Hacker News
1Local Wikipedia: Wikipedia, but only for things a half hour walk from you
- Local Wikipedia shows geocoded Wikipedia pages within about a half-hour walk of you. If fewer than 25 items are nearby, it expands to the closest 25 within roughly 6 miles, and caps results at 500. Your location stays in your browser and isn't included in shared links. The site is not affiliated with Wikipedia.
Nearby feature useful but limited
Users find Wikipedia's nearby-places feature useful but uneven. Rural areas yield few or distant results, while cities get flooded with streets, schools, and housing developments. Users want adjustable radius, keyword filters, and easier manual location entry.Local places feel more interesting
Commenters find articles about nearby places more compelling than distant ones, which they attribute to proximity, personal relevance, and an evolved tendency to favor one's own group. Some also note that a plane crash feels more significant when it happened nearby.Alternatives for discovering places
Commenters point to other resources for discovering places, including Atlas Obscura, Wikivoyage, and Chattymaps. One commenter also describes a project that unlocks location-based audio stories by visiting the site.
2Docker is weird
- The author built their own container runtime and found Docker surprising: containers are just restricted processes, images are simply folders, Go is an awkward fit for such tools, and Docker usually runs as root. Building your own tool is a great way to learn.
Docker's origins and lasting impact
The creator joined the thread, and commenters recalled his 2013 PyCon lightning talk. Many credited Docker with replacing Vagrant and LXC and reshaping local development and deployment.What made Docker unique
Commenters debated whether the container tech or the UX drove adoption. Dockerfiles, layer caching, registries, and momentum were cited, and some noted that chroot and earlier research already existed.Daemon and architecture criticism
Critics called the client/server daemon model needlessly root-dependent and pointed to Podman as a daemonless alternative. Defenders argued the daemon enabled consistent cross-platform developer experience.
3Tell HN: I unknowingly broke the PHP RCE embargo. I am sorry
- The author apologizes for posting about upcoming PHP remote code execution vulnerabilities (CVE-2026-109618 and CVE-2026-109619), which disrupted the coordinated disclosure schedule and added workload. They say they posted believing no embargo existed, having been told so by a contact, and intended only to alert PHP users to apply updates promptly. They accept full responsibility and say they have learned their lesson.
Embargo responsibility for uninformed third parties
Most commenters agree OP did not break the embargo because they were never told about it and were given false assurances. Some argue the breach happened earlier in the chain, and others say the process failed rather than any one person.Meaning of "take full responsibility"
Some see the phrase as a sincere admission of fault, while others call it empty corporate language that implies reparations OP never offered. Several commenters defend OP's apology as honorable.Blameless process over individual blame
Commenters urge a blameless post-mortem and a system that doesn't rely on individuals being perfect. Others note that unreliable links in the chain may need to be identified and bypassed.
4Today, we hacked what we'd set out to hack: the iPhone
- Researchers announce they have hacked the iPhone and present a series of demos covering attacks on the Signal app, Apple's iMessage, and an "Apple Persistent Threat." Full releases are promised later.
Jailbreaking and iOS openness
Commenters lament Apple's walled garden and lost root-access freedoms of early Android/iOS days, debating whether to vote with wallets, regulate, or buy open devices like Fairphone, Pixel, or Murena.PDF parsing as attack vector
PDFs are seen as a huge attack surface since they support scripts, images, and complex codecs; users suggest treating documents as dumb by default, while Lockdown Mode only partly mitigates one-click exploits.AI-driven vulnerability discovery
Researchers lament that frontier AI cyber-capable models are restricted to trusted insiders, while others argue LLMs help both attackers and defenders, and patching hardware-level bugs remains hard.
