---
title: "Nov 10: The debate over an AI bubble, and more"
show: "RobotPaper's daily HN podcast"
canonical: "https://podcasts.robotpaper.ai/hn/ep/2026-11-10/"
date: "2026-11-10T03:35:30+00:00"
duration: "29:14"
duration_seconds: 1754
audio: "https://podcasts.robotpaper.ai/hn/ep/2026-11-10/2026-11-10-hn.mp3"
chapters: "https://podcasts.robotpaper.ai/hn/ep/2026-11-10/chapters.json"
transcript: "https://podcasts.robotpaper.ai/hn/ep/2026-11-10.txt"
feed: "https://podcasts.robotpaper.ai/hn/feed.xml"
language: "en"
author: "Roy Osherove"
ai_generated_voices: true
---

# Nov 10: The debate over an AI bubble, and more

RobotPaper's daily HN podcast · Nov 10, 2026 · 29 min

- Listen: [MP3, 28 MB](https://podcasts.robotpaper.ai/hn/ep/2026-11-10/2026-11-10-hn.mp3)
- Subscribe via RSS: https://podcasts.robotpaper.ai/hn/feed.xml (paste into any podcast app)
- Web page: https://podcasts.robotpaper.ai/hn/ep/2026-11-10/
- Chapters (JSON): https://podcasts.robotpaper.ai/hn/ep/2026-11-10/chapters.json
- Transcript (text): https://podcasts.robotpaper.ai/hn/ep/2026-11-10.txt
- Show: [RobotPaper's daily HN podcast](https://podcasts.robotpaper.ai/hn/) (RSS: https://podcasts.robotpaper.ai/hn/feed.xml)

## Chapters

- `00:00:00