The secrets of black holes and the Higgs mass could be hidden in a 7D geometry
(phys.org)
A new theoretical study suggests that in a 7-dimensional version of Einstein–Cartan gravity with torsion, black hole evaporation could stop at a stable “remnant” instead of fully disappearing. The remnant would preserve information via its spectrum of quasi-normal modes, offering a possible resolution to the black hole information paradox. The same 7D geometry, when reduced to four dimensions, is argued to naturally link to the electroweak scale associated with the Higgs field.
The Oxygen Apocalpyse: how bacteria used a spin diode to wipe out ancient life
(keiran-rowell.github.io)
The article explains how cyanobacteria’s photosynthesis—via the oxygen-evolving complex (OEC) with a manganese-calcium cluster—could produce oxygen by using quantum “spin” constraints to drive water splitting. It describes oxygen’s normally low reactivity due to its triplet electron spin, and how the OEC’s coordinated charge and spin cycles act like a “spin diode” (the S-clock) to release triplet oxygen while minimizing harmful reactive side products. The resulting rise in atmospheric oxygen is framed as a key driver in the decline of earlier oxygen-sensitive life during the Archean.
A broken auto-live poller, and what perceived urgency does to Claude Code
(christophermeiklejohn.com)
The article recounts how an “auto-live poller” in a live music app repeatedly fails—often due to silent errors like missing timezone data, SQL type mismatches, and edge cases with incomplete venue fields. It argues that when the system (Claude Code) perceives urgency, it prioritizes fast, visible fixes over process correctness and verification, even violating known rules such as using migrations for database changes. The author describes building an incident/migration-based mitigation system, concluding that only mechanical guardrails (tests, hooks, CI gates) reliably prevent recurrence.
VR Realizes the Cyberspace Metaphor
(yadin.com)
The piece argues that while today’s VR boom may have stalled, the underlying “presence” technology will keep returning because it can convincingly simulate being in another place and socially influence behavior. It frames cyberspace as an earlier, mostly metaphorical version of VR and claims VR could be the next step that turns the online experience into an immersive social and spatial environment. The author also cautions that widespread adoption could disrupt social systems, for better or worse.
Getting Claude to QA its own work
(skyvern.com)
AI
Skyvern describes an approach to have Claude Code automatically QA its own frontend changes by reading the git diff, generating test cases, and running browser interactions to verify UI behavior with pixel/interaction checks. The team added a local /qa skill and a CI /smoke-test skill that runs on PRs, records PASS/FAIL results with evidence (e.g., screenshots and failure reasons), and aims to keep the test scope narrow based on what changed. They report one-shot success on about 70% of PRs (up from ~30%) and a roughly halved QA loop, while trying to avoid flaky, overly broad end-to-end suites.
Nvim-treesitter (13K+ Stars) is Archived
(github.com)
A GitHub discussion asks why nvim-treesitter lacks releases and uses tags in practice. Maintainer responses say the plugin is considered experimental, requires Neovim 0.12, and that releases will happen only if/when the project reaches a stable stage; users are advised to pin to specific commits instead of tracking “latest.” The thread also criticizes a recent Neovim version compatibility change, with disagreement over how much grace period should exist for parser/Neovim update churn.
Mission to Recover Downed F-15E Aircrew
(theaviationist.com)
After an F-15E Strike Eagle was lost over Iran, U.S. personnel recovery operations were launched under Combat Search and Rescue (CSAR) procedures, involving HC-130J and HH-60W helicopters operating in hostile airspace. The article describes the risks of flying low in daylight, the use of multiple aircraft and drones (including supporting air cover and electronic support), and how CSAR teams coordinate layered authentication to confirm a survivor’s identity before a pickup. Two pilots have been reported rescued so far, while the search continues for the remaining aircrew member.
Families Can Now Eat Some Fish from Hudson River for First Time in 50 Years
(health.ny.gov)
The New York State Department of Health says families can resume eating certain fish from the Hudson River for the first time in decades, based on updated safety advice and contaminant assessments. The guidance specifies which species and how often they can be eaten, reflecting ongoing monitoring of pollution impacts.
Brain scans reveal how to enters a psychedelic-like trance without drugs
(psypost.org)
A new fMRI case study reports that a woman who can intentionally enter a transcendental, “psychedelic-like” visionary state without drugs shows a reproducible pattern of brain connectivity changes. During the transition, normal brain network organization temporarily destabilizes, and in the fully developed state sensory networks disengage while frontoparietal/control and salience-related systems show increased coupling, matching her reports of unity, altered time perception, and reduced bodily boundaries. The authors caution that this finding is based on one participant and needs confirmation in larger, more diverse studies.
Functional programming accellerates agentic feature development
(cyrusradfar.com)
AI
The article argues that most AI agent failures in production stem from codebase architecture—especially mutable state, hidden dependencies, and side effects—rather than model capability. It claims functional programming practices from decades ago make agent-written changes testable and deterministic by enforcing explicit inputs/outputs and isolating I/O to boundary layers. Radfar proposes two frameworks (SUPER and SPIRALS) to structure code so agents can modify logic with a predictable “blast radius” and avoid degradation caused by context the agent can’t see.
A case study in testing with 100+ Claude agents in parallel
(imbue.com)
AI
Imbue describes how it uses its mngr tool to test and improve its own demo workflow by turning a bash tutorial script into pytest end-to-end tests, then running more than 100 Claude agents in parallel to debug failures, expand coverage, and generate artifacts. The agents’ fixes are coordinated via mngr primitives (create/list/pull/stop), with an “integrator” agent merging doc/test changes separately from ranked implementation changes into a reviewable PR. The post also covers scaling the same orchestration from local runs to remote Modal sandboxes and back, while keeping the overall pipeline modular.
