Not if New Zealand refuses to go along! Just kidding!
What is the urge to join the EU by remote countries like Canada or now Australia?
I did not watch the 20 minutes long video. I suspect this is more idle talk than anything else.
Wednesday, September 16, 2026
Could Australia join the European Union? Really!
Could Genetic Therapies Finally Cure Cystic Fibrosis?
Good news! Only a few years away from application!
US House of Representatives passes third war powers resolution to limit military action in Iran
So far, it seems, these resolutions did not make much of a difference.
"The House on Tuesday passed a Democratic-backed war powers resolution that would limit President Trump’s military actions in Iran.
The lower chamber voted 220-204 to pass the resolution, ...
It was the third time the House has passed a similar war powers resolution, but the number of Republicans in support increased from four to seven. ...
“The administration hasn’t just continued to wage this illegal war in direct violation of the Constitution ..."
Kremlin Favored Russian General just Awarded Medal for Fake City Capture Dies in Action
Another Russian general bites the dust!
Defense Minister Andrei Belousov awards General Anton Grunis on August 28 Russia’s highest medal for bravery “Hero of the Russian Federation”
Chart of the day
France experiments with smaller tank units to evade drone attacks. Really!
Are tanks not almost as obsolete as knights in shining armor when gunpowder was used?
A company intends to mine uranium from the oceans
Good news!
"... The ocean has over 4 billion metric tons of uranium in it, enough for 50,000 years or so. ...
Fluxnium’s technology uses specially made polymer fibers that attract the uranium that’s dissolved in seawater. The company licensed the chemistry from the Department of Energy, and it has continued to work on the production and configuration of the fibers themselves.
The key has been to increase the surface area of the fibers, ... When the technology was demonstrated at a national lab, it could extract uranium at more than $200 per pound, which is more than twice the price of Western suppliers today. But by increasing the surface area, Fluxnium has been able to draw more uranium out of the water, bringing costs down. ..."
Scientists find various viruses in mosquito poop
Amazing stuff!
"... When the scientists sequenced the genetic material present in the mosquito excrement, they discovered a complete, high-quality genome of West Nile virus. The analysis also revealed evidence of a Hedwig virus, a pathogen that was first identified in Europe in a snowy owl and had never before been reported in the Americas, as well as several viruses that may be new to science."
From the abstract:
"Nearly 30% of emerging infectious disease events worldwide are transmitted by arthropod vectors, and this proportion continues to rise.
Rapid and accurate detection is critical for directing vector control interventions, thereby reducing the likelihood of widespread transmission.
Surveillance of infected mosquitoes can provide an early warning of impending human infection; however, conventional virus testing relies on processing large pools of mosquitoes and requires labor-intensive pre-processing. During rapidly developing epidemic or panzootic events, these delays may limit the effectiveness of public health responses.
Mosquito excreta has recently emerged as a promising alternative substrate for pathogen detection.
Sugar-fed mosquitoes regularly excrete gut contents, offering a rich source of nucleic acids. In this study, we developed and applied custom superhydrophobic excreta-collection funnels that efficiently aggregate excreta produced by field-collected Culex mosquitoes into attached microcentrifuge tubes.
Shotgun metagenomic sequencing of this material revealed a diverse RNA virome, including both globally distributed viruses and those reported here for the first time from the Americas.
Beyond virus detection, additional analyses enabled confirmation of host mosquito species and identification of trypanosomatid parasites, demonstrating the broader utility of mosquito excreta for integrated surveillance.
We anticipate that methods and devices of this type will become valuable components of vector surveillance programs, particularly in remote or resource-limited settings where repeated collections are challenging.
Overall, our findings highlight the potential of excreta-based monitoring to improve early detection of emerging or unknown pathogens of One Health importance, refine our understanding of mosquito virome biogeography, and facilitate the discovery of previously undescribed viruses."
Mosquito Waste Reveals Viruses Hiding in Plain Sight (original news release) "Mosquitoes can carry viruses that threaten human health, making surveillance and prevention a continuing public health priority."
Dana Price sets up a ‘mosquito toilet’ for disease surveillance.
JudgeGPT Experiment Boosts Case Closures in Pakistan Courts since 2024
Amazing stuff!
