Mumbai last year into joint work across clean energy, technology, investment, education, and security, according to [Gulf News](https://gulfnews.com/world/asia/india/uks-prince-william-and-princess-kate-head-to-india-to-meet-pm-modi-attend-earthshot-prize-ceremony-1.500748474). ### Mumbai’s Earthshot Debut and India’s Clean-Tech Showcase The centerpiece of the trip took place a day earlier on November 17 in Mumbai, marking the first time the sixth annual Earthshot Prize—founded by Prince William in 2020 to award **£1 million** each to five environmental initiatives per year—was hosted on Indian soil. Hosted by filmmaker and television personality Karan Johar, the ceremony spotlighted clean mobility and industrial decarbonization, with Indian electric-vehicle manufacturer Mahindra Last Mile Mobility among the headline £1 million laureates. Choosing India as host reflected both geopolitical calculus and domestic market scale. When announcing Mumbai as the 2026 venue earlier this year, William cited India’s scale in climate innovation, biodiversity, and youth demographics. Alongside the formal awards, the royal couple leaned into accessible public diplomacy—riding an electric rickshaw in Mumbai and playing gully cricket alongside Indian batting icon Sachin Tendulkar—imagery designed to project continuity a decade after their 2016 tour of India and Bhutan without the colonial nostalgia that has complicated royal visits elsewhere in the Commonwealth. ### Balancing Climate Rhetoric With Trade Realities Behind the ceremonial choreography, both governments face practical tests in converting diplomatic goodwill into durable industrial results. Under CETA and Vision 2035, bilateral success will be measured less by royal engagements than by tariff reductions, clean-energy capital flows, technology transfers, and mobility negotiations between London and New Delhi. Still, Wednesday’s New Delhi talks underscored a clear convergence: for Britain, India’s clean-tech manufacturing base and consumer market are indispensable partners; for Modi’s government, hosting the Earthshot Prize and welcoming the UK’s next head of state reinforces India’s positioning as a bridge-builder in climate innovation and global trade. ### Sources & Further Reading - [BBC News (United Kingdom — New Delhi Talks & Royal Itinerary)](https://www.bbc.co.uk/news/articles/cq62ped2ypy4o) - [Hindustan Times (India — Modi Statement & CETA Context)](https://www.hindustantimes.com/india-news/prince-william-kate-meet-pm-modi-in-delhi-on-final-day-of-india-visit-101795013178375.html) - [Gulf News (UAE — British High Commission & Vision 2035 Framing)](https://gulfnews.com/world/asia/india/uks-prince-william-and-princess-kate-head-to-india-to-meet-pm-modi-attend-earthshot-prize-ceremony-1.500748474) ~WorldNews ~UnitedKingdom ~India ~Diplomacy ~EarthshotPrize
Prince William and Catherine Meet Modi in New Delhi as UK–India Ties Lean Into Earthshot and CETA Momentum
0
Discuss 8
Slovenia has quietly suspended tourist, student, and employment visa applications for Russian citizens, extending the crackdown beyond Russia’s borders to Russian passport holders applying through its embassy in neighboring Serbia. According to notices posted on November 17 by visa operator VFS Global and confirmed across independent reporting on November 18, 2026, the new guidelines took effect retroactively on **November 16** without an official public explanation or sunset date—making Ljubljana the latest Schengen member to tighten entry ahead of a bloc-wide European Commission proposal expected in January 2027. ### What Categories Remain Open—and Why Serbia Was Swept In Under the revised guidelines issued by Slovenia’s Foreign Ministry—whose diplomatic footprint in Moscow runs through the [Slovenian Embassy in the Russian Federation](https://www.gov.si/en/representations/embassy-moscow/)—visa centers in Russia now process Russian applications across only four narrow exemptions, as documented by [Meduza](https://meduza.io/en/news/2026/11/18/slovenia-stops-taking-tourist-visa-applications-in-russia-extends-restrictions-to-russians-in-serbia) and [The Moscow Times](https://www.themoscowtimes.com/2026/11/18/slovenia-halts-tourist-visas-for-russian-citizens-a94301): visits to family members who are EU citizens, family reunification with EU citizens (Category D national visas), business and conference travel carrying prior approval from the Slovenian Embassy, and international transport workers including truck drivers, seafarers, and pilots. What makes Slovenia’s rollout notably sharper than routine consular adjustments in Moscow is its extraterritorial reach into **Serbia**, which has absorbed tens of thousands of Russian emigrants since 2022 and frequently acts as a third-country Schengen application hub. According to emigrants’ rights project **Kovcheg**, Slovenia’s embassy in Belgrade began canceling pre-booked appointments for Russian nationals on November 16 and 17, scrapping at least nine confirmed slots and turning away another applicant at the door. In a cancellation notice shared by Kovcheg, the embassy stated explicitly that the restrictions apply exclusively to applicants holding Russian citizenship under updated Foreign Ministry directives. ### Small Volume, Larger Schengen Signal On paper, Slovenia was never a mass gateway for Russian tourism into the European Union. European Commission figures cited by the Association of Tour Operators of Russia (ATOR) and Russian travel-industry outlet [RusTourismNews](https://www.rustourismnews.com/2026/11/18/slovenia-stops-accepting-tourist-visa-applications-from-russians/) show Slovenia issued just **1,254 Schengen visas** to Russian nationals in 2025—more than 90% of them single-entry—out of roughly **680,000** issued across the entire EU. Yet the political timing is unmistakable. As [UA.News](https://ua.news/en/world/sloveniia-pripinila-vidavati-turistichni-vizi-gromadianam-rosiyi) and European policy reporting indicate, member states are actively debating tighter Schengen restrictions—including geographically limited visas—amid mounting warnings from security services over Kremlin-linked hybrid operations and sabotage across Europe. On October 11, Finnish Prime Minister Petteri Orpo urged the bloc to halt Russian tourist visas outright on security grounds, sharpening a split between frontline states that stopped issuing tourist visas after 2022 and southern hubs such as France, Italy, Spain, and Greece that have kept tourist channels open. ### What Changes for Russian Travelers and Émigrés For Russians still inside the country, Slovenia’s shift closes one more minor Schengen avenue just as winter travel planning begins, pushing remaining demand toward southern EU consulates. For Russian residents in Serbia—many of whom lack long-term EU residency and rely on local Schengen embassies for business, study, or family mobility—the Belgrade cancellations mark an unsettling precedent: citizenship-based restrictions can now follow applicants into non-EU third countries regardless of where they live. Until Ljubljana publishes formal policy reasoning or Brussels tables its January 2027 Schengen package, Russian nationals outside the four narrow exemptions face an indefinite freeze on Slovenian entry. ### Sources & Further Reading - [Meduza (Independent Russian Press — VFS Global & Kovcheg Report)](https://meduza.io/en/news/2026/11/18/slovenia-stops-taking-tourist-visa-applications-in-russia-extends-restrictions-to-russians-in-serbia) - [The Moscow Times (Independent / Europe — Embassy Cancellations & EU Visa Context)](https://www.themoscowtimes.com/2026/11/18/slovenia-halts-tourist-visas-for-russian-citizens-a94301) - [RusTourismNews (Russia — ATOR Data & Travel Sector Impact)](https://www.rustourismnews.com/2026/11/18/slovenia-stops-accepting-tourist-visa-applications-from-russians/) - [UA.News (Ukraine — Security Context & Schengen Divide)](https://ua.news/en/world/sloveniia-pripinila-vidavati-turistichni-vizi-gromadianam-rosiyi) - [Ministry of Foreign and European Affairs of Slovenia (Official Embassy in Moscow)](https://www.gov.si/en/representations/embassy-moscow/) ~WorldNews ~Slovenia ~Russia ~Schengen ~Visas
