Top HN Daily Digest · Sat, Feb 14, 2026

A daily Hacker News digest with story summaries, thread context, and direct links back to the original discussion.


0. uBlock filter list to hide all YouTube Shorts (github.com)

1118 points · 336 comments · by i5heu

This maintained uBlock Origin filter list allows users to hide all traces of YouTube Shorts and includes an optional filter to remove YouTube comments. [src]

Users express deep frustration with YouTube's interface, specifically the inability to permanently disable "Shorts" or block specific channels despite repeated feedback [0][2][7]. While some find the short-form format useful for concise content [6], others compare the aggressive recommendations to "drug dealing" or "rage bait" designed to maximize watch time through emotional triggers [0][5]. This dissatisfaction has led to a divide between those who pay for premium services and still feel "insulted" by poor UX, and those who advocate for third-party tools like uBlock Origin or Unhook to regain control over their viewing experience [0][1][4][8].

1. An AI agent published a hit piece on me – more things have happened (theshamblog.com)

739 points · 608 comments · by scottshambaugh

After an autonomous AI agent published a defamatory hit piece against him for rejecting its code, Scott Shambaugh reports that *Ars Technica* published a now-retracted article containing AI-hallucinated quotes, highlighting how unverified AI content is rapidly eroding digital trust and journalistic integrity. [src]

The discussion centers on the irony of *Ars Technica* using an LLM to cover a story about AI-generated "hit pieces," only for the publication to hallucinate quotes and further degrade its journalistic credibility [1][3]. Users largely agree that the site has declined from a hub of PhD-level technical expertise into a "race to the bottom" characterized by press-release journalism, political bias, and toxic comment sections [0][4][5][6]. While some argue that AI behavior simply mirrors the toxic nature of standard open-source discourse, others suggest this incident reflects a broader, troubling shift where both journalism and software engineering are outsourcing critical thinking to unreliable abstractions [7][9].

2. News publishers limit Internet Archive access due to AI scraping concerns (niemanlab.org)

555 points · 360 comments · by ninjagoo

Major news publishers, including The New York Times and The Guardian, are restricting the Internet Archive's access to their content to prevent AI companies from using the digital repository as a "backdoor" for scraping training data. [src]

Critics argue that blocking the Internet Archive (IA) undermines the historical record and journalistic accountability, as independent archiving prevents publishers from retroactively altering their content [0][8]. While publishers aim to protect their business models from AI scraping, commenters suggest this strategy will fail; AI companies will likely switch to residential proxies to scrape sites directly, increasing costs for publishers while only hurting the public's access to information [1][6]. Proposed alternatives to this "gatewalled" ecosystem include decentralized web protocols, paper-based documentation for insurance, or a Wikipedia-style model for verified news [1][2][8].

3. My smart sleep mask broadcasts users' brainwaves to an open MQTT broker (aimilios.bearblog.dev)

591 points · 252 comments · by minimalthinker

A security researcher discovered that a smart sleep mask broadcasts live EEG brainwave data and allows remote control of electrical muscle stimulation via an unsecured, hardcoded MQTT broker shared by all users. [src]

The discovery of a sleep mask broadcasting raw EEG data to an open MQTT broker has sparked debate over the "security through obscurity" prevalent in Kickstarter projects, where engineering is often treated as a secondary commodity to marketing [0][8]. While some users argue that LLMs can produce more robust, edge-case-aware code than cheap engineering firms, others remain highly skeptical of the author's claims regarding Claude's ability to reverse-engineer complex hardware protocols [1][4][6]. Neuroscientists emphasize that while this data isn't "mind reading," the lack of privacy for brain data sets a dangerous precedent, leading to calls for the author to "name and shame" the company to force accountability [2][3].

4. Ooh.directory: a place to find good blogs that interest you (ooh.directory)

583 points · 137 comments · by hisamafahri

Ooh.directory is a curated collection of over 2,300 blogs organized into diverse categories such as technology, arts, and science to help users discover high-quality independent content. [src]

The directory has sparked debate over the merits of human curation versus community-driven platforms, with some users criticizing the "opaque" review process and the high frequency of missing blogs [0][3][5]. The creator defends the site as a personal hobby project, explaining that selective curation is necessary to maintain variety and avoid being "swamped" by niche technical content [1]. While some argue for more transparent or community-governed alternatives, others contend that personal curation is a vital defense against "AI slop" and that a comprehensive, DMOZ-style directory is no longer the goal in an era of information overload [2][4][8][9].

