SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

Researchers Allege Data Fraud in Major Procrastination Study

0
·1 views

Investigators at Data Colada have published evidence suggesting fraud in a widely cited study on procrastination. The findings raise serious concerns about the integrity of the research, which has influenced academic understanding of the topic. Data Colada has previously exposed manipulation in other high-profile psychology studies. The allegations, if confirmed, could prompt retractions and a reassessment of conclusions drawn from the affected research.

Read the full story at Hacker News

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Analysis of 559 AI coding rule files finds 63% of content is not instructions

A developer building a tool to verify AI coding agent compliance analyzed 559 public configuration files — including CLAUDE.md, AGENTS.md, and Copilot instructions — from projects like PyTorch, Kubernetes, and Elasticsearch. Parsing over 23,700 items revealed that 62.9% of content in these files is documentation such as directory listings, architecture notes, and reference tables rather than actual directives. Of the genuine rules identified, only 43.5% are mechanically verifiable, such as 'never commit to main,' while the remaining 56.5% involve subjective behaviors like 'surface bad news first' that require human judgment to assess. The 63/37 split between non-instructions and rules remained largely stable as the corpus grew from 40 to 559 files across different formats, suggesting the distinction is a property of language rather than file structure. The developer has released a free, locally-run open-source tool called RuleReceipt to help users audit their AI agent rule files.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AI-Powered Spatial Intelligence Set to Reshape Autonomous Systems by 2026

The global geospatial intelligence market is projected to surpass $18 billion by 2026, fueling rapid advances in autonomous vehicles, robotics, and urban planning. AI-driven world models and multimodal systems are replacing static 2D maps with real-time, three-dimensional environmental data, enabling tools like Atlas Navi's engine to predict traffic conditions hours in advance. Digital twins — virtual replicas of real-world infrastructure — are being adopted by cities like Dubai and Singapore to manage climate risks and improve urban maintenance. The U.S., China, and Europe are each racing to lead the sector, with companies such as Waymo, Baidu, and Mistral developing competing frameworks for autonomous and logistics applications. Multimodal AI, combining geospatial imagery, sensor feeds, and human instructions, is also improving safety in high-risk industries like construction and enabling driverless vehicles to respond to pedestrian gestures and verbal commands.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Security Report Flags Oracle Manipulation as Top Risk for ether.fi's $4.26B Platform

A DeFi security research team has published a risk assessment of ether.fi Stake, a liquid-staking protocol holding approximately $4.26 billion in total value locked across Ethereum and multiple Layer 2 networks. The report identifies oracle manipulation as the most critical systemic threat, given that the protocol's reward distribution, slashing safeguards, and cross-chain bridge operations all depend on on-chain price feeds. Researchers outlined six key attack vectors, including flash-loan-driven price skews, single-source feed compromise, and governance-controlled oracle parameter changes, with flash-loan attacks rated highest in overall risk. The report warns that even a modest 5% price distortion could result in over $200 million in misallocated value through inflated receipt minting, erroneous reward calculations, or forced liquidations. Researchers concluded with a prioritized remediation roadmap to address the identified vulnerabilities.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Apify MCP Connectors Turn a G2 Review Scraper Into an Automated Dedup Pipeline

A developer maintaining a G2 Software Reviews Scraper on Apify—used by 124 product and sales teams—identified a recurring manual workflow: users were downloading CSVs weekly and hand-comparing them to spot new competitor reviews. When Apify launched MCP connectors in July, the developer used them to automate this process by building a companion Actor that calls the existing scraper, deduplicates results against the user's own Supabase database, and writes a digest to Notion. The companion was kept separate from the live scraper to avoid disrupting existing users during experimentation. A key design goal was security: the system accesses users' databases and Notion workspaces without the developer ever handling their credentials directly. The build surfaced several unexpected issues, including a database entering sleep mode mid-run and review data arriving wrapped in a prompt-injection guard.

0 comments Read more at DEV Community
Researchers Allege Data Fraud in Major Procrastination Study · ShortSingh