Slovenia Halts Tourist, Study, and Work Visas for Russians—and Extends Ban to Applicants in Serbia
0
Discuss 4
A menos de veinticuatro horas del [cara a cara decisivo del 19-N](https://www.mushimu.com/post/9603) frente a Alberto Núñez Feijóo y con la campaña electoral del 29-N entrando en su tramo final, el presidente del Gobierno, **Pedro Sánchez**, acudió este miércoles 18 de noviembre al plató de *Espejo Público* en Antena 3 con una doble estrategia discursiva: blindar las cuentas del PSOE frente a las pesquisas de la Unidad Central Operativa (UCO) de la Guardia Civil y trazar una línea divisoria entre los casos de corrupción que admite como reales y los procedimientos judiciales que afectan a su entorno familiar y al expresidente José Luis Rodríguez Zapatero. ### Financiación del PSOE: «cantidades nimias» frente al precedente de Gürtel Ante las preguntas de la periodista Susanna Griso sobre el riesgo de que la investigación abierta en la Audiencia Nacional sobre presuntas irregularidades contables arroje novedades en plena recta electoral, Sánchez expresó *«absoluta tranquilidad»* y rechazó de plano que su partido haya recurrido a financiación ilegal. Según recogieron tanto la propia cadena [Antena 3](https://www.antena3.com/noticias/espana/sanchez-admite-casos-corrupcion-gobierno-pero-defiende-psoe-nos-hemos-financiado-irregularmente_202611186afd674a103a0e0a474eb2ed.html) como [El País](https://elpais.com/espana/elecciones-generales/2026-11-18/sanchez-en-antena-3-absoluta-tranquilidad-el-psoe-no-se-ha-financiado-irregularmente.html), el jefe del Ejecutivo defendió que los socialistas remiten sus donaciones al Tribunal de Cuentas —organismo al que acudió la UCO para contrastar información— y calificó los pagos en metálico para comidas bajo sospecha de *«cantidades nimias»*. Para sostener que no existe una trama estructural sino episodios acotados, Sánchez contrastó la situación actual del PSOE con la condena firme al Partido Popular en el caso Gürtel y los 40 millones de euros localizados en Suiza al extesorero Luis Bárcenas. *«En el caso del PP hubo una condena, en nuestro caso no es así; hay una investigación»*, argumentó, situando al PSOE *«a años luz»* de PP y Vox en materia de transparencia. Al mismo tiempo, reconoció la gravedad política del caso que salpicó a **José Luis Ábalos** —condenado a 24 años y tres meses de prisión por la pieza de las mascarillas del caso Koldo— y al exsecretario de Organización Santos Cerdán, subrayando que asumió responsabilidades políticas y pidió perdón, aunque trató de encapsular el daño asegurando que Ábalos fue el único ministro afectado entre una *«miríada»* de nombramientos. ### Causas «reales» frente a «ficticias», Plus Ultra y la tensión con Griso El punto de mayor carga jurídica y electoral llegó cuando el debate pasó de la cúpula socialista a las causas abiertas sobre **Begoña Gómez**, su hermano **David Sánchez** —cuyos salarios en Badajoz analiza también el [Tribunal de Cuentas](https://www.mushimu.com/post/9550)— y el exfiscal general del Estado, Álvaro García Ortiz. Como detalla [20minutos](https://www.20minutos.es/nacional/sanchez-dice-que-ya-ha-asumido-responsabilidades-politicas-por-corrupcion-distingue-casos-reales-ficticios_7062500_0.html) y el despacho de [Infobae / Agencia EFE](https://www.infobae.com/espana/agencias/2026/11/18/sanchez-asegura-que-el-psoe-no-se-financio-irregularmente-y-considera-que-estan-en-cuestion-cantidades-nimias/), Sánchez contrapuso el caso Ábalos, que tildó de *«real»*, con esos procedimientos, que describió como *«ficticios»*. Sin llegar a pronunciar el término *lawfare*, acusó a sectores judiciales y opositores de estar *«forzando el derecho penal»* por una razón estrictamente partidista: *«la imposibilidad que tiene el PP de ganar las elecciones»*. En la misma línea se pronunció respecto al rescate de **53 millones de euros** a la aerolínea Plus Ultra durante la pandemia: respaldó la integridad de la presidenta de la SEPI, Belén Gualda, y negó que José Luis Rodríguez Zapatero mediara ante él para facilitar la operación, deslizando que le resulta *«bastante casualidad»* que las investigaciones patrimoniales apunten únicamente a ese expresidente socialista. La entrevista vivió su momento más áspero al abordar la crispación en los mítines y el cántico de *«hijo de puta»* dirigido contra el presidente, que Sánchez vinculó al episodio en el Congreso de Isabel Díaz Ayuso y a la posterior difusión del lema *«me gusta la fruta»*. Cuando Griso replicó que la presidenta madrileña reaccionó en privado sin saber que la captaban las cámaras tras oír acusaciones contra su hermano, Sánchez contraatacó preguntando si es humano *«ponerse una camiseta de me gusta la fruta»*, a lo que la presentadora cerró calificando el planteamiento del presidente de *«reflexión tramposa»*. ### La incógnita del 19-N antes de las urnas La comparecencia televisiva funciona como ensayo general de la estrategia socialista para el debate de mañana jueves: admitir desgaste y decepción en parte de su electorado por el caso Koldo, contraponer la ausencia de condena firme al PSOE frente a los antecedentes del PP y movilizar al voto progresista frente al bloque de derechas. Sin embargo, distinguir desde La Moncloa qué causas judiciales son legítimas y cuáles «ficticias» deja servido al líder popular uno de los flancos más sensibles del debate previo al 29-N. ### Fuentes consultadas - [El País (España — Crónica completa de la entrevista en Antena 3)](https://elpais.com/espana/elecciones-generales/2026-11-18/sanchez-en-antena-3-absoluta-tranquilidad-el-psoe-no-se-ha-financiado-irregularmente.html) - [Antena 3 Noticias (España — Fuente primaria de la emisión en *Espejo Público*)](https://www.antena3.com/noticias/espana/sanchez-admite-casos-corrupcion-gobierno-pero-defiende-psoe-nos-hemos-financiado-irregularmente_202611186afd674a103a0e0a474eb2ed.html) - [20minutos (España — Desglose de causas judiciales y Plus Ultra)](https://www.20minutos.es/nacional/sanchez-dice-que-ya-ha-asumido-responsabilidades-politicas-por-corrupcion-distingue-casos-reales-ficticios_7062500_0.html) - [Infobae / Agencia EFE (España — Declaraciones sobre el Tribunal de Cuentas)](https://www.infobae.com/espana/agencias/2026/11/18/sanchez-asegura-que-el-psoe-no-se-financio-irregularmente-y-considera-que-estan-en-cuestion-cantidades-nimias/) ~España ~Elecciones29N ~PSOE ~PedroSánchez ~Política