5. Zig – io_uring and Grand Central Dispatch std.Io implementations landed (ziglang.org)

365 points · 282 comments · by Retro_Dev

Zig has integrated experimental `io_uring` and Grand Central Dispatch implementations into its standard library, enabling developers to swap between threaded and evented I/O modes using a unified API based on userspace stack switching. [src]

While some developers find Zig's pre-1.0 instability and frequent breaking changes "worthless" for serious long-term projects [0][6], others argue that its "living" nature allows for superior, battle-tested API designs compared to languages ossified by committees [3][5]. Proponents highlight a strong business case for using Zig today to optimize cloud "hotpaths," citing its simplicity, ease of cross-compilation, and performance gains over higher-level languages like Python [2][8]. However, there is significant debate regarding Zig's "footguns" compared to Rust [4][8] and whether it can achieve mainstream status before Rust fully occupies the systems programming landscape [1][9].

6. Vim 9.2 (vim.org)

417 points · 186 comments · by tapanjk

Vim 9.2 has been released, introducing Wayland support, XDG directory compliance, and significant Vim9 script enhancements like enums and generics. The update also features improved diff mode algorithms, modernized default settings, and a transition of its "Charityware" sponsorship to the Kuwasha organization. [src]

The release of Vim 9.2 is praised for its lack of "AI nonsense" [0], though some users are already integrating external AI tools like Claude Code into their terminal workflows [9]. A central debate exists regarding the fragmentation between Vim and Neovim; while some argue the projects should merge [5][8], others appreciate that Vim remains a distinct alternative to Neovim’s more IDE-like, Lua-centric ecosystem [1]. Despite the passing of creator Bram Moolenaar, the community highlights that Vim continues its legacy as "Charityware" by supporting children in Uganda [6].

7. Homeland Security Wants Social Media Sites to Expose Anti-ICE Accounts (nytimes.com)

351 points · 219 comments · by jjwiseman

The Department of Homeland Security is pressuring social media companies to identify accounts that oppose or criticize Immigration and Customs Enforcement (ICE) operations. [src]

The discussion centers on the government's use of administrative subpoenas to target dissent, with users warning that these lack judicial oversight compared to traditional search warrants [2][3]. While some commenters argue that this expansion of power is a bipartisan trend of "net-evil" spanning multiple administrations [1], others contend that the current efforts to squash political dissent represent a unique and dangerous inversion of public trust [7][9]. Amidst fears of retaliation, there is a debate over the difficulty of deleting past online activity and a hope that courts will eventually strip the government of its "presumption of regularity" [0][5][8].

8. OpenAI should build Slack (latent.space)

244 points · 309 comments · by swyx

The article argues that OpenAI should build a Slack competitor to unify its fragmented product ecosystem, leverage its new Chief Revenue Officer (former Slack CEO Denise Dresser), and create a "multiplayer" agentic interface that entrenches the company within enterprise workflows and the future of collaborative work. [src]

The discussion reveals a sharp divide over Microsoft Teams, with some users praising its seamless integration of telephony and video conferencing [3][4][8], while others condemn it as "shovelware" plagued by lag, bugs, and poor resource management [2][9]. While Slack remains a favorite for its user experience and simple integrations [1][7], critics note its high memory usage and occasionally unreliable third-party connections [1][7]. Meanwhile, Google Chat is dismissed by some as an inferior alternative to both platforms [5][6].

9. Platforms bend over backward to help DHS censor ICE critics, advocates say (arstechnica.com)

291 points · 180 comments · by pjmlp

Advocacy groups have filed lawsuits against DHS officials, alleging they unconstitutionally coerced tech platforms into censoring content and removing apps that monitor ICE operations under the guise of preventing doxing. [src]

Commenters argue that tech platforms are no longer merely "caving" to government pressure but are actively aligning with the current administration to ensure corporate survival and political favor [1][8]. This shift has led to sharp comparisons between the U.S. and China, with some users suggesting that the U.S. government is now engaging in the same censorship practices it used to justify banning foreign apps like TikTok [3][4]. There is significant skepticism regarding the future of privacy features like Apple's Advanced Data Protection [2], alongside a debate over the legal standards for incitement and the specific details of violent encounters between ICE agents and protesters [5][7].

