securitylab_nJuly 12, 2026🇷🇺Translated from Russian

Universe Proves Ordinary After All: High-Profile DESI Study Claiming Giant Aligned Cosmic Structures Collapses Following Data Correction

A high-profile paper published in the journal Nature that claimed to detect giant, directionally aligned filaments in the cosmic web has been effectively invalidated after independent researchers identified fundamental errors in the data analysis.

The original study examined observations from the Dark Energy Spectroscopic Instrument (DESI), which has produced one of the largest three-dimensional maps of the universe to date. The dataset included 47 million galaxies and quasars covering more than 11 billion years of cosmic history within the 13.8-billion-year lifespan of the universe.

According to the authors, the filaments of the cosmic web appeared significantly longer than predicted by existing models and, more strikingly, showed a preferred orientation rather than random alignment. Such a finding would have challenged the cosmological principle, the foundational assumption that the universe is statistically homogeneous and isotropic on sufficiently large scales, with no preferred directions.

An independent cosmologist who re-examined the work discovered that the team had used luminosity distance instead of the conventional comoving distance when measuring large-scale structure. These two distance measures serve different purposes and are not interchangeable for studies of cosmic-web geometry. In addition, the analysis did not incorporate the fact that distances between distant objects change over time due to the ongoing expansion of the universe.

After substituting the correct distance metric and applying the necessary expansion correction, the apparent giant aligned structures disappeared. The revised results showed a cosmic web that remains inhomogeneous on moderate scales but exhibits no mysterious preferred direction, fully consistent with the cosmological principle.

One of the original authors has disputed the critique, arguing that the re-analysis focused on the distribution of structures rather than their orientation. The independent reviewer maintains that the identified errors affect both types of measurements and are independent of the chosen analytical approach.

The controversy has raised broader questions about the peer-review process at leading journals. Although Nature typically requires papers to present results capable of significantly advancing a field, reviewers cannot realistically verify every line of code or calculation. The authors of the disputed study also bypassed the common practice of posting a preprint on the arXiv server, depriving the community of an opportunity for early scrutiny. Scientific embargoes imposed by journals further limited pre-publication discussion, highlighting ongoing tensions between the desire for high-impact announcements and the need for thorough, open verification.

Related articles

HabrOther

MEPhI Opens 2026 Admissions for Online Cybersecurity Master's Program with Yandex Practicum

The National Research Nuclear University MEPhI, in partnership with Yandex Practicum, is accepting applications for its online master's program in Cybersecurity for the 2026 intake. The two-year program leads to a state diploma in Information Security under code 10.04.01 and a professional retraining certificate from Yandex Practicum. Students can choose from four specialized tracks covering AppSec, DevSecOps, network security, and AI security. Admission is fully online and includes document submission via Gosuslugi, an entrance exam, and a motivation letter requiring at least 80 points. The program runs entirely remotely with evening and weekend classes, allowing students to combine studies with work while accessing student benefits and an educational loan at a subsidized 3% rate.

安全客Other

360 Group Launches NanoWork Enterprise AI Platform with Built-in Security and Opens Nationwide Channel Partner Recruitment

On July 28 at the Beijing National Convention Center, 360 Group founder Zhou Hongyi officially unveiled NanoWork, a next-generation enterprise intelligent agent work platform. The platform is designed to bridge the gap between powerful AI models and real-world business tasks by enabling multi-agent collaboration, on-demand model scheduling, and 24/7 cloud operation across diverse scenarios. NanoWork was developed through extensive real-world testing involving 100,000 intelligent agents, coverage of 630 positions over 150 days, consumption of 350 trillion tokens, and collection of 56,000 feedback items. A core emphasis is placed on native security features drawn from 360 Group's two decades of cybersecurity experience to prevent errors that could lead to actual data loss or permission breaches. The company is now actively recruiting city-level channel partners across China to help deploy the solution in local industries and activate existing customer bases with AI capabilities.

AntiMalwareOther

Google Enables Document Backup to Drive in Stable Play Services 26.26 Release

Google has rolled out automatic document backup from Android devices to Google Drive in the stable version of Google Play Services 26.26. The feature, which the company prepared for nearly a year, adds a new Documents option in Settings on Pixel phones under Accounts and backup. It remains disabled by default to avoid uploading the Downloads folder without user consent. Once enabled, supported files including PDF, DOC, PPT, XLS, ZIP and even APK files are copied to a new Android backups folder on Drive, with separate subfolders created for each device. The backup consumes storage quota and offers no automatic two-way sync, requiring manual cleanup when disabled. Traces of the capability first appeared in August 2025, followed by an official mention in February 2026 and beta testing before the current stable deployment.

HabrOther

Bridging Manual and Automated Testing: InfoWatch Engineer Outlines Unified Quality Process

InfoWatch senior test engineer Mikhail Shalepo has published a detailed article describing how his team built a reproducible process that links manual and automated testing into a single quality system. The approach addresses the growing complexity of server-side products that undergo quarterly releases and require extensive matrix testing across multiple operating systems and installation scenarios. Shalepo explains that the regulation focuses on decision points rather than prescribing exact test volumes, ensuring that choices about automation candidates, smoke-test gates, failure ownership, and manual compensation are documented and visible to the entire team. The framework divides work into two main blocks—feature development and pre-release regression—each containing preparation, execution, and completion stages. Key outputs include linked artifacts in TMS Scale, explicit Definition of Done criteria, and traceable coverage data that prevents reliance on individual knowledge. The article also shares practical lessons, including why separate mind maps and tables failed to stay current and how the team now integrates automation status directly with test cases.