5Kimi K3.1 Pro and K3.1 Pro Highspeed
Kimi's usage limits make subscriptions impractical
Many users report the $99 plan hitting the 5-hour quota within minutes or under an hour, calling limits far too low versus Claude or Codex. Suggestions include pairing Kimi for planning with cheaper models like DeepSeek for coding.Kimi withholds cyber capability for safety reasons
Moonshot removed cyber-related training data from K3.1 Pro's open weights, citing safety. Commenters debate whether this is responsible restraint, a cop-out, regulatory positioning, or a loss for defensive security work.Kimi's blunt personality and benchmark standing
Users praise Kimi's argumentative, non-sycophantic style, which contrasts with US models that flip-flop. Artificial Analysis ranks K3.1 Pro 4th overall, prompting debate on whether China is close to the US frontier.
6Every AI model has the same "controversial" take on humans
- Thirteen AI models from ten labs were asked for their most controversial opinion of humans, three times each. Most gave the same take: people pick what their group believes, then invent reasons. Eight gave it every time. A few dissented, including Claude Opus 5.5, which argued humans are more decent than they believe.
Human decency and trust
Commenters share stories of strangers returning lost wallets and offering hospitality, but many argue this trust belongs to specific high-trust societies, and that cynicism about human nature is overstated.Model convergence on shared views
Commenters attribute the models' matching "controversial" takes to shared training data, cynical internet text, and RLHF smoothing out variety, rather than independent reasoning or real insight.Moral progress or constancy
Debate over whether morality changes over time. Some argue norms shift with material conditions and society, while others argue core imperatives like the Golden Rule have stayed constant for millennia.
7Ask HN: If the models are so good, why doesn't OpenAI just start 100 companies?
- The post asks why frontier AI labs like OpenAI and Anthropic primarily sell model access (tokens) instead of building their own applications. It argues that if the models are so good, the lab could earn more by launching products, and that falling token prices suggest more profit lies in the application layer than in the underlying utility.
Selling shovels beats mining gold
Commenters argue AI labs profit more from selling tokens than running businesses, much as McKinsey advises rather than operates. Companies also need people for relationships, execution, and judgment, not just brains.Commoditization would erode AI profits
If OpenAI or Anthropic could build a business with agents, others could too, and its value would collapse. Some suggest labs know profits are distant and rely on funding that depends on the hype.Real-world expertise limits AI businesses
Building the product is the easy part; knowing what to build, finding product-market fit, and managing context remain hard for LLMs, so human agency is still needed.
8Investor-owned utilities charge 23% more for electricity than public providers
- Since 2020, investor-owned utility electricity prices rose 34% versus 19% for public utilities, against 24% inflation. In 2025, homes paid 17.8¢/kWh at investor-owned utilities, 14.5¢ at public utilities, and 14.1¢ at co-ops. Most of the gap opened during the 2020-2023 energy price spike.
Privatized utilities failing consumers
Commenters cite Alberta, Ontario, and Texas as examples where deregulation and private ownership raised power bills, guaranteed utility profits, and led to deferred maintenance and deadly outages.Government efficiency versus profit motive
Debate over whether public bureaucracy is inherently inefficient. Some argue monopolies serving the public shouldn't be profit-driven, while others note government suffers political churn and that public services need slack, not pure optimization.Cooperatives as an alternative model
Members describe customer-owned cooperatives that build renewables, keep rates near market price, and pay dividends. Others praise municipal trash and water service over competing private providers.
9Jared Kushner Hacked Phone Reveals Secret Negotiations, Millions in Transactions
- Dutch intelligence secretly monitored Jared Kushner and found he controlled a UAE bank account funded by about $750 million from Emirati and Saudi royals for off-the-books foreign policy goals, routed through Azerbaijan and a UAE charity.
Flagging of the post
Users say a popular thread with 357 points and 169 comments was flagged and buried after eight hours. Some suspect bots, others suspect users who dislike political content in tech discussions.Kushner's hacked phone and government compromise
Commenters say the US government is widely assumed to be compromised after reports that Dutch intelligence accessed Jared Kushner's phone. Debate covers his unofficial envoy role, accountability, and possible pardons.Media bias over Hunter Biden's laptop
Commenters contrast the scant coverage of Kushner's phone with the heavy coverage of Hunter Biden's laptop. Some blame left-leaning media, while others say the laptop story drew attention for its salacious content.