The situation in Pakistan: "With a backlog of 2.26 million cases and fewer than two judges per 100,000 people—compared to 22 in the EU and eight in Brazil"
"... in Pakistan, a large-scale trial of a specially designed AI tool for judges found the technology—together with appropriate training–boosted the number of cases resolved by 6.3 percent with no obvious drop in the quality of judgments. ...
They built a custom tool combining OpenAI’s GPT-4 large language model (LLM) with a knowledge base of nearly 130,000 Pakistani judicial opinions and statutes, to help judges with legal research and drafting judgments.
They began offering the tool in 2024 to 1,559 trial judges—roughly half the country’s justices. ...
AI tools for judges are already being rolled out in Brazil and India ...
Some judges were also already using AI chatbots, Mehmood says, but commercial offerings performed poorly on Pakistani legal queries, frequently hallucinating case law. ..."
AME Agent Swarms Quietly Rewrite the software development life cycle Workflow at AMD
Amazing stuff! Impressive!
"The impact of AI on software development has been both profound and ever-evolving. Last year, I wrote about AMD’s plans to use AI not just for generating new lines of code, but also for other steps in the software development lifecycle (SDLC), such as triaging problems, debugging code, and testing the software. At the time, we were hoping for a 25 percent productivity boost from AI use over the course of two or three years.
But with each new release, the capabilities of large language models (LLMs) improve dramatically—accelerating software development, increasing the quality of AI-generated code, and fundamentally reshaping how software is engineered.
Now, just one year later, we have surpassed our productivity target, achieving a 30 percent overall productivity boost through AI. On top of that, we are rethinking not only how we use AI within the SDLC, but the structure of the SDLC itself. ...
AMD began developing AI systems for code generation, testing automation, bug analysis, and code review in 2024. At the time, our objective was to achieve 25 percent AI-generated production code by 2027 while gradually automating larger portions of the SDLC. ...
By this metric, we have crossed the 20 percent mark at the beginning of this year and are now progressing towards 50 percent across entire codebase. In some software components, more than 80 percent of the code is now generated using AI. ...
Agentic AI has enabled us to include AI in every step of the life cycle:
For code analysis and triage, agents are trained to analyze problem reports, identify and group similar requests, and highlight which code snippets are likely to need modification.
For debugging and code generation, agents are directed to analyze a bug request and implement required code changes.
For testing, the agents generate unit tests, and if those are passed, identify necessary integration and product-level tests.
And finally, for the approval and release stage, agents prepare architecture summary, code change review, and full test results for engineers’ review and approval—and, if approved, integrate the changes into the next release. ...
A good example is our AI-driven effort to resolve issues in our Radeon Software eXperience (RSX). RSX is a user interface component that allows users to configure and monitor graphics driver behavior.
In October 2025, we began using AI agents to automatically debug and fix reported RSX issues. Out-of-the-box AI tools delivered limited results, resolving only 6 percent of issues. ...
At the same time, advances in models and agent run-times further increased effectiveness. Together, these improvements significantly increased our resolution rate from 6 percent to more than 75 percent of RSX issues resolved by agentic loop. ..."
Virtual cell model predicts best cancer drug
Good news! Cancer is history (soon)!
Again Nature journal uses the horrible ideological term "people" instead of women!!!
triple-negative breast cancer (TNBC)
"A new ‘virtual cell model’ could help to choose more personalized treatments for women people with hard-to-treat triple-negative breast cancer. Researchers trained the prototype artificial-intelligence-based model, called ProteinTalks, on millions of protein measurements collected over time from breast cancer cell lines that were treated with dozens of antitumour drugs. The model did a good job of predicting how cells would respond to drugs it hadn’t seen before and also real-world clinical outcomes. “This is the first time that a virtual cell model goes out of the laboratory and is tested in a clinical scenario,” ...
In experiments using biopsies taken from individuals with TNBC, the virtual cell model showed promising results in identifying the same drugs that would prove to be effective when given to women people. The authors say this raises the possibility of more personalized care for TNBC, which accounts for 15–20% of breast cancer cases. ...
In their study, Guo and his colleagues aimed to develop a virtual cell model of breast cancer cells using proteomics data. They trained an AI model on more than 38 million protein measurements collected from 18 breast cancer cell lines, 16 of which were TNBC cells. The researchers treated the cells with 63 antitumour drugs that have been approved by the US Food and Drug Administration, and 59 drug combinations. ..."
