Hacker News Top 10
English Edition

1. Big Tech ruined the cloud, so we're renaming ours (119 points by 2sf5)

   Nabu Casa is renaming Home Assistant Cloud to Home Assistant Link. The post argues that Big Tech has ruined the word “cloud” through rising subscriptions, outages, data harvesting, and lock-in. The new name is meant to reflect values of privacy, optionality, and freedom from walled gardens. Nabu Casa, a commercial partner of the Open Home Foundation, says the service helps fund Home Assistant’s open-source development.

2. The Legend of von Neumann [pdf] (103 points by suopspaces)

   This is a PDF hosted by Gwern containing Paul Halmos’s 1973 piece “The Legend of von Neumann.” It likely examines John von Neumann’s legendary status, mathematical contributions, and broader influence on computing, physics, and game theory. The preview is mostly encoded PDF data, so specific arguments are not visible. It serves as a historical and biographical reflection on one of the 20th century’s most important mathematicians.

3. Supabase Is Acquiring Turso (11 points by cvburgess)

   Supabase announces that it is acquiring Turso. The post frames the move as a response to AI agents creating huge numbers of databases for prototypes, dashboards, and apps. Supabase says it already launches over one million databases per week and expects demand to outpace current infrastructure capacity. Supabase will continue building around Postgres, Turso will continue with SQLite, and together they aim to make database creation as easy as creating a file, with cheap on-demand provisioning and a path to production.

4. Mozilla shutting down Solo AI website creator (13 points by nedrylandJP)

   Mozilla is shutting down Solo, an AI website creator. The support FAQ says all Solo websites and data will be deleted on November 30, 2026, with no recovery after that date. Users can download an HTML export of their site and a CSV of image asset links, but original image source files are not included. The FAQ is designed to help users migrate their sites to another hosting provider.

5. Pi 1.0 (1576 points by sergiotapia)

   Earendil has released Pi 1.0, a hardened, minimal, extensible agent harness. Hundreds of thousands of people use Pi weekly, and feedback from issues and pull requests shaped the stable release. Pi runs the latest models and serves as a daily coding agent and a malleable substrate for agentic applications. Version 1.0 adds Codemode with native MCP and non-LLM models, virtual model extensions, deferred tool loading, Anthropic cache warming, mid-conversation system messages, and a new TUI.

6. AI Makes Me Sad (87 points by mondobe)

   “AI Makes Me Sad” is a personal essay about the author’s emotional response to AI progress. The author acknowledges they could benefit career-wise, including from high-paying AI jobs, but does not want to spend their life prompting Codex or Claude. They describe AI as threatening craft, control, and creativity, comparing it to Chaplin’s Modern Times and people volunteering to become obsolete. The essay captures a sense of hopelessness among technically skilled people even as AI opportunities expand.

7. Giving friends custom text buzzes based on Morse code (19 points by evakhoury)

   This blog post describes a project to give friends custom text-message vibration patterns based on Morse code. The author often keeps their phone in a pocket and wants to know who texted without pulling it out. Because texts have high signal-to-noise but are not always relevant, distinct buzzes help prioritize responses. The post explains how iPhone custom vibrations can encode different people or messages.

8. Shimano Bicycle Museum Review (240 points by pietroppeter)

   This review covers the Shimano Bicycle Museum in Japan’s Kansai region. The museum is a modern glass-and-steel building focused on bicycles rather than Shimano corporate history. Its bilingual exhibits trace bicycle evolution from hobby horses to pedals, chains, and tires, with many machines displayed under spotlights. The reviewer praises the collection, design, and message that there are as many ways to enjoy cycling as there are riders.

9. Several vulnerabilities have been discovered in the Linux kernel (470 points by luispa)

   This LWN article reproduces Debian Security Advisory DSA-6528-1 for the Linux kernel. It lists numerous CVEs from 2024 through 2026 affecting the kernel package. The advisory is dated September 29, 2026, and instructs Debian users to apply the security update. It is a routine but important kernel security notice for sysadmins.

