Science

Archived Code Decides Which Supernova Results Hold Up

By Jonas Eriksen / Sep 18, 2026

Archived code, not just data, decides whether supernova classifications survive scrutiny. A closer look at how smoothing choices and missing error bars reshape published results.

Science

Rubin Observatory's LSST Camera Turns Every Night Into a Repeated Survey

By Alice Chen / Sep 18, 2026

Rubin Observatory's 3.2-gigapixel LSST Camera turns the southern sky into a ten-year time-lapse, letting astronomers measure motion, not just position.

Science

Reproducibility Audits and Journal Sanctions Diverge on Retraction Reach

By Karim Osman / Sep 18, 2026

Reproducibility audits test claims; journal sanctions test conduct. The two tracks run on different clocks, different evidence, and produce different retraction reach.

Tech

Mutual TLS Deployment Stalls Where Certificate Rotation Lacks Automation

By Yusuke Tanaka / Oct 20, 2026

Mutual TLS demos cleanly but stalls at renewal. A practical look at what mTLS changes on the wire, where manual rotation breaks, and automation patterns that survive.

Tech

Developer Tooling Careers Trade Depth for Breadth in Hiring

By Lucas Mendes / Oct 20, 2026

Developer tooling careers reward breadth in hiring loops, but depth in package managers, build systems, and CI pays off long-term. How specialists can bridge the gap.

Tech

Why Token Rotation Logs Miss the Sessions Attackers Keep

By Lucas Mendes / Oct 20, 2026

Token rotation logs report success while stolen sessions stay live. A payments team post-mortem shows why session state escapes audit and how to close the gap.

Science

Priming Studies Survive Contact Only Where Labs Logged Their Negatives

By Renu Shah / Sep 18, 2026

Priming effects diverge sharply under replication. Semantic priming holds up; social priming often does not. The deciding factor is whether null results got logged.

Tech

Maintainer Contracts Rarely Cover the On-Call Hours Behind a Security Fix

By Lucas Mendes / Oct 20, 2026

Security fixes ship on time, but the unpaid on-call hours behind them rarely appear in maintainer contracts. A breakdown of who bears the cost and how to fix it.

Science

Liquid Electrolytes Outrun Solid Polymer Cells Above Room Temperature

By Karim Osman / Sep 18, 2026

A focused explainer on why liquid electrolytes outperform solid polymer cells above room temperature, what the conductivity data show, and what the comparison leaves out.

Science

Kahneman and Tversky Asked 100 Subjects About a Disease

By Jonas Eriksen / Sep 18, 2026

How Kahneman and Tversky's 1981 Asian disease problem revealed framing effects, why identical outcomes produced opposite choices, and what replications show.

Tech

Editors and Package Managers Split Over Who Pays for CI Minutes

By Lucas Mendes / Oct 20, 2026

CI minutes have become a real budget line. Editors, package managers and platforms disagree about who absorbs the cost, and the contracts decide the winner.

Tech

Nvidia's Driver Stack Carries the Correctness Burden Its Hardware Cannot

By Deepa Iyer / Oct 20, 2026

Nvidia's GPUs do not guarantee correct math on their own. The driver and CUDA runtime carry that promise, and the burden is shifting toward users.

Tech

Postgres Wire Protocol Changes and the Client Libraries That Must Track Them

By Lucas Mendes / Oct 20, 2026

How Postgres client libraries track wire protocol changes, where message parsing breaks, and the testing practices that keep drivers compatible across server versions.

Science

Fluid Dynamicists Carry a Turbulence Model Into Galaxy Formation

By Alice Chen / Sep 18, 2026

How fluid dynamicists' k-epsilon turbulence closures migrated into galaxy formation simulations, what procedural choices make them work, and where the analogy breaks down.

Science

Grant Review Panels Score Methods Sections Before Results Sections

By Karim Osman / Sep 18, 2026

Grant panels often score methods sections before reading results. This methodology breakdown explains how review order shapes funding decisions and how to audit it.

