New science, in plain language. Checked before you read a word.

Every two days we take 10 brand-new scholarly papers, screen each one for retractions and a live DOI, and rewrite them so a curious person can actually read them.

This screen checks process signals. It does not tell you the findings are true.

Get it free, every two days

Three papers free every issue. No spam. Leave in one click.

In this issue 10 papers · 3 free

01 Science AI in scientific publishing: Slower, worse, and more expensive Free
02 Science Cross-cohort analysis of expression and splicing quantitative trait loci in TOPMed Free
03 Nature Regression to the mean can explain saturation of geomagnetic storms Free
04 Nature People use fast and flat simulation to reason about new games Subscribers
05 Proceedings of the National Academy of Sciences Response duration tracks confidence Subscribers
06 Proceedings of the National Academy of Sciences Polarons are probes of the dynamic nanoscale environments found in electrochemically doped π-conjugated polymers Subscribers
07 Nature Climate Change Intercity connectivity enhances urban mobility resilience to extreme rainfall Subscribers
08 Nature Communications Quasi-solitons in Rydberg atom chains Subscribers
09 Nature Communications β-arrestin recruitment facilitates a direct association with G proteins Subscribers
10 Nature Medicine An anti-PMEL antibody-drug conjugate with a Gq/11 inhibitor payload in GNAQ/GNA11-mutant melanomas: a phase 1 trial Subscribers

The papers Read free

Science·16 July 2026

H. Holden Thorp

AI in scientific publishing: Slower, worse, and more expensive

This essay argues that AI is not making science publishing faster or cheaper. AI tools can write research papers very quickly. But people must still check those papers by hand for mistakes and fake results. That checking takes extra time and money. The author compares this to old factory history. New machines and tech often promised to save workers effort. Instead they often made bosses demand more work, not less, while profits went mostly to people at the top. The author says something similar is now happening with AI in science publishing.

AI speeds up writing papers, but checking them by hand makes publishing slower and pricier.

The number: No key statistic identified
The caveat: This is an opinion piece, not a data study. It does not test its claims with numbers or experiments.

For your life: today, nothing. This is basic research.

Integrity screen: passed (3 checks) Checked 17 July 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Science). Read the source

Science·16 July 2026

Peter Orchard, Thomas W. Blackwell et al.

Cross-cohort analysis of expression and splicing quantitative trait loci in TOPMed

Genes are turned on and off by small differences in our DNA. Scientists studied over 14,000 samples of gene activity from blood, lung, and other body tissues. They found tens of thousands of extra, less obvious DNA links that control how genes turn on and off. When they matched these links to a huge health study covering 164 different traits, they found 10,611 matches. Most of those matches involved these newer, less obvious genetic links, not the well known ones.

Many hidden genetic controls of gene activity had not been found before.

The number: 10,611 genetic signals linked to diseases or traits were matched to gene activity controls, including 7,096 tied to the newly found, less obvious ones.
The caveat: This study only shows patterns in gene activity data. It does not prove which genes actually cause which diseases. Larger studies in the future may find even more of these hidden genetic links.

For your life: today, nothing. This is basic research.

Integrity screen: passed (3 checks) Checked 17 July 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Science). Read the source

Nature·15 July 2026

Nithin Sivadas, David Sibeck et al.

Regression to the mean can explain saturation of geomagnetic storms

Scientists once thought Earth's magnetic response to strong solar wind hits a ceiling during extreme space weather. This new study shows that "ceiling" was actually a math illusion. It comes from a common statistical trap called regression to the mean, caused by uncertain measurements of the solar wind's timing and strength. Once the errors are fixed, Earth's response keeps growing steadily with solar wind strength. It never levels off. This means the biggest geomagnetic storms could hit Earth twice as hard as scientists had believed.

Earth's magnetic storms may be twice as strong as once believed, with no true limit.

The number: The true impact of extreme geomagnetic storms can be twice as large as previously thought.
The caveat: The paper shows the old "leveling off" idea came from biased math, not real physics. But it does not rule out every other possible explanation for the original pattern.

For your life: nothing to act on yet. Early research, worth watching.

Integrity screen: passed (3 checks) Checked 17 July 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Nature). Read the source

7 more papers are in the full issue.

The same screening, the full reading, every two days. 2 euros a month, or 20 a year, and you can leave in one click.

Nature·15 July 2026

Katherine M. Collins, Cedegao E. Zhang et al.