Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Get the latest international news and world events from around the world.

Log in for authorized contributors

100 scientific papers I’ve read in full over the past year

Ever since high school, I’ve had a personal tradition of sharing the scientific papers I read (in full) with my online community. About a year ago, I started cataloguing my posts and periodically sharing batches of papers here on Substack. So far, I have made 16 Substack posts, each with 5–10 papers and my comments. In total, this has resulted in a compilation of 100 papers and commentaries. Together, these papers represent a massive accumulation of human knowledge, progress, and innovation. Millions more exist out there on the internet. I will acknowledge that there’s a lot wrong with the academic publishing system and that it deserves massive reform. But seeing all of this amazing work at once nonetheless gives me hope that the collective intellect of the human species is capable of great things.

I’ve kept track of the 100 scientific papers that I’ve read in full over the past ~1 year, writing short summaries and posting them on social media. Here is the full compilation!

Mission · Oak Lab

Before starting Oak Lab, both Sutton and Javed were working at Keen Technologies—an AGI startup founded by legendary game developer John Carmack. They chose to break away to pursue a fundamentally different path toward understanding intelligence. Sutton bluntly describes current deep learning methods as “weak and inefficient,” arguing that today’s AI models are hitting a wall because of how they learn.

Today’s frontier models (like ChatGPT or Claude) are trained on massive, static, pre-collected internet datasets. Sutton argues this is “learning from someone else’s experience.” Because these datasets are frozen, the models cannot independently discover truly new knowledge, adapt in real time, or evaluate their own outputs.

Sutton’s model explicitly shifts away from the “turn-based” prompt-and-response loop of modern LLMs. By running the FC-STOMP cycle continuously on streaming data, Oak Lab expects to build agents that discover optimal, creative survival and problem-solving strategies that completely bypass the limits of human intuition.

OaK architecture discovers temporal abstractions grounded in experience that are both self-verifiable and useful for planning.

Biodegradation of polyethylene by the marine fungus Parengyodontium album

Year 2024 Marine fungus that eats plastic.

Plastic pollution in the marine realm is a severe environmental problem. Nevertheless, plastic may also serve as a potential carbon and energy source for microbes, yet the contribution of marine microbes, especially marine fungi to plastic degradation is not well constrained. We isolated the fungus Parengyodontium album from floating plastic debris in the North Pacific Subtropical Gyre and measured fungal-mediated mineralization rates (conversion to CO2) of polyethylene (PE) by applying stable isotope probing assays with 13 C-PE over 9 days of incubation. When the PE was pretreated with UV light, the biodegradation rate of the initially added PE was 0.044%/day. Furthermore, we traced the incorporation of PE-derived 13 C carbon into P. album biomass using nanoSIMS and fatty acid analysis.

Gravitational waves reveal hidden populations within black hole mergers

Since gravitational waves were first detected in 2015, instruments including LIGO, Virgo and KAGRA have picked up a steady stream of signals from colliding black holes, building a catalog that now numbers in the hundreds. Yet despite this wealth of data, a fundamental question has remained stubbornly unresolved: How do these black holes actually form?

Now, two independent research teams have used fresh theoretical approaches to comb through the data, and both arrived at a similar conclusion: Merging black holes don’t form a single uniform group, but instead separate into distinct subpopulations, each bearing the fingerprints of different formation mechanisms. Both studies have been published in Physical Review Letters.

World-first neutron lens brings sharp focus to structures inside materials and objects

Researchers at Paul Scherrer Institute (PSI) have developed the world’s first achromatic lens for neutron imaging. The lens overcomes a longstanding obstacle in the field: focusing neutrons of different wavelengths well enough to form a sharp, magnified image. With the lens, researchers can now image thick samples and follow processes inside bulky equipment such as furnaces, cryostats or pressure cells.

Neutrons can provide unique insights into the structure of materials—but they are hard to manipulate. Neutrons, like X-rays, are produced as a beam at research facilities such as the Swiss Spallation Neutron Source SINQ and are used to image inside materials and objects. Unlike X-rays, however, neutrons can penetrate deeply into many metals while remaining highly sensitive to light elements such as hydrogen and lithium. In this way, they can be used to observe oil, polymer or lithium distribution inside dense metallic structures such as engines or batteries, reveal water uptake in plants or nondestructively examine priceless archaeological artifacts.

Yet the same weak interaction with matter that makes neutrons such a useful tool also makes them notoriously difficult to deflect or focus—a fact that has limited the development of advanced imaging techniques. Now, PSI scientists have reported in Nature Communications a new type of lens that overcomes this barrier.

Cytoskeletal remodeling with rhythmic changes lead to axon development

The researchers have uncovered a mechanism that determines why a neuron usually forms a single, long extension called “axon” – a phenomenon that is fundamental to how our brain functions. Contrary to the common view that external cues drive axon formation, the team of scientists comes to the conclusion that its growth originates primarily inside the cell. Their work, based on cell cultures and was published in the journal “Nature” reveals how a neuron’s structure is remodeled to generate the axon.

Neurons in the brain and spinal cord form a vast network in which each cell receives many inputs but sends output through only a single, long extension: the “axon”. “If our neurons had multiple axons, this would cause chaos in the brain,” says the senior author. “Nature has therefore found a clever way to make sure that neurons generate only one axon. This applies not only to humans, but across the entire animal kingdom. So, we’re dealing with very fundamental processes that shape the wiring of the brain and nervous system.”

Whole cross-sectional human ultrasound tomography

An impressive paper where entire human torso cross sections were imaged via ultrasound. I’m also curious if this has any ties to Midjourney’s announcement of a similar-sounding technology in their new biomedical division.

An ultrasound tomography system composed of a custom 512-element circular receiver array combined with a single-element transmitter that rotates around the participant enables whole cross-sectional ultrasound imaging of the human body.

/* */