Tech

Postgres Replication Slots Hold WAL Until Disk Fills

By Yusuke Tanaka / Oct 20, 2026

Replication slots pin WAL segments indefinitely, filling disk and halting writes. Learn how slots retain WAL, the signals to watch, and the settings that cap the damage.

Tech

Chrome and Firefox Repaint Differently When CSS Grid Changes

By Yusuke Tanaka / Oct 20, 2026

Chrome and Firefox invalidate and repaint CSS Grid changes differently. A look at Blink's invalidation sets, Gecko's display lists, and how to profile both.

Science

Lab Priming Talks Stall When Raters Guess the Hypothesis

By Renu Shah / Sep 18, 2026

Priming talks often rely on human raters who infer the hypothesis from video clips. That inference inflates effects and stalls replication. Here is the procedural fix.

Tech

Machine-Learning Teams Rewrite Their Deployment Playbooks as Inference Standards Shift

By Yusuke Tanaka / Oct 20, 2026

Serving engines like vLLM and TensorRT-LLM now define wire behavior. What changes for ML teams at the API level, in on-call rotations, and in hiring loops.

Science

Replication Studies Stall Where Grant Cycles End

By Karim Osman / Sep 18, 2026

Why replication attempts in psychology and behavioural economics stall at the grant cycle, and what shared pools, standing funds and null-result journals could change.

Science

Counting What a Catalyst Costs Before Counting What It Makes

By Jonas Eriksen / Sep 18, 2026

Catalysis papers report activity, not the infrastructure and synthesis costs behind it. A research-economics look at what the literature leaves out and how to read it.

Science

Spectrograph Throughput Sets the Cost of a Survey Night

By Renu Shah / Sep 18, 2026

How spectrograph throughput drives the real cost of a survey night, why funding favors new instruments over operations, and what survey teams can audit before proposing.

Tech

Open Source Foundations Fund Compliance Work From Trademark Royalties

By Sara Park / Oct 20, 2026

Open source foundations increasingly fund license compliance work through trademark royalties from certification marks, creating a business model with real governance tensions.

Science

Cold Atoms Settle a Quasiparticle Mass Debate in Two Labs

By Jonas Eriksen / Sep 18, 2026

Two cold-atom experiments measured quasiparticle effective mass and disagreed by roughly 30 percent. Here is what shaped the verdict, and how to read such claims.

Tech

Android Fragmentation Forces Cross-Platform Stacks Into Per-Device Testing

By Sara Park / Oct 20, 2026

Android fragmentation turns one cross-platform codebase into hundreds of test targets. A practical breakdown of device tiering, snapshot matrices, and hardening the mobile pipeline.

Science

Python Scripts Move Into Economics Departments and Rewrite What Counts as a Result

By Karim Osman / Sep 18, 2026

Python has become standard in applied economics, turning code into evidence and forcing departments to budget for reproducibility they never planned to fund.

Science

The Redshift Drift Test Measures Expansion Without Assuming a Model

By Renu Shah / Sep 18, 2026

Redshift drift measures cosmic expansion directly, without assuming a dark energy model. Here is how the test works, what today's instruments can do, and what it cannot yet settle.

Tech

Check Which Maintainers Hold Signing Keys Before Your Next Update

By Yusuke Tanaka / Oct 20, 2026

Code signing only proves what a key says. Here is how to check which maintainers actually hold signing keys, where those keys live, and what to fix this quarter.

Science

Ocean Cores Rewrite the Timing of the Last Ice Retreat

By Alice Chen / Sep 18, 2026

Deep-sea sediment cores and oxygen isotopes in foraminifera shells have reshaped the chronology of the last ice retreat, and with it the craft of dating ancient climate.

Science

Two Pipette Calibrations Decide Whether a Grid Cell Finding Replicates

By Renu Shah / Sep 18, 2026

Grid cell findings often fail to replicate. The culprit may be pipette calibration drift, a mundane variable that shifts spike timing and goes unlogged in methods sections.

Tech

Certificate Pinning Breaks Silently When Rotation Runs Behind Revocation

By Deepa Iyer / Oct 20, 2026

Certificate pinning fails quietly when revocation outruns rotation. A practical breakdown of pin sets, overlap windows, revocation drills and handshake telemetry.