Editorial Team

LLMs read tokens, not letters. See how byte-pair encoding builds a vocabulary, and why that explains strawberry miscounts, shaky arithmetic, and uneven costs.

Neural scaling laws explained: what Kaplan (2020) and Chinchilla (2022) found about parameters, data, and compute, and where the power laws stop working.

Mechanistic interpretability reverse-engineers neural networks into features and circuits. What it has found in GPT-2 and Claude, and what it cannot yet do.

Horizontal gene transfer lets microbes pass genes sideways through transformation, conjugation, and transduction, and it blurs the tree of life.

Apoptosis is programmed cell death. Learn how caspases, mitochondria, and Bcl-2 proteins carry it out, and what goes wrong in cancer and autoimmunity.

Prions are misfolded proteins that spread by converting normal ones, with no DNA or RNA. How scrapie, kuru, and mad cow disease revealed the mechanism.

Telomeres are repeat-DNA caps on chromosome ends that shorten with each cell division. See how telomerase works and what telomere length does not prove.

The Cambrian explosion was a burst of animal diversification about 539 million years ago. Here is the fossil evidence, the timing, and the main explanations.

Christiaan Eijkman's chickens in Java developed beriberi on white rice and recovered on brown, starting the path to vitamin B1 and a 1929 Nobel Prize.

How torsion balances, the Apollo 15 feather drop, and the MICROSCOPE satellite tested whether all matter falls alike, now to a few parts per quadrillion.

In 1912 Victor Hess carried electroscopes to 5,350 m in a balloon and found radiation rising with height. Here is how the flights worked and what they proved.

On April 8, 1911, Kamerlingh Onnes watched the resistance of mercury vanish in liquid helium. How the experiment worked and why it surprised him.