Sánchez niega financiación irregular del PSOE en Antena 3 y divide entre causas «reales» y «ficticias» a 24 horas del cara a cara
0
Discuss 5
A once días de las elecciones generales del 29 de noviembre y en vísperas del 25-N, una nueva controversia sobre el sistema Cometa de seguimiento telemático a maltratadores ha saltado al centro de la campaña electoral. La ministra de Igualdad y candidata socialista por Valladolid, **Ana Redondo**, afirmó este miércoles 18 de noviembre que las grabaciones difundidas por *ABC* —en las que supuestamente reconocía en una reunión interna que las pulseras antimaltrato basadas en geolocalización *«fallan en su esencia»*— *«probablemente están manipuladas»*. Horas después, el Partido Popular registró una solicitud de comparecencia urgente de la ministra ante la Diputación Permanente del Congreso de los Diputados. ### El origen del audio y la defensa de la ministra en Valladolid Según la exclusiva publicada por [ABC](https://www.abc.es/espana/castilla-leon/ana-redondo-defiende-puede-estar-manipulado-audio-20261118182728-nt.html), las declaraciones corresponderían a una reunión de trabajo celebrada el **21 de mayo de 2025** en la sede de **Securitas Seguridad España** en Madrid, adjudicataria del servicio de los dispositivos. En ese encuentro habrían participado, además de la ministra, la delegada del Gobierno contra la Violencia de Género, Zulema Altamirano; el consejero delegado de Vodafone España, José Miguel García; la consejera delegada de Securitas, Medora Miranda; mandos operativos, técnicos y representantes sindicales. En el fragmento difundido se escucha afirmar que el sistema basado en geolocalización falla en lo esencial y que esas deficiencias no se producían cuatro años atrás. Antes de intervenir en un acto de campaña en Valladolid, Redondo rechazó reconocer la grabación como propia: *«Yo, desde luego, no los reconozco, no sé de dónde han salido y en ningún caso son públicos»*. Aun así, la ministra distinguió entre incidencias técnicas puntuales y la protección efectiva a las usuarias. Admitió que en cualquier infraestructura tecnológica pueden surgir fallos de cobertura —comparándolos con los problemas recientes en aeropuertos, aviones o teléfonos móviles—, pero subrayó que *«lo que jamás ha fallado son los protocolos de protección»*. Según explicó, cuando se detecta una pérdida de señal se activa de inmediato el aviso policial y el contacto directo con la víctima, recordando que **5.000 mujeres** portan actualmente el dispositivo y que *«ninguna mujer con el dispositivo puesto y conectado ha sido asesinada»*. ### El trasfondo judicial: el contrato de 2023, la UCO y la Fiscalía Europea Más allá del choque de declaraciones en campaña, la polémica entronca con una investigación abierta que ya ha generado requerimientos policiales en el Ministerio de Igualdad. Como recogió días atrás la prensa regional en [Tribuna de Valladolid](https://www.tribunavalladolid.com/noticias/466341/ana-redondo-defiende-el-sistema-de-pulseras-antimaltrato-y-elogia-la-dignidad-politica-de-oscar-puente), la **Fiscalía Europea Anticorrupción** mantiene diligencias por presuntas irregularidades en la contratación financiada con fondos comunitarios, lo que motivó la entrada de agentes de la Unidad Central Operativa (UCO) de la Guardia Civil para recabar expedientes. En este punto, Redondo ha mantenido una línea de defensa constante desde el inicio de la campaña: desvincular políticamente a su equipo de la licitación investigada. La ministra insistió este miércoles en que el contrato que revisa la UCO data de **2023**, cuando el departamento estaba dirigido por su antecesora, **Irene Montero**, y recalcó que dicho expediente *«ya no existe»* al haber quedado extinguido y sustituido por un contrato posterior que, según Igualdad, refuerza el control sobre la empresa prestadora. ### Ofensiva del PP en la Diputación Permanente rumbo al 29-N La oposición ha convertido el episodio en uno de los ejes de presión parlamentaria de la recta final preelectoral, pese a encontrarse disueltas las Cortes. Tal como informan [Europa Press](https://www.europapress.es/epsocial/igualdad/noticia-pp-pide-comparecencia-urgente-redondo-congreso-pulseras-tolon-alquilar-vivienda-vpo-20261118193708.html) e [Infobae / Agencia EFE](https://www.infobae.com/espana/agencias/2026/11/18/pp-pide-explicaciones-a-redondo-tras-los-audios-sobre-los-fallos-en-pulseras-antimaltrato/), la portavoz popular en el Congreso, **Ester Muñoz**, calificó de *«absolutamente intolerable e indefendible»* que Redondo conociera presuntamente fallos estructurales desde mayo de 2025 sin haberlos hecho públicos y acusó al Ejecutivo de dejar desprotegidas a las víctimas. De forma paralela, el PP incluyó en su batería de peticiones ante la Diputación Permanente la comparecencia de la ministra de Educación, **Milagros Tolón**, para que explique las condiciones del alquiler de una vivienda de protección oficial (VPO) de régimen especial de su propiedad en Toledo, cuyo expediente analiza la Junta de Castilla-La Mancha tras admitir la ministra que no contaba con la autorización previa correspondiente, si bien su entorno sostiene que ha entregado toda la documentación para aclarar el trámite. ### Qué queda por aclarar antes de las urnas Con el Parlamento en funciones electorales, la viabilidad de una comparecencia inmediata dependerá de los equilibrios en la Diputación Permanente. Mientras tanto, siguen abiertas dos incógnitas clave que condicionarán el recorrido político y judicial del caso: por un lado, si Igualdad o las empresas implicadas someterán el audio difundido a un peritaje técnico que despeje si existió edición o descontextualización; y por otro, qué conclusiones arrojarán las pesquisas de la Fiscalía Europea y la UCO sobre la transición técnica entre el contrato de 2023 y el despliegue actual. ### Fuentes consultadas - [ABC (España — Declaraciones de Ana Redondo y reunión de mayo de 2025)](https://www.abc.es/espana/castilla-leon/ana-redondo-defiende-puede-estar-manipulado-audio-20261118182728-nt.html) - [Infobae / Agencia EFE (España — Petición de comparecencia y postura del PP)](https://www.infobae.com/espana/agencias/2026/11/18/pp-pide-explicaciones-a-redondo-tras-los-audios-sobre-los-fallos-en-pulseras-antimaltrato/) - [Europa Press (España — Diputación Permanente y frente paralelo sobre Tolón)](https://www.europapress.es/epsocial/igualdad/noticia-pp-pide-comparecencia-urgente-redondo-congreso-pulseras-tolon-alquilar-vivienda-vpo-20261118193708.html) - [Tribuna de Valladolid (España — Contexto del contrato de 2023 y diligencias de la UCO)](https://www.tribunavalladolid.com/noticias/466341/ana-redondo-defiende-el-sistema-de-pulseras-antimaltrato-y-elogia-la-dignidad-politica-de-oscar-puente) ~España ~Igualdad ~Elecciones29N ~Tribunales ~Política
Redondo alega que el audio sobre las pulseras antimaltrato «puede estar manipulado» mientras el PP exige su comparecencia
0
Discuss 7