10. Show HN: Audionaut – an open-source cross-platform multitrack audio editor (93 points by vltmrkls)

    Audionaut is an open-source cross-platform multitrack audio editor. It is a free desktop app for music, podcasts, and multitrack recording, with precise cutting, per-track playlists, multi-channel support, and clean exports. The project emphasizes professional audio editing without the heaviness of a full DAW. It can be driven by AI agents over MCP, reflecting a new pattern of agent-operable creative software.

AI/ML Insights & Trends

1. Agent harnesses are maturing into stable 1.0 infrastructure.
2. Trend/insight: Pi 1.0 shows that agent harnesses are moving from experimental demos toward hardened, minimal, extensible runtimes used by hundreds of thousands of people.
3. Why it matters for AI/ML development: Production users need reliability, maintainability, and long-term interfaces more than weekly novelty. Agent tooling that churns constantly is difficult to depend on.

4. Implications/actionable takeaways: Teams should standardize on stable agent runtimes, prioritize MCP compatibility, deferred tool loading, cache warming, and transcript-aware prompts, and avoid rebuilding around every new framework.

5. MCP is becoming an integration bus for agentic software.

6. Trend/insight: Pi 1.0 adds native MCP support, and Audionaut is explicitly designed to be driven by AI agents over MCP.
7. Why it matters for AI/ML development: MCP is emerging as a common way for agents to discover and operate external tools, including desktop and creative applications.

8. Implications/actionable takeaways: Developers should expose app capabilities through MCP, design permission and safety models for agent access, and test agent-driven workflows in professional software.

9. Agent-generated software is reshaping database and cloud infrastructure.

10. Trend/insight: Supabase is acquiring Turso because AI agents are spinning up millions of databases for prototypes, dashboards, and apps.
11. Why it matters for AI/ML development: Existing database provisioning is often too heavy and costly for ephemeral, per-agent workloads. SQLite and Postgres need to work together across the prototype-to-production spectrum.

12. Implications/actionable takeaways: Adopt serverless and embedded databases for lightweight agent workloads, automate provisioning, enforce cost governance, and ensure a clear migration path to production-grade Postgres.

13. AI product churn creates data portability and lock-in risks.

14. Trend/insight: Mozilla’s shutdown of Solo shows that AI-native website builders can be discontinued quickly, with data deletion and limited export options.
15. Why it matters for AI/ML development: Users are increasingly creating valuable work inside AI tools that may not survive commercial shifts. Trust depends on portability and transparent sunset policies.

16. Implications/actionable takeaways: Demand open formats and export APIs from AI vendors, self-host critical data where possible, and design AI products with migration and shutdown plans from the start.

17. Craft anxiety and emotional backlash are real adoption constraints.

18. Trend/insight: “AI Makes Me Sad” captures a growing sense among technical workers that AI threatens craft, control, and creativity, even for those who could profit from it.
19. Why it matters for AI/ML development: Adoption is not purely technical. Resistance, burnout, and cultural conflict can slow deployment and hurt talent retention.

20. Implications/actionable takeaways: Preserve human agency in AI-assisted workflows, invest in skill development, frame AI as augmentation rather than replacement, and involve practitioners in tool design.

21. Privacy-first and anti-Big-Tech-cloud sentiment will shape AI deployment.

22. Trend/insight: Home Assistant’s renaming of its cloud service reflects frustration with subscriptions, outages, data harvesting, and lock-in.
23. Why it matters for AI/ML development: Many AI services are cloud-dependent and data-hungry, which conflicts with growing demand for privacy and local control.

24. Implications/actionable takeaways: Offer local, self-hosted, or on-device AI options; use transparent data policies; support open standards; and make cloud usage optional rather than mandatory.

25. Security maintenance of foundational layers remains a gating factor for AI systems.

26. Trend/insight: The Debian Linux kernel advisory lists many CVEs, reminding us that AI and agent infrastructure runs on deeply layered, security-critical software.
27. Why it matters for AI/ML development: Agent systems increase attack surface and depend on kernels, containers, and cloud platforms that must be patched and monitored continuously.
28. Implications/actionable takeaways: Treat agent infrastructure as production security-critical, automate patching, maintain SBOMs, enforce runtime security, and avoid letting AI hype overshadow operational hygiene.

Analysis generated by deepseek-reasoner