From the abstract:
"Artificial intelligence-empowered virtual cell models represent an emerging approach for in silico drug discovery, yet most existing approaches lack large-scale, time-resolved perturbation proteomics data and interpretable frameworks for predicting therapeutic responses.
Here we generated more than 38 million temporal protein-abundance measurements from systematically perturbed breast cancer cell lines, and developed ProteinTalks, a virtual cell model.
Central to ProteinTalks is the synergy of this large-scale dynamic proteomic resource and the model architecture, enabling a new pretraining framework that learns transferable dynamical latent representations from temporal proteome trajectories. By modelling how proteins respond conditionally to different perturbations, this approach enables the model to function as an operational tool for diverse drug discovery tasks: predicting drug efficacy and synergy, discovering new drug combinations, probing proteins associated with drug resistance, stratifying patient responses and prioritizing drug candidates for patient organoids.
It also shows robust transferability, extending beyond cell lines to patient-derived organoids and clinical biopsies, generally achieving higher performance than the selected benchmark implementations under the evaluated protocols.
Together, ProteinTalks shows how scalable pretraining of transferable dynamic representations enables operational, dynamics-aware, proteomics-based virtual cell models to advance in silico drug discovery."
AI model predicts which breast-cancer drugs work best (behind paywall) "Model trained on millions of protein measurements can gauge drug effectiveness in tissue samples taken from people with triple-negative breast cancer."
An operational perturbation proteomics-based virtual cell model (no public access, but article above contains link to PDF)
A perturbation proteomics-based foundation model for virtual cell construction (preprint, open access)
Fig. 1 Overview of this perturbation proteomics study.
Fig. 3 Development and performance of the ProteinTalks model.
Scientists engineer ready-to-use cancer-fighting T cells for solid tumors
Good news! Cancer is history (soon)!
"Key takeaways
- ... researchers have developed a way to mass-produce cancer-fighting T cells from blood stem cells found in cord blood, engineered to target a protein found in many solid tumors — creating uniform batches instead of custom treatments for each patient.
- This way, scientists ensure that all the T cells they produce carry the same single tumor-targeting receptor, avoiding the random receptors that can attack healthy tissue and cause a dangerous condition called graft-versus-host disease.
- In mouse models of ovarian cancer and melanoma, a single dose kept tumors in check and extended survival, outperforming conventional donor T cells, which caused toxic side effects in the same tests.
...
Rather than starting with mature, donor-derived T cells, the researchers began a step earlier — with blood stem cells found in cord blood, which naturally give rise to every type of blood and immune cell. They then added a gene for a receptor that targets NY-ESO-1, a protein found in many solid tumors. Fragments of NY-ESO-1 get pushed to the outer surface and displayed there, name-tag style. The researchers then grew these engineered stem cells into T cells in the lab. ...
“From a small number of cord blood stem cells, we can generate trillions of therapeutic cells — enough for thousands of doses — within about six weeks,” ... “At an estimated $5,000 per dose, this approach would be far more accessible than today’s therapies.” ..."
From the highlights and abstract:
"Highlights
• HSPC-derived AlloESO-T cells enable scalable, feeder-free T cell manufacturing
• AlloESO-T cells overcome autologous complexity, HLA restriction, and donor variability
• AlloESO-T cells display tumor homing, durable persistence, and TCR/NKR dual targeting
• AlloESO-T cells show minimal GvHD/CRS risk and stable hypoimmunogenicity
Summary
Adoptive T cell therapy for solid tumors is limited by autologous manufacturing complexity and, in allogeneic settings, risks including graft-versus-host disease (GvHD), HLA restriction, and donor variability.
We develop a scalable, feeder-free platform to differentiate gene-engineered hematopoietic stem and progenitor cells (HSPCs) into allogeneic, NY-ESO-1-specific cytotoxic T (AlloESO-T) cells.
Product phenotype, function, tumor homing, and safety are assessed against solid tumor models and benchmarked to peripheral blood mononuclear cell (PBMC)-derived TCR-engineered T cells. AlloESO-T cells display a uniform cytotoxic phenotype, with dual tumor targeting through a transgenic TCR and natural killer receptors.