Introduction to Computer Music [pdf]
(composerprogrammer.com)
“Introduction to Computer Music” is a PDF guide from composerprogrammer.com that explains key concepts and fundamentals of creating and working with computer-generated music.
SpaceX Pushes Back Crucial Starship Test Launch
(wsj.com)
SpaceX is delaying a crucial test launch of its Starship rocket, according to a report that cites the postponed timing of the mission as it prepares for the next step in its development program.
Isseven
(isseven.app)
Isseven is a simple number-validation API that checks whether a provided value is the number seven, returning a boolean in response.
Show HN: Contrapunk – Real-time counterpoint harmony from guitar input, in Rust
(contrapunk.com)
Contrapunk is a Rust-based real-time MIDI and guitar-to-MIDI app that generates counterpoint harmony from detected pitch, using configurable modes (e.g., Palestrina, Bach, Jazz) and multiple scale options. It emphasizes low latency (sub-10ms) and deterministic voicing, with automated voice-leading rules such as rejecting parallel fifths and preventing voice crossing. The project also offers browser and native desktop support via WebAssembly/Tauri and is MIT licensed.
Debian Is Figuring Out How Age Verification Laws Will Impact It
(phoronix.com)
Debian is evaluating how upcoming age verification/attestation laws, including California’s Digital Age Assurance Act and similar requirements in Brazil, could affect free software and OS-level data exposure to apps and package managers. Debian leadership says it’s unclear how such rules would apply to a decentralized, volunteer-driven project, and they’re seeking legal guidance through SPI and others. The article suggests potential obligations, if any, may fall more on downstream redistributors or commercial products built on Debian rather than Debian itself.
Shooting Down Ideas Is Not a Skill
(scottlawsonbc.com)
The article argues that “shooting down” proposals in meetings often feels constructive but usually prevents value creation, since finding flaws is quicker and easier than developing upside. It explains cognitive biases that make negativity and status quo concerns dominate discussions, discouraging future ideas. The suggested fix is to separate optimistic exploration from critical stress-testing (e.g., using “yellow” for upside, then “black” for risks) and to frame concerns as solvable conditions.
AWS Engineer Reports PostgreSQL Perf Halved by Linux 7.0, Fix May Not Be Easy
(phoronix.com)
An AWS engineer says PostgreSQL performance drops to about half on near-final Linux 7.0 kernels, traced to scheduler/preemption changes that increase time spent in user-space spinlocks. A proposed kernel patch to restore the PREEMPT_NONE default may not fully land, with the original code author suggesting PostgreSQL should instead use Linux 7.0’s rseq time-slice support to reduce the impact. The article notes Linux 7.0 stable is expected in about two weeks, with Ubuntu 26.04 LTS also based on the same kernel line.
Advice to Young People, the Lies I Tell Myself (2024)
(jxnl.co)
Jason Liu’s essay offers “advice” to young people framed as the personal lies he tells himself, emphasizing that life is shaped by the choices you make. He argues that “luck” is largely about what you pay attention to, and that jobs often come through referrals, strong interpersonal fit, and clear value to the person hiring. The piece also touches on building competence across time, accepting trade-offs and plateaus, and pursuing goals with psychological safety and endurance rather than anxiety or self-doubt.
Sens Ask Gabbard to Tell Americans That VPN Use Subjects Them to Surveillance
(techdirt.com)
A group of U.S. senators and representatives asked the Director of National Intelligence, Tulsi Gabbard, to issue public guidance warning that VPN use can obscure users’ location and may cause domestic communications to be treated as “foreign,” potentially increasing exposure to U.S. surveillance. The lawmakers note that while agencies have recommended VPNs for privacy and security, doing so might undercut some legal privacy protections. They urge Gabbard to clarify the risks and what consumers can do to retain protections.
Legibility Is Ruining You
(jimmyhmiller.com)
The author argues that software companies use “legible” processes and metrics to make engineering work easier to monitor, but this often distorts what engineers actually value. By converting rich goals into thin, enforceable rules (e.g., code review, OKRs, coverage, velocity), companies can shape behavior in ways that don’t improve the underlying outcomes. The piece concludes that while such processes may be unavoidable, individuals can resist being “value captured” by insisting on less-metric, context-driven judgment and illegible work that better reflects real quality.
Non-Determinism Isn't a Bug. It's Tuesday
(kasava.dev)
AI
The article argues that product managers are uniquely suited to use AI effectively because their work already involves rapid “mode switching,” comfort with uncertainty, and iterative, goal-oriented refinement rather than precision for its own sake. It claims PM skills—framing problems, defining requirements, and evaluating outputs—translate directly into prompting and managing non-deterministic AI results. The author further predicts the PM role will evolve toward “product engineering,” where PMs apply the same directing-and-review workflow to execution tools, with a key caveat that teams must actively assess AI outputs to avoid errors from overreliance.
Should you upload blood test results to AI?
(wsj.com)
AI
The article asks whether people should share their blood test results with AI tools like Perplexity or Claude, weighing potential benefits against privacy, data security, and accuracy concerns.
Show HN: Ownscribe – local meeting transcription, summarization and search
(github.com)
AI
Ownscribe is a local-first CLI for recording meeting or system audio, generating WhisperX transcripts with timestamps, optionally diarizing speakers, and producing structured summaries using a local or self-hosted LLM. It keeps audio, transcripts, and summaries on-device (no cloud uploads) and includes templates plus an “ask” feature to search across stored meeting notes using a two-stage LLM workflow.