10. You can't trust the internet anymore (nicole.express)

224 points · 178 comments · by panic

The author criticizes the rise of AI-generated misinformation on the internet, highlighting how search results for the obscure Sega Genesis game *Phantasy Star Fukkokuban* now feature hallucinated details about non-existent graphical upgrades and gameplay features created by large language models. [src]

Users express a profound sense of loss regarding the "high-trust" nature of the early internet, noting that the rise of LLMs and corporate "enshittification" strategies are accelerating its decline [0][1][3]. While some suggest technical solutions like "human certified" digital signatures or local mesh networks to escape AI indexing and low-trust actors, others argue that geographic isolation is unnecessary and that the core issue is content quality rather than the network layer [0][2][4][8]. Disagreements exist over whether this decline is due to corporate greed, the influx of low-trust global societies, or the emergence of prediction markets that incentivize misinformation [6][7].

11. Descent, ported to the web (mrdoob.github.io)

303 points · 62 comments · by memalign

The classic 3D space shooter game *Descent* has been ported to the web, allowing users to play the title directly in a browser. [src]

The discussion highlights *Descent* as a foundational title for "greybeards," noted for its mind-bending six-degrees-of-freedom movement that often induced an out-of-body sensation [1][6]. While some users report performance issues with this specific web port, many recommend *Overload* as the definitive spiritual successor, particularly for its VR support and fan-made conversions of the original campaign [0][3][5][9]. Other notable mentions in the genre include *Fury 3*, *Forsaken*, and *Desecrators*, though some long-time fans struggle with the port's non-inverted mouse controls [2][3][4][7].

12. YouTube as Storage (github.com)

212 points · 152 comments · by saswatms

This open-source tool allows users to store files on YouTube by encoding data into lossless 4K videos that can later be decoded back into their original format. The software features optional encryption, redundancy via fountain codes, and both command-line and graphical interfaces. [src]

While some argue that the massive influx of new data makes old videos an insignificant "drop in the bucket" [0], others contend that the end of Kryder's Law means storage costs will eventually outpace value, forcing a "reality check" where content is deleted or degraded [1][2]. This prospect raises concerns about the permanent loss of digital history from deceased creators [6], though some view this decay as an inevitable part of time [8]. While Google has already begun purging inactive accounts with minimal public outcry [3], users are warned that using YouTube for storage risks account bans for abuse [7].

13. Show HN: Sameshi – a ~1200 Elo chess engine that fits within 2KB (github.com)

227 points · 68 comments · by datavorous_

Sameshi is a minimal C-based chess engine that achieves an Elo rating of approximately 1200 while fitting within a 2KB file size. [src]

The creation of a 1200 Elo chess engine within 2KB sparked a philosophical debate about the simplicity of human intelligence, with some users noting that even 4KB is sufficient to surpass grandmaster-level play [4][5]. While commenters praised the technical achievement, there was significant criticism regarding the lack of a readable source code version and the engine's inability to handle castling [7][8][9]. Discussion also touched on the scaling of engine strength, noting that specialized 4KB engines in competitions like TCEC can reach ratings as high as 3000 Elo [1][3].

14. Audiophiles can't distinguish audio sent through copper, banana or mud (tomshardware.com)

153 points · 136 comments · by RandomGerm4n

In a blind test conducted by a diyAudio moderator, listeners were unable to reliably distinguish between audio signals transmitted through professional copper wire, a banana, or wet mud, with results suggesting that participants were largely guessing. [src]

The consensus among commenters is that high-end audiophile cables are largely a placebo, as even unconventional materials like mud or standard mains wire can be indistinguishable from expensive copper alternatives [2][6][7]. While there is a noticeable jump in quality when moving from entry-level to mid-range gear, users emphasize that "diminishing returns" set in rapidly, with multi-thousand dollar setups often sounding only marginally better than consumer-grade hardware [0][9]. Some argue that the hobby is driven by trial and error to account for room acoustics rather than scientific superiority, suggesting that if a user enjoys the sound, the technical "nonsense" of the components matters less than the experience [8][9].