Relative to PBMC-derived counterparts, AlloESO-T cells show superior cytotoxicity, selective solid-tumor homing, durable killing persistence, and resilience to immune evasion.
They also maintain low GvHD and cytokine release syndrome risk, while retaining stable hypoimmunogenic features. These findings establish HSPC-derived AlloESO-T cells as an off-the-shelf, mono-specific cytotoxic T cell therapy with scalable manufacturing, enhanced efficacy, and improved safety, which support broad applicability of AlloESO-T cells across solid tumors."
Scalable generation of hematopoietic stem cell-engineered off-the-shelf mono-specific cytotoxic T cells targeting solid tumors (open access)
Graphical abstract
Figure 1 Generation and characterization of Allo/15ESO-T cells
Data Scarcity and Model Sparsity: Mixtures-of-Experts Overfit More to Repeated Data
This could be an interesting new paper by Luke Zettlemoyer and his team!
From the abstract:
"As the supply of human-written text is exhausted, it has become standard practice to repeat language model training data.
Prior work has studied data repetition for densely activated Transformers, but the effects of data repetition remains largely unexplored for recently dominant sparse architectures such as Mixture-of-Experts (MoE), despite their increased compute efficiency.
We vary data repetition rates across single- and multi-domain data mixes, and across MoE settings, including expert count and granularity. We consistently find, for models ranging from 80M to 1B active (8.5B total) parameters, that MoEs degrade more rapidly under data repetition.
This effect increases with sparsity, dictated by total rather than active parameters. While 80M dense models can repeat data over 8x with minimal degradation, MoEs instead begin to suffer at 4x, and deteriorate rapidly, ceding their performance benefits in all-unique data settings to underperform dense models after 32x.
We experiment with existing regularization methods as a potential remedy. We find that some methods, such as dropout, can mitigate overfitting. In particular, with strong masking-based regularization, MoEs are able to outperform dense models even when data is repeated more than 64 times. However, no method fully matches the performance of all-unique training data.
Finally, we analyze internal mechanisms correlated with MoE overfitting in high repetition regimes, and find that MoE routing universally stabilizes early in training, and that expert specialization correlates with overfitting to repeated data. In sum, our work addresses the adverse interactions between sparsity and data repetition: we present evidence for the core mechanisms of overfitting and its potential remediation, and suggest promising avenues for future methods to reduce over-specialization in model parameters by disrupting memorization patterns."
Wild dogs make the longest recorded journey of any terrestrial mammal in Africa
Amazing stuff!
"... Researchers from the Zambian Carnivore Programme and Montana State University wanted to understand how this endangered predator navigates human-dominated landscapes to better help with conservation efforts. So they tracked groups of wild dogs in Zambia after they separated from their packs.
In doing so, they recorded a record-breaking journey. Three of the dogs traveled a staggering 4,126 kilometers (2,564 miles) across the country, the longest journey ever recorded for an African land mammal. ..."
Extraordinary movements connecting distant ecosystems by African wild dogs (Lycaon pictus) (open access)
African wild dog (Lycaon pictus) in the Okavango Delta, Botswana.
Tuesday, September 15, 2026
The Overpopulation Panic Dates Back to Odysseus’s Time
Recommendable! Food for thought!
"... In Greek mythology, the Trojan War was a deadly conflict that Zeus orchestrated to fight human overpopulation. According to a surviving fragment of The Cypria (a prequel to Homer’s Iliad and Odyssey):
There was a time when the countless tribes of men, though wide-dispersed, oppressed the surface of the deep-bosomed earth, and Zeus saw it and had pity and in his wise heart resolved to relieve the all-nurturing earth of men by causing the great struggle of the Trojan War, that the load of death might empty the world. And so the heroes were slain in Troy, and the plan of Zeus came to pass. ..."
"... A study published recently in the journal Sustainable Development recommends that humanity aim to more than halve the global population, to about 4 billion, in order to improve long-term sustainability." What a nonsense!
The EPA repealed rules limiting greenhouse gas emissions from coal and natural gas power plants
Good news! Bravo President Trump! Global warming is a hoax and climate change is a religion!
"... a move the agency said would help “unleash” American energy. Environmental groups vowed immediate court challenges."