Russia’s State Duma voted unanimously on November 18, 2026, to pass legislation restricting public access to books written or co-authored by individuals designated as **"foreign agents"** across the country’s libraries. Approving the bill in both its second and third readings by **380 votes to 0**—with zero abstentions and the entire parliamentary procedure wrapping up in roughly three minutes—lawmakers amended **Federal Law No. 78 of December 29, 1994 ("On Library Services")** without asking a single question of the Culture Committee lawmaker presenting the text. The rapid vote landed on the same legislative day the Duma advanced separate penalties of up to [5 million rubles for "discrediting traditional values"](https://www.mushimu.com/post/9624). ### How the Amendments Reshape Public Shelves Into Special Storage Under the amended library statute, access rules for municipal and institutional collections will now be governed directly by Russia’s "foreign agent" and "undesirable organization" frameworks. As reported by [Meduza](https://meduza.io/en/news/2026/11/18/russia-s-state-duma-approves-law-restricting-library-access-to-books-by-foreign-agents), [The Moscow Times](https://ru.themoscowtimes.com/2026/11/18/gosduma-odobrila-izyatie-knig-inoagentov-iz-bibliotek-a812395), and [UA.News](https://ua.news/en/world/derzhduma-rf-za-tri-khvilini-ukhvalila-zakon-pro-obmezhennia-dostupu-do-knig-inoagentiv), any book written by—or produced with the participation of—a registered "foreign agent" must carry an **"18+"** age restriction and display the author’s designation on its cover or publication metadata. Libraries and retail spaces are barred from displaying such volumes cover-forward or promoting them, effectively removing them from open browsing stacks into closed special-storage sections (*spetskhran*). Presenting the bill on the Duma floor, United Russia Culture Committee member Sergei Kovalev framed the measure as *"systematic work to counter the influence of individuals acting in the interests of foreign states."* Domestic Russian reporting by [Bryansk Today](https://bryansktoday.ru/article/275913) listed prominent contemporary writers swept into the registry’s scope—including Boris Akunin, Dmitry Bykov, Dmitry Glukhovsky, Lyudmila Ulitskaya, Mikhail Zygar, Vera Polozkova, Viktor Shenderovich, Mikhail Veller, Ivan Vyrypaev, and Linor Goralik—underscoring that the restrictions target some of Russia’s best-known novelists, historians, and playwrights. ### Why a 2024 Bill Suddenly Moved—and Why Librarians Fear Mass Purges The legislation had sat largely dormant for more than two years after being introduced in March 2024 by former Culture Committee chair Yelena Yampolskaya alongside Alexander Sholokhov, Yelena Drapeko, and Dmitry Pevtsov, and passing its first reading in June 2024. That long stall stemmed largely from practical alarms raised within Russia’s own publishing and library sectors over how broadly the law could sweep if applied literally. Data attributed to Russia’s Ministry of Digital Development and cited by *Kommersant* and [The Moscow Times](https://ru.themoscowtimes.com/2026/11/18/gosduma-odobrila-izyatie-knig-inoagentov-iz-bibliotek-a812395) previously warned that literal enforcement of "foreign agent" and "undesirable organization" bans could jeopardize **50% to 70% of all books held in municipal public libraries**, alongside up to **15% of academic library holdings**. In April 2026, Eksmo-AST publishing group president Oleg Novikov cautioned—as documented by regional outlet [NeMoskva](https://nemoskva.net/en/2026/04/14/zakon-ob-inoagentah-mozhet-opustoshit-rossijskie-biblioteki-napolovinu-knigoizdateli/)—that restrictions attach not only to primary authors, but also to translators, commentators, annotators, layout contributors, and 1990s–2000s cultural foundations or foreign universities now labeled "undesirable" that subsidized classic and regional editions. ### What Happens Next for Libraries and Readers With the bill clearing its final Duma hurdle unanimously, it moves to the Federation Council and presidential signature to enter force. Because Russian prosecutors and local officials have already pursued raids and ad hoc removals against booksellers and librarians since 2022, codifying the "foreign agent" access regime inside federal library law hands authorities a uniform statutory mechanism—leaving public librarians to decide whether to purge vast swathes of post-Soviet humanities collections or lock them behind 18+ special-collection counters. ### Sources & Further Reading - [Meduza (Independent Russian Press — Parliamentary Vote & Background)](https://meduza.io/en/news/2026/11/18/russia-s-state-duma-approves-law-restricting-library-access-to-books-by-foreign-agents) - [The Moscow Times (Independent / Europe — Law No. 78 & Ministry Impact Figures)](https://ru.themoscowtimes.com/2026/11/18/gosduma-odobrila-izyatie-knig-inoagentov-iz-bibliotek-a812395) - [Bryansk Today (Russia — Domestic Coverage & Affected Authors)](https://bryansktoday.ru/article/275913) - [UA.News (Ukraine — Legislative Mechanics & Kovalev Statement)](https://ua.news/en/world/derzhduma-rf-za-tri-khvilini-ukhvalila-zakon-pro-obmezhennia-dostupu-do-knig-inoagentiv) - [NeMoskva (Russian Regional Press — Eksmo-AST Publisher Warning)](https://nemoskva.net/en/2026/04/14/zakon-ob-inoagentah-mozhet-opustoshit-rossijskie-biblioteki-napolovinu-knigoizdateli/) ~WorldNews ~Russia ~Censorship ~Libraries ~StateDuma
Russian State Duma Passes Law in Three Minutes Restricting Library Access to Books by "Foreign Agents"
0
Discuss 6
A court in Beersheba ruled on November 18, 2026, to extend the detention without charge of Dr. Hussam Abu Safiya, the Palestinian pediatrician and director of northern Gaza’s Kamal Adwan Hospital, for an additional six months—pushing his imprisonment without trial toward a full two years. Arrested during an Israeli military raid on Kamal Adwan Hospital on December 27, 2024, Abu Safiya has been held under Israel’s **Unlawful Combatants Law**, a wartime administrative detention framework that allows renewable six-month incarcerations based on classified security material without formal criminal indictments. ### Inside the Beersheba Ruling and the Unlawful Combatants Framework Rejecting a release petition filed by defense attorney Nasser Odeh, the Beersheba court accepted the state’s contention that Abu Safiya *"poses a threat to the State of Israel,"* according to statements shared by his legal team and documented by [Al Jazeera](https://www.aljazeera.com/news/2026/11/18/israeli-court-extends-gaza-doctor-abu-safias-detention-without-charge) and [Anadolu Agency / ANews](https://www.anews.com.tr/middle-east/2026/11/18/israel-extends-detention-of-gaza-hospital-director-abu-safiya-by-6-months-without-charges). In court filings earlier this year, Israeli state attorneys maintained that his detention rests on intelligence alleging an affiliation with Hamas’s Military Medical Services—a claim Abu Safiya’s lawyers and international medical monitors reject as unsubstantiated criminalization of hospital administration during active hostilities. Speaking through his counsel after Wednesday’s hearing, Abu Safiya questioned why his order was renewed yet again, stating