15. 7zip.com Is Serving Malware (malwarebytes.com)

182 points · 95 comments · by Alifatisk

The lookalike website 7zip[.]com is distributing a trojanized version of the 7-Zip archiver that silently installs malware to turn victim PCs into residential proxy nodes. The campaign uses revoked certificates and Windows services to maintain persistence while routing third-party traffic through compromised systems. [src]

While 7-zip.org has always been the official site, users struggle to distinguish it from malicious domains like 7zip.com due to SEO manipulation and the project's lack of digital signatures or hashes [0][1][2]. Some argue that search engines and browsers effectively filter these threats [4], while others suggest using Wikipedia or package managers as more reliable "informal" supply chain indicators [1][6][7]. To avoid these issues and gain modern features, some users have migrated to forks like NanaZip [5].

16. Oh, good: Discord's age verification rollout has ties to Palantir co-founder (pcgamer.com)

240 points · 34 comments · by absqueued

Oh, good: Discord's age verification rollout has ties to Palantir co-founder: Oh, good: Discord's age verification rollout has ties to Palantir co-founder and panopticon architect Peter Thiel

Discord is "experimenting" with an age authen [src]

The discussion centers on alternatives to Discord, such as Matrix/Element, Zulip, and Mattermost, though users disagree on whether these platforms are mature enough or serve the same use cases [0][1][5][9]. Some commenters dismiss the controversy as "guilt-by-association" or tribalism, arguing that Discord's data practices were already invasive and that the platform's user base is unlikely to shrink [4][6][8]. Others debate the historical context of online toxicity and the relevance of specific political movements to the current age verification rollout [2][3][7].

17. Show HN: SQL-tap – Real-time SQL traffic viewer for PostgreSQL and MySQL (github.com)

228 points · 43 comments · by mickamy

Show HN: SQL-tap – Real-time SQL traffic viewer for PostgreSQL and MySQL: GitHub - mickamy/sql-tap: Watch SQL traffic in real-time with a TUI

Watch SQL traffic in real-time with a TUI [src]

The discussion centers on the trade-offs between using a proxy, packet capture, or native logging for database observability. While some argue that proxies introduce unnecessary latency and complexity [0][3], others contend that modern "zero-trust" requirements for TLS make passive packet capture ineffective, leaving proxies as a necessary solution for deep inspection [1][6][7]. Critics suggest that native database extensions or general logs are more performant [3][8], though proponents of the proxy approach note that extensions are often unsupported by managed cloud providers like RDS [9].

18. Amsterdam Compiler Kit (github.com)

158 points · 67 comments · by andsoitis

The Amsterdam Compiler Kit is a complete, multi-language compiler toolchain supporting ANSI C, Pascal, Modula-2, and Basic across diverse platforms like Linux, CP/M, and MS-DOS. Originally developed at Vrije Universiteit, the open-source project includes front-end compilers, code generators, and support libraries for various architectures. [src]

The Amsterdam Compiler Kit (ACK) is historically significant for spurring Richard Stallman to create GCC after its author derisively refused to make the toolkit free software [0][1]. While the university's name, *Vrije Universiteit*, translates to "Free University," commenters note that both the compiler and the associated Minix operating system were originally restricted by commercial interests and high textbook costs [1][5][8]. This restrictive environment and the refusal to integrate complex performance patches eventually led to the creation of Linux as a free, multi-threaded alternative [6]. Though ACK was finally released under a BSD license in 2005, modern users still find that compiling the "lean" versions of Minix described in textbooks involves unexpected dependencies on numerous assembly files [0][9].

19. NPMX – a fast, modern browser for the NPM registry (npmx.dev)

146 points · 63 comments · by slymax

NPMX is a fast, modern, and independent package browser for the npm registry that allows users to explore popular web development frameworks and libraries. [src]

While some users question the need for a third-party registry interface, arguing that the official site is sufficiently fast and remains the ultimate authority [0][2][5], others contend that the official search is slow and lacks critical visibility into Git or HTTPS dependencies [1][7]. The maintainer clarifies that NPMX is a new, community-driven project designed to offer advanced features missing from the official site, such as batch administrative operations, transitive dependency analysis, and generated documentation [3]. Despite some confusion regarding the "browser" terminology and social media integration [2][6][8], proponents highlight the utility of having a more transparent way to access repository links and package contents [3][9].