"The EPA announced ... a final rule repealing regulations on carbon and toxic emissions from fossil fuel plants. The action would represent one of the Trump administration’s biggest blows yet against emissions reduction regulations put in place by Democrats. ..."
Downed in Iran US Airman Reveals Details of his Rescue
What a story!
"An Air Force officer shot down over Iran in April described his 50-hour fight to survive in his first public interview Sunday. Identified only by his call sign, Bravo, he fell at up to 100 mph after an Iranian missile shredded his parachute.
He prayed as he fell: “Lord, your will be done. But if I’m getting through this, I need help.” Bravo broke his back, an arm, and a shoulder, then climbed a 7,000-foot ridgeline as Iranian forces came within a few hundred meters.
His rescue led to the U.S. putting more than 90 troops on the ground in Iran for the first in 46 years, aided by a CIA deception campaign. When their planes got stuck in the desert sand, the rescue force destroyed the aircraft rather than leave them behind.
Bravo was pulled out at sunrise on Easter Sunday, April 5."
Israel and Saudi Arabia discuss intelligence aid amid growing Houthi attacks, British advisers are to assist Saudi Arabia
Good news! When will Saudi Arabia sign the Abraham Accords?
"... Saudi Arabia also asked Britain for assistance in repelling the Houthis from Bab al-Mandab Strait and in helping defend against strikes on its oil infrastructure, Bloomberg reported.
British Prime Minister Andy Burnham has agreed to send British military advisers to assist Saudi Arabia, but he has not yet committed to sending any additional aid, the report said. ..."
How Fairphone built the ethical, repairable Fairphone Gen 6+
This alternative smartphone is not cheap!
"... Fairphone, a Dutch smartphone maker that has just entered the US market, thinks about phones differently.
The new $650 Fairphone Gen 6+ is designed both for longevity and easy repairability. ..."
62,000 People Arrested for Speech Crimes in Britain over Past Five Years
Concerning! Bad news! The UK used to be a beacon of free speech!
"Censorship laws such as the Malicious Communications Act 1988, the Communications Act 2003, and the Online Safety Act 2023 have been used by both Labour and so-called Conservative governments to arrest 62,199 people for “communication offences”, including offensive jokes online, between January 1, 2021 and December 31, 2025, the Big Brother Watch pressure group has revealed.
The report was based on police data from 38 of the 45 territorial police forces in the UK, obtained following a series of Freedom of Information requests sent to all forces. However, the civil liberties group noted that because some forces failed or refused to provide their data, the true number of people arrested for speech crimes was likely much higher. ..."
An experiment that recreates coronal mass ejections in the lab
Amazing stuff!
"Some of the most iconic images of the Sun have captured the explosive force of coronal mass ejections (CMEs), in which huge clouds of magnetized plasma erupt from the solar atmosphere. But CMEs have only rarely been sighted on other stars, a discrepancy that has puzzled astrophysicists for many years.
Now an international research team has devised an experiment that recreates these stellar eruptions in the lab, providing clear evidence that CMEs can be suppressed by strong magnetic fields. The same experimental approach could also be used to investigate the role that magnetic confinement plays in other stellar activity. ..."
"... The current study ... combines astrophysical simulations, high-energy laser-plasma experiments, and advanced 3D plasma numerical modelling. The results show that strong magnetic fields of active stars can completely suppress CMEs before they escape into space. “The unique combination of theory, laboratory tests, and numerical simulations provides the first experimental evidence supporting a long-standing prediction that stellar magnetic fields can confine these large-scale eruptions.” ..."
From the abstract:
"Solar coronal mass ejections (CMEs) are routinely observed, but as of yet there exist few convincing detections of stellar CMEs. A reason for this could be that the stronger magnetic fields of these stars, compared to that of our Sun, would prevent CME to form and escape.
Here we combined astrophysical simulations, measurements of scaled high-energy laser-driven plasma flows, and 3D magnetohydrodynamic modeling to test this hypothesis.