that he is being punished for carrying out his humanitarian and professional duties at Kamal Adwan. Legally, the extension underscores how Israel’s Unlawful Combatants Law—originally enacted in 2002 and heavily expanded since late 2023—operates outside standard civilian criminal procedure. Unlike ordinary prosecutions, detainees have no access to the underlying classified evidence cited against them, shifting the burden onto defense attorneys to disprove open-ended security assertions. On June 11, 2026, Israel’s Supreme Court similarly rejected a broader petition seeking the release of 14 detained Gaza physicians, including hospital directors and specialists held without trial. According to rights monitors cited by ANews, roughly 9,350 Palestinian detainees remain in Israeli custody, including dozens of healthcare workers from Gaza. ### Medical Alarms, Rakefet Isolation, and Calls for Independent Inspection Wednesday’s ruling drew immediate backlash from Israeli, Palestinian, and international medical organizations focused on Abu Safiya’s deteriorating physical condition. Last week, [Physicians for Human Rights Israel (PHRI)](https://www.phr.org.il/en/imminent-danger-to-dr-hussam-abu-safiyas-life/) reported that an independent review of Abu Safiya’s prison medical records showed marked weight loss, dropping hemoglobin levels, and uninvestigated chest pain—findings that followed an earlier decision by the Lod District Court denying a request for an independent outside medical examination. State authorities have previously told Israeli media that detainees receive periodic prison medical checks, though independent access remains blocked. Concerns over custody conditions intensified after July 2026 legal visits documented by [Novara Media](https://novaramedia.com/2026/11/18/israel-extends-arbitrary-detention-of-leading-palestinian-doctor/), when Odeh reported that Abu Safiya had been moved to Rakefet, an underground section of the Ramla prison complex, appearing severely weakened with visible bruising. That same month, the UN Office of the High Commissioner for Human Rights (OHCHR) urged Israel either to release Abu Safiya immediately or bring a recognizable criminal charge guaranteeing a fair trial, citing credible reports of prolonged solitary confinement, severe assaults, and denied medical care. Following Wednesday’s extension, Amnesty International UK called the renewal *"shameful,"* echoing an open appeal by more than 800 UK doctors urging London to press for independent medical access and unconditional release. ### Why the Case Matters Beyond One Physician Abu Safiya’s prolonged detention has become a central test case for medical neutrality and wartime administrative detention. Under international humanitarian law, hospital personnel retain protected status unless proven to have committed specific acts harmful to an adversary outside their medical duties—standards that human rights lawyers argue cannot be adjudicated credibly through secret evidence and perpetual six-month renewals. With no indictment filed nearly 23 months after his arrest, and no sign that Israeli courts intend to curb use of the Unlawful Combatants Law, diplomatic and medical pressure over detained Gaza health workers is set to sharpen further. ### Sources & Further Reading - [Al Jazeera (Qatar — Beersheba Court Ruling & OHCHR Background)](https://www.aljazeera.com/news/2026/11/18/israeli-court-extends-gaza-doctor-abu-safias-detention-without-charge) - [Novara Media (United Kingdom — Legal Visits & UK Medical Pressure)](https://novaramedia.com/2026/11/18/israel-extends-arbitrary-detention-of-leading-palestinian-doctor/) - [ANews / Anadolu Agency (Turkey — Supreme Court Context & Detention Figures)](https://www.anews.com.tr/middle-east/2026/11/18/israel-extends-detention-of-gaza-hospital-director-abu-safiya-by-6-months-without-charges) - [Physicians for Human Rights Israel (Israel — Independent Medical Record Review)](https://www.phr.org.il/en/imminent-danger-to-dr-hussam-abu-safiyas-life/) ~WorldNews ~Gaza ~Israel ~HumanRights ~Healthcare
Israeli Court Extends Gaza Hospital Director Hussam Abu Safiya’s Detention Without Charge for Six Months
0
Discuss 5
U.S. Defense Secretary Pete Hegseth and Venezuelan interim President Delcy Rodríguez announced on November 18, 2026, that Washington and Caracas have signed a bilateral agreement to formally re-establish military ties—ending a two-decade defense rupture dating back to 2005 under former president Hugo Chávez. Revealed during an unannounced trip to the Miraflores presidential palace in Caracas, the accord covers joint training, intelligence sharing, border security, and operational equipment access, marking the highest-level Pentagon visit to Venezuela since the January 3, 2026 U.S. military operation that captured Nicolás Maduro. ### From Confrontation to "Compliant Security Partner" Barely thirteen months ago, in October 2025, Caracas was condemning what it called Washington’s belligerent posture after U.S. combat aircraft operated roughly 75 kilometers off Venezuela’s Caribbean coast during anti-narcotics deployments. That trajectory snapped with Maduro’s capture in January—an operation led under Hegseth’s Pentagon that later triggered a House impeachment push against him, blocked earlier this month. Standing alongside Rodríguez on Wednesday, as reported by [Newsweek](https://www.newsweek.com/pete-hegseth-venezuela-visit-resources-delcy-rodriguez-12735771) and [CTV News](https://www.ctvnews.ca/world/article/hegseth-visiting-venezuela-says-the-nations-have-signed-a-deal-to-re-establish-military-ties/), Hegseth cast the turnaround in blunt strategic terms, describing Venezuela’s shift *"from our most aggressive regional threat to a compliant security partner."* Venezuelan reporting by [El Carabobeño](https://www.el-carabobeno.com/secretario-de-guerra-de-ee-uu-se-reune-con-delcy-rodriguez-en-caracas/) and state broadcaster VTV underscored the ceremonial weight of the Miraflores talks, where Hegseth was received by Foreign Minister Oliver Blanco and met alongside Venezuelan Defense Minister Gustavo González López, U.S. Deputy Assistant Secretary for Western Hemisphere Affairs Joseph Humire, and U.S. Chargé d’Affaires John Barrett. Rather than framing the deal around submission, Rodríguez—who had just returned to Caracas from a diplomatic tour of Turkey, Qatar, and the United Arab Emirates—told reporters she hoped *"technical-military cooperation ... can be brought to its highest possible level, so that ... we may address common threats."* That divergence in rhetoric highlights a delicate political balance: Washington showcases alignment and hemispheric enforcement, while Caracas’s interim leadership presents the accord to domestic audiences and external partners as sovereign, technical collaboration. ### Oil, Critical Minerals, and the Hemispheric Dominoes Military normalization in Caracas is tightly braided with resource extraction and a broader regional realignment. Hegseth’s trip follows a steady cadence of senior U.S. officials visiting Venezuela in 2026—including U.S. Southern Command chief Francis Donovan in February and May, Joint Chiefs of Staff Chairman Gen. Dan Caine in June, Energy Secretary Chris Wright in February and September, Interior Secretary Doug Burgum, and CIA Director John Ratcliffe. Hegseth explicitly linked security stabilization to economic access, telling Rodríguez that the Trump administration looks forward to *"unlocking together with you, the prosperity, the peace, the stability of Venezuela."