Science

Current Picks

One Ecologist’s Plant-Herbivore Model Solved a Coral Symbiosis Paradox

By Jonas Eriksen / Jun 11, 2026

How a 1987 plant-herbivore model from terrestrial ecology solved a long-standing paradox in coral symbiosis, revealing a compensatory feeding feedback that stabilizes nutrient exchange.

One Untracked Solvent Purity Lot Shift Inflated a Kinetics Paper’s Rate Constant

By Renu Shah / Jun 11, 2026

A 23% jump in a reported rate constant was traced to a 0.03% water difference between solvent lots. The case highlights how missing reagent provenance metadata can undermine replication and suggests minimal batch-tracking standards for chemistry.

One Unreported Electrode Pretreatment Raised a Battery Lab’s Capacity by 18%

By Alice Chen / Jun 11, 2026

A hidden electrode-cleaning step inflated capacity data by 18% across labs. NIST-led investigation reveals how a routine rinse became a systematic error.

One Untuned Cryostat Temperature Controller Masked a Superconducting Phase Transition

By Jonas Eriksen / Jun 11, 2026

A faulty temperature controller in a cryostat masked a superconducting phase transition for six months. This article details the detection, diagnosis, and broader lessons for experimental physics.

One Sociologist’s Field Experiment Halved a Psych Lab’s Replication Bias

By Alice Chen / Jun 11, 2026

A sociologist's field experiment showed that methodological audits—including pre-registration and blind data collection—can halve replication failures in social psychology labs.

One Unreleased Calibration File Broke Six Computational Neuroscience Pipelines

By Karim Osman / Jun 11, 2026

A single unreleased calibration file for MRI gradient nonlinearities caused six major preprocessing pipelines to produce contradictory results. The error, hidden for years, eroded effect sizes and inflated false positives.

One Funder’s Single-Subject Cost Cap Shrank Rodent Neuroimaging Cohorts by a Quarter

By Renu Shah / Jun 11, 2026

A major charity's US$1,500-per-animal cap on rodent imaging costs reduced cohort sizes by roughly 25% across labs, undermining statistical power for small-effect studies.

One Untracked Detector Bias Voltage Shift Compromised a Dark Matter Search

By Jonas Eriksen / Jun 11, 2026

A 0.3% drift in photomultiplier bias voltage at the LUX-ZEPLIN detector mimicked a dark matter signal, hiding a true WIMP signal for years. A graduate student's forensic analysis of telemetry logs revealed the flaw.

One 0.003 Arcsecond Star Tracker Error Mapped a Planet to the Wrong Star

By Karim Osman / Jun 11, 2026

A tiny star tracker glitch in Gaia led astronomers to misattribute an exoplanet to the wrong star. The error, 0.003 arcseconds, wasted years of follow-up and reshaped how the field vets astrometric data.

One Unreported Precatalyst Activation Step Doubled a Cross-Coupling Yield

By Renu Shah / Jun 11, 2026

A trace ammonium chloride contaminant stabilizes a Ni(I) dimer intermediate, doubling the yield of a nickel-catalyzed C–N coupling reaction. The finding explains why many published yields may be underestimates.

One Uncalibrated Two-Photon Microscope Laser Priced a Lab Out of Longitudinal Imaging

By Alice Chen / Jun 11, 2026

A single uncalibrated laser can halt longitudinal imaging for months, revealing how equipment costs distort neuroscience research and funding.

One Grant Agency’s Per-Cage Fee Rule Halved Primate Social Behavior Studies

By Renu Shah / Jun 11, 2026

A per-cage fee hike by the US National Institutes of Health inadvertently halved primate social behavior research, shifting incentives toward single housing and altering the course of behavioral neuroscience.

One Grant Agency’s No-Ship-Core Rule Forced a Pacific Sediment Transect Rethink

By Karim Osman / Jun 11, 2026

A grant agency's ban on ship-based coring mid-campaign forced a Pacific sediment transect to rely on autonomous gliders. An independent audit later revealed major gaps in the data, leading to a hybrid approach that improved quality and cut costs.

One Untracked Anode Porosity Parameter Biased Three Battery Capacity Studies

By Karim Osman / Jun 11, 2026

A single unmeasured porosity parameter inflated capacity gains in three battery studies from 2022–2024, exposing a reproducibility gap in materials science.

One Unanalyzable Python Script Blocked a Computational Epidemiology Paper for Two Years

By Jonas Eriksen / Jun 11, 2026

A single Python script with no docstrings and hardcoded paths held a computational epidemiology paper in peer review for two years. The story reveals how funding incentives, infrastructure costs, and journal practices discourage code hygiene.

One Untuned Interferometer Port Fixed a Dark Matter Search Null Result

By Renu Shah / Jun 11, 2026

A null result in a dark matter search was traced to a mis-set optical interferometer port. A cross-disciplinary fix from quantum optics and LIGO's port-tuning methods resolved the issue, turning a null into candidate events.

One Unpublished Polymerization Catalyst Recipe Doubled a Battery Lab’s Anode Capacity

By Renu Shah / Jun 11, 2026

A single unpublished catalyst recipe doubled a battery lab's anode capacity from ~360 to ~720 mAh/g. This feature explains the chemistry, evidence, and limitations of the method.

One Unarchived Monte Carlo Seed Code Collapsed a Galaxy Formation Simulation

By Alice Chen / Jun 11, 2026

A missing Monte Carlo seed code made a galaxy formation simulation irreproducible, costing millions of CPU-hours and spurring new archiving standards across computational science.

One Grant Agency’s Per-Animal Cost Limit Cut Rodent Neuroimaging Cohorts by a Third

By Renu Shah / Jun 11, 2026

A single agency's per-animal cost cap forced rodent neuroimaging labs to shrink cohorts by a third, eroding statistical power and shifting research toward cheaper but narrower methods.

One Unversioned Climate Model Parameter Produced 3 °C Spread in 2100 Projections

By Alice Chen / Jun 11, 2026

A single unversioned parameter controlling ice nucleation in cloud models generated a 3°C spread in 2100 temperature projections, revealing deep reproducibility challenges in computational climate science.