Simulations show that, in a 100 G stellar dipole field, low-plasma
In the laboratory, a laser-produced plasma stream scaled to stellar CME conditions propagates freely at low applied magnetic fields (approximately 30 G stellar equivalent), but becomes unstable and halts entirely when the field is increased to
Analytical estimates and numerical simulations suggest that the sudden disruption of the flow is induced by a kink instability. Even though the laboratory environment does not fully reproduce the complex magnetic flux-rope topology of CMEs, these results provide the first laboratory-scale evidence that strong stellar magnetic fields can fully suppress CME propagation, offering a physical explanation for their lack in stellar observations and highlighting the role of magnetic confinement in stellar evolution and exoplanet space weather."
Stellar eruptions in the laboratory: First experimental evidence for their suppression in strong magnetic fields (original news release)
Experimental Evidence for Coronal Mass Ejection Suppression in Strong Stellar Magnetic Fields (no public access)
Experimental evidence for coronal mass ejection suppression in strong stellar magnetic fields (preprint, open access)
Computer simulation of a coronal mass ejection for a star (green) with a strong magnetic field (lines). The eruption is confined by the magnetic field and thus does not escape the star.
A Theoretical Analysis of Generalization Dynamics in Neural Networks under Gradient Descent with Weight Decay
This could be an interesting new theoretical paper by Mikhail Belkin and his team!
From the abstract:
"Understanding generalization remains a central challenge in machine learning because it requires jointly considering data, architecture, and training dynamics.
In this paper, we develop a theoretical framework that characterizes how these factors jointly shape generalization performance throughout training.
More precisely, we study a broad class of neural networks trained under the ℓ2 loss by gradient descent (GD) with weight decay, and prove the convergence of GD to a neighbourhood of the global minimizers of the empirical loss.
By partitioning the space based on the input data, we then decompose the population error into data error, optimization error, and prediction variation error, and bound them separately.
In particular, for the prediction variation error, which measures the oscillations of the learned function, we propose (local) approximate homogeneity and derive explicit cellwise and layerwise bounds for its evolution along the training trajectory.
These bounds yield two important implications: a necessary condition of improved generalization explains differences in layerwise generalization behavior; a sufficient condition describes delayed generalization and provides a theoretical characterization of grokking."
Bottom-up Modeling of Repeated Elements via Single Image Analysis-by-Synthesis
This could be an interesting new paper by Alexei A. Efros and his team!
From the abstract:
"We address the problem of discovering repeated elements from a single image. In contrast to existing approaches that depend on large annotated datasets, curated multi-image collections, or object segmentation masks, we show that a single image can suffice to learn a meaningful object model in a completely bottom-up fashion, without any prior knowledge beyond a coarse scale prior.
Our method learns a tunable image-space prototype of the repeated elements through a reconstruction objective, enabling the model to identify and synthesize consistent object instances within the same image.
Experiments on 116 real images from the FSC-147 dataset demonstrate that our method successfully learns coherent element models and captures intra-category variation on challenging images.
Qualitative results reveal superior reconstructions and interpretable decompositions compared to classical decomposition, joint alignment, and 3D object modeling methods, while maintaining a simple 2D formulation. These results suggest that meaningful object discovery can emerge from single image learning alone."
The Yemeni Houthis are about 350,000 strong!
I bet you did not know that!
Why this fact so rarely mentioned in Western news media is strange to say the least!
Who allowed this terrorist group to become so strong?
How much money and support is Iran giving the to Houthis?
"The group's manpower can be broken down into three main categories:
Core Militants and Veteran Troops: The group maintains a core faction of 20,000 to 40,000 highly trained, war-hardened militants. This core is heavily augmented by formal army units and commanders who remained in northern Yemen following the country's civil fragmentation, giving the Houthis access to professional military expertise.
Active Standing Army: According to official administrative records in Sana'a and independent intelligence reports from the Middle East Institute, the Houthis command an active-duty and affiliated force of over 200,000 soldiers divided across seven military regions.
Massive Recent Mobilization: The Houthis capitalized heavily on public sentiments regarding the Gaza conflict and reactions to Western airstrikes in the Red Sea. This allowed them to conduct a massive recruitment drive, ballooning their total roster to 350,000 personnel.
..." (Google Search result)
Monday, September 14, 2026
US Weapons Already In Orbit; Russia, China Brace For Space War?
It was expected to happen for some time! The militarization of space goes on!
How much have Russia and China already deployed?
Subscribe to:
Posts (Atom)