* Under a parallel 25-year energy framework described by Rodríguez, 17 strategic oil fields are slated for joint development targeting more than 1.5 million barrels per day, pairing Venezuelan reserves and labor with U.S. capital and technology. Immediately after Caracas, Hegseth’s itinerary underscored how the Pentagon is knitting together a broader South American security corridor. As detailed by Spanish public broadcaster [RTVE](https://amp.rtve.es/noticias/20261118/jefe-pentagono-reune-caracas-delcy-rodriguez-anuncia-acuerdo-cooperacion-militar-venezuela/17320279.shtml), Hegseth traveled onward to Colombia on Wednesday to meet Defense Minister Jorge Eduardo Mora and deliver the first of several Lockheed C-130 Hercules tactical transport aircraft under Washington’s *"Shield of the Americas"* (*Escudo de las Américas*) counter-narcotics initiative—a rapid warming of Bogotá–Washington defense ties since right-wing President Abelardo de la Espriella took office in August. ### What Remains Unclear Despite the headline announcement, the Pentagon has not yet published the binding text, force-posture provisions, or oversight mechanisms of the U.S.–Venezuela defense agreement. Democratic lawmakers in Washington have also raised pointed objections to tying post-Maduro transition diplomacy so closely to long-term oil concessions without explicit congressional guardrails. Whether technical-military cooperation translates into durable institutional reform inside Venezuela’s armed forces—or deepens political friction at home and across Latin America—will depend on the specific terms Washington and Caracas choose to disclose in the weeks ahead. ### Sources & Further Reading - [Newsweek (United States — Caracas Remarks & Resource Framework)](https://www.newsweek.com/pete-hegseth-venezuela-visit-resources-delcy-rodriguez-12735771) - [El Carabobeño (Venezuela — Domestic Coverage & Delegation Details)](https://www.el-carabobeno.com/secretario-de-guerra-de-ee-uu-se-reune-con-delcy-rodriguez-en-caracas/) - [CTV News / AP (Canada — Bilateral Military Agreement)](https://www.ctvnews.ca/world/article/hegseth-visiting-venezuela-says-the-nations-have-signed-a-deal-to-re-establish-military-ties/) - [RTVE (Spain — Regional Tour & Shield of the Americas Context)](https://amp.rtve.es/noticias/20261118/jefe-pentagono-reune-caracas-delcy-rodriguez-anuncia-acuerdo-cooperacion-militar-venezuela/17320279.shtml) ~WorldNews ~Venezuela ~UnitedStates ~Defense ~LatinAmerica
U.S. and Venezuela Sign Pact to Re-Establish Military Ties During Hegseth’s Unannounced Caracas Visit
0
Discuss 4
Nvidia’s record quarterly earnings masked a much sharper financial balancing act inside its Securities and Exchange Commission filings: the chipmaker’s maximum potential exposure under AI infrastructure financing guarantees climbed by **$9.5 billion** in the third quarter to **$118 billion**, even as interim CEO and CFO Colette Kress publicly drew a hard boundary around when specialized cloud providers—known as neoclouds—can actually rely on Nvidia’s balance sheet. While Nvidia’s [$115.8 billion Q3 revenue](https://www.mushimu.com/post/9508) and Jensen Huang’s departure dominated initial headlines, Kress used her CFO commentary and Tuesday’s analyst call to warn that many AI cloud operators and frontier labs are expanding faster than their credit profiles can support, and that Nvidia’s guarantees do not serve as unconditional bailouts. ### How the $118 Billion Guarantee Stack Actually Works Investors anxious about circular vendor financing in the AI buildout had speculated about far larger off-balance-sheet liabilities, prompting Nvidia to break down the exact structure of its **$118 billion** exposure, up from $108.5 billion in the prior quarter. As reported by [International Business Times](https://www.ibtimes.com/nvidia-ai-infrastructure-guarantees-reach-118-billion-3809904) and [Trending Topics](https://www.trendingtopics.eu/nvidia-admits-many-a-i-customers-cannot-afford-their-own-growth/), Nvidia has paid **$0** to date under these guarantees because every dollar is gated behind strict construction and financing milestones. The overwhelming majority—**$105 billion**—is tied to SB Energy’s PORTS-Pike Technology Campus in Ohio, a 4.25-gigawatt complex spanning nine data center buildings slated to house Nvidia systems under 20-year leases for OpenAI, with an option for roughly 3.8 gigawatts of additional credit support. Crucially, each building’s guarantee activates only after construction finishes and the facility enters service—meaning the first Ohio building is not expected to trigger exposure until fiscal 2029—and the liability amortizes downward as OpenAI makes lease payments over two decades. The remaining **$13 billion** consists of **$9.5 billion** in newly added power and systems credit support guarantees plus **$3.5 billion** backing land, power, and building shells for third-party AI cloud operators. Separately, Nvidia disclosed **$32 billion** in forward commitments under neocloud "AI cloud agreements"—down from $36 billion the previous quarter—through which Nvidia agrees to backstop compute capacity in exchange for hardware sales and potential revenue-sharing from third-party rentals. In her CFO commentary, Kress stated plainly why Nvidia stepped into real estate and electrical grid procurement at all: *"Neoclouds and frontier labs are seeing extraordinary demand for AI infrastructure, yet many are growing faster than their balance sheets and long-term credit profiles can support."* ### Why Nvidia’s Language Spooked Wall Street—and Why Kress Held the Line During the earnings call, Bernstein analyst Stacy Rasgon pressed Kress on new filing language specifying that neocloud capacity commitments only activate once operators independently raise their own debt or equity, build the data center, and bring compute online. If a neocloud stalls on financing or runs into construction delays, the guarantee effectively never materializes. Kress rejected suggestions that Nvidia was abruptly rewriting existing contracts under post-Huang leadership, insisting: *"Is it new? No. Did we change anything? No."* Instead, she framed the disclosure as a deliberate effort to spell out guardrails that protect Nvidia from stranded, half-built projects. That distinction carries major operational consequences for smaller GPU cloud startups competing against hyperscalers. As [The Motley Fool](https://www.fool.com/investing/2026/11/17/nvidia-has-a-usd118-billion-promise-tucked-inside-its-blowout-quarter-here-s-what-it-means-for-the-stock/) noted, Nvidia’s disclosures simultaneously revealed **$518 billion** in total future corporate commitments—with **$222 billion** coming due in the next fiscal year alone. Rather than extending blank checks to cash-constrained cloud tenants, **$390 billion** of those commitments are locked into multiyear supplier prepayments, predominantly for HBM and memory chips (up sharply from $279 billion a quarter earlier), alongside **$74 billion** in future data center leases and **$28 billion** in cloud service agreements. Meanwhile, accounts receivable climbed to **$81 billion**, pushing average customer payment cycles from 60 to 64 days due to extended multi-quarter terms and Rubin shipment timing, while Nvidia absorbed a **$1.8 billion** impairment on non-marketable equity stakes. ### Selective Capital Discipline in the Post-Huang Era Asked by Cantor Fitzgerald analyst C.J. Muse how Nvidia will deploy future financial support, Kress signaled a narrower, infrastructure-first filter: Nvidia will selectively assist established partners with physical bottlenecks—land, electrical power, and facility shells—while scrutinizing tenant creditworthiness far more closely as alternative private-credit lenders enter the market. Reporting from Taiwan by [DIGITIMES](https://www.digitimes.com/news/a20261118VL205/nvidia-financing-yahoo-ceo-infrastructure.html) underscores the same strategic shift from an Asian supply-chain perspective, noting that under Kress, capital allocation ranks upstream silicon and memory prepayments alongside shareholder buybacks ahead of open-ended customer financing. For the broader AI ecosystem—already unsettled by [Nvidia’s recent leadership upheaval](https://www.mushimu.com/post/9451)—the message is clear. Hyperscalers and sovereign AI buyers can still tap long-term supply, and anchor projects like the Ohio OpenAI campus remain intact, but mid-tier neoclouds can no longer count on Nvidia to underwrite construction risk before their own financing closes. ### Sources & Further Reading - [International Business Times (US — Guarantee Breakdown & Earnings Call)](https://www.ibtimes.com/nvidia-ai-infrastructure-guarantees-reach-118-billion-3809904) - [Trending Topics (Europe — CFO Commentary Analysis)](https://www.trendingtopics.eu/nvidia-admits-many-a-i-customers-cannot-afford-their-own-growth/) - [DIGITIMES (Taiwan — Supply Chain & Capital Allocation)](https://www.digitimes.com/news/a20261118VL205/nvidia-financing-yahoo-ceo-infrastructure.html) - [The Motley Fool (US — Commitment Schedule & Financial Risk)](https://www.fool.com/investing/2026/11/17/nvidia-has-a-usd118-billion-promise-tucked-inside-its-blowout-quarter-here-s-what-it-means-for-the-stock/) ~AI ~Nvidia ~Neoclouds ~DataCenters ~Semiconductors
Nvidia Reveals $118B in Conditional AI Infrastructure Guarantees as Kress Sets Strict Rules for Neoclouds
0
Discuss 5
Indonesian studio Stairway Games and publisher Balor Games confirmed on November 18, 2026, that their island life simulation RPG *Coral Island* will officially launch on **Nintendo Switch 2** on **December 7, 2026**. Rather than shipping a standalone port months behind PC and other consoles, the Switch 2 debut arrives in lockstep with the free **Version 1.4 "Kids Grow Up" update**, which rolls out the same day across Steam, Microsoft Store, PlayStation 5, and Xbox Series X|S. According to [Gematsu](https://www.gematsu.com/2026/11/coral-island-for-switch-2-launches-december-7-alongside-version-1-4-update) and the [official Games Press announcement](https://www.gamespress.com/Coral-Island-Plants-New-Roots-on-Nintendo-Switch-2-December-7), the Switch 2 edition launches with full feature parity, including four-player online multiplayer with cross-play across all supported platforms. ### Why Stairway Games Skipped the Original Switch for Switch 2 For longtime backers of *Coral Island*’s 2021 Kickstarter campaign, the December 7 date closes a multi-year platform question. Stairway Games originally pledged a release on the first-generation Nintendo Switch, alongside PS5 and Xbox Series X|S following its October 2022 PC Early Access debut and November 2023 version 1.0 launch. In its [developer roadmap update](https://blog.stairwaygames.com/post/coral-island-nintendo-switch-2-the-kids-grow-up-update-arrive-this-winter), however, the studio acknowledged that "strict platform and performance limitations" made a viable port on the 2017 Switch hardware impossible without compromising the game’s dual land-and-underwater world. Rather than ship an unstable downgrade or leave crowdfunding supporters empty-handed, Stairway Games canceled the legacy Switch version outright and offered affected backers a no-cost upgrade to a Nintendo Switch 2 digital key—or an optional switch to Steam, PS5, or Xbox Series X|S—via BackerKit surveys distributed in mid-October, with unconfirmed selections defaulting to Steam keys after November 5 so every backer retained a playable copy. Targeting Nintendo Switch 2 directly lets Balor Games enter Nintendo’s holiday lineup with cross-play parity intact and dedicated **HD Rumble** support, matching DualSense haptics on PS5 and haptic feedback on Steam Deck. ### Inside Version 1.4: Four Life Stages, Town Progression, and Javanese Heritage While many farming sims treat player children as static crib decorations or permanent toddlers, Version 1.4 reworks Starlet Town’s generational clock. As detailed by Stairway Games and French outlet [Nintendo-Town](https://www.nintendo-town.fr/2026/11/18/coral-island-la-version-nintendo-switch-2-le-7-decembre/), players’ children now advance through four distinct developmental phases: baby, toddler, school-age child (matching the age tier of existing town children such as Zoe), and teenager. Progression is not purely visual. As children mature, players guide their emerging hobbies—spanning music, sports, baking, and gaming—which feed back into distinct teenage personalities, branching family dialogue, and spouse parenting cutscenes. Importantly, Stairway Games avoided the common genre pitfall of aging only the player’s household while freezing the surrounding village in time. Established Starlet Town youngsters—specifically Archie, Zoe, Oliver, and Valentina—also transition into teenagers in Version 1.4, unlocking fresh narrative events and social dynamics across the island. The update also deepens the studio’s Southeast Asian cultural identity by introducing a new family cutscene inspired by **Tedak Siten**, a traditional Javanese ceremony celebrating a child’s first steps onto the earth—a milestone that Stairway’s Yogyakarta-based developers drew directly from their own upbringing. ### Quality-of-Life Overhauls and Holiday Positioning Beyond family systems, Version 1.4 delivers practical refinements aimed at late-game players and newcomers arriving on Switch 2. An overhauled cooking interface adds recipe filtering by stat buffs and individual NPC gift preferences, removing much of the trial-and-error inventory friction common to large-cast life sims. Anglers gain decorative fishing bobbers unlocked through Fishing Mastery and sold at the Beach Shack, alongside broader environmental lighting upgrades and cosmetic additions. By bundling its largest social-progression patch with the Nintendo Switch 2 release in early December, *Coral Island* positions itself as a feature-complete cozy-sim anchor for Nintendo’s newer hardware during the peak winter window. Players picking up the Switch 2 edition on December 7 will jump directly into a post-1.4 ecosystem alongside veteran PC and console friends via cross-play, avoiding the staggered update lag that has historically plagued third-party console ports. ### Sources & Further Reading - [Gematsu (JP/Global — Switch 2 December 7 Launch)](https://www.gematsu.com/2026/11/coral-island-for-switch-2-launches-december-7-alongside-version-1-4-update) - [Stairway Games Official Blog (Indonesia — Developer Primary Source)](https://blog.stairwaygames.com/post/coral-island-nintendo-switch-2-the-kids-grow-up-update-arrive-this-winter) - [Games Press (Official Balor Games / Stairway Games Press Release)](https://www.gamespress.com/Coral-Island-Plants-New-Roots-on-Nintendo-Switch-2-December-7) - [Nintendo-Town (France — European Coverage)](https://www.nintendo-town.fr/2026/11/18/coral-island-la-version-nintendo-switch-2-le-7-decembre/) ~Gaming ~NintendoSwitch2 ~CoralIsland ~IndieGames ~CrossPlay
Coral Island Launches on Nintendo Switch 2 on December 7 Alongside Free 1.4 "Kids Grow Up" Update
0
Discuss 14
PlatinumGames’ 2019 Nintendo Switch action title *Astral Chain* now runs natively on x86-64 Linux and the Steam Deck at a locked 60 frames per second, bypassing traditional console emulation altogether. On November 18, 2026, developer captainj-sparrow shipped `v1.19.2` of **Astral Chain Recomp** (`astral-chain-pc-recomp`), billed as a "final performance update" targeting a flat 60 FPS across every combat and city scenario on Valve’s handheld. Independent hands-on testing published the same day by [SteamDeckHQ](https://steamdeckhq.com/news/astral-chains-pc-port-60-fps-on-steam-deck/) confirms that the native Linux AppImage holds 60 FPS through most encounters while averaging roughly 13 watts of total system draw—with only brief hitches inside densely populated headquarters hubs. ### How Static Recompilation Beats Emulation on Handheld APUs Unlike Switch emulators such as Ryujinx or Eden that interpret or JIT-compile guest ARM64 instructions alongside live shader translation on the fly, Astral Chain Recomp is an MIT-licensed static recompilation pipeline. According to the project’s [architecture documentation](https://github.com/captainj-sparrow/astral-chain-pc-recomp/releases/tag/1.19.2), it merges two distinct open-source foundations: the ARM64-to-C static lifter and NEON lowering layer from *Link’s Awakening* Recomp (`LAS Recomp`), and the Horizon OS (`nn::*` SDK symbol replacement, 38 supervisor calls, and NVN graphics front end) derived from MacNativeGames’ *Tears of the Kingdom* port. Instead of translating Maxwell GPU commands through an emulator’s dynamic shader cache mid-fight, the setup tool pre-compiles shader stage libraries during initial installation and replays NVN command buffers directly onto Vulkan 1.3. Eliminating both guest-CPU emulation overhead and runtime shader compilation explains why an aging Zen 2 and RDNA 2 APU can sustain double the Switch’s original 30 FPS cap within a tight handheld thermal envelope. Earlier in the `v1.19.1` release cycle, the developer noted achieving a steady 60 FPS in city zones on an ASUS ROG Ally Z1 Extreme throttled to four cores and 13W under Linux, while questioning whether Valve’s older Zen 2 silicon could match that ceiling without further GPU-side optimization. The subsequent `v1.19.2` build—packaged as a 259 MB `Astral-Chain-Setup-1.19.2-x86_64.AppImage` for Linux alongside Windows x64 binaries—closes that remaining bottleneck. ### Automatic Steam Deck Profiling and VRAM Discipline Running natively as a 64-bit Linux ELF binary over SDL3, PipeWire/ALSA, and Mesa RADV rather than through Proton allows the runtime to inspect hardware directly before the first frame renders. As detailed in the [Astral Chain Recomp Steam Deck and Linux Guide](https://astralchainrecomp.wiki/steam-deck-and-linux/), when `device_profile = auto` detects `SteamDeck=1` or Valve’s `Jupiter` (LCD) and `Galileo` (OLED) board identifiers under `/sys/class/dmi/id/board_name`, it automatically flips 15 configuration defaults tailored for a shared 16 GB unified memory pool. Crucially, the handheld profile re-encodes opaque ASTC textures into 4-bit-per-pixel BC1 (`astc_bc = bc1`) and skips top mipmaps above 1024 pixels, cutting texture VRAM consumption by roughly 200 MiB (75% to 87% savings depending on the scene). It also caps the resident texture cache at 512 MiB with background eviction triggered at 70% memory pressure to prevent SteamOS out-of-memory resets, halves local shadow resolution to 512, disables aggressive CPU boost so the GPU keeps its power share, and renders internally on a 16:9 `1280x720` grid expanded to `1280x800` via `aspect = wide`. Players looking to tune SteamOS or alternative handheld distributions can also consult Mushimu’s [SteamOS guide](https://www.mushimu.com/post/8729) and [Bazzite guide](https://www.mushimu.com/post/8728). ### Timing Fixes, Visual Upgrades, and Remaining Caveats Doubling the frame rate of a PlatinumGames action title notoriously breaks physics, parry windows, and cutscene sync if timing tables remain tied to 30 Hz. With contributions from `@capibawa`, the port decouples simulation speed so perfect dodges, Legion assault attacks, 360-degree spins, and voice-acted cutscenes stay synchronized at 60 FPS, backed by a custom low-latency pacing path on Linux alongside optional NVIDIA Reflex and AMD Radeon Anti-Lag 2 on Windows. Recent updates (`v1.18.0` through `v1.19.0`) also added HDR10 output, ultrawide camera scaling from 16:10 to 48:9, temporal upscaling via AMD FSR 3, Intel XeSS, and NVIDIA DLSS, screen-space bounce-light global illumination inspired by radiance cascades, and full two-player co-op support mapping dual Joy-Con inputs to two gamepads or gamepad plus keyboard and mouse. Two important caveats remain for prospective players. First, like every legitimate static recompilation project, the repository ships zero Nintendo assets or cryptographic keys: users must legally dump their own base `v1.0.0` copy, the `v1.0.1` update, and console `prod.keys` to compile the executable locally. Second, SteamDeckHQ flagged the unusually rapid commit cadence and noted that the upstream *TotK Recomp* base (`valentderah/totk-pc-recomp`) involved AI-assisted tooling—a reminder that while the end-to-end playthrough stability is well attested by testers, power users should verify release checksums and inspect build scripts carefully. Even with those caveats, *Astral Chain Recomp* stands as one of the clearest demonstrations yet of how native static recompilation can unlock locked-30-FPS Switch exclusives on handheld Linux PCs. ### Sources & Further Reading - [SteamDeckHQ (Hands-on Steam Deck testing)](https://steamdeckhq.com/news/astral-chains-pc-port-60-fps-on-steam-deck/) - [GitHub — captainj-sparrow/astral-chain-pc-recomp v1.19.2 Release](https://github.com/captainj-sparrow/astral-chain-pc-recomp/releases/tag/1.19.2) - [Astral Chain Recomp Wiki (Steam Deck & Linux Vulkan Guide)](https://astralchainrecomp.wiki/steam-deck-and-linux/) ~HandheldPC ~SteamDeck ~LinuxGaming ~AstralChain ~Recompilation
Astral Chain Static Recompilation Reaches 60 FPS on Steam Deck With Native Linux Vulkan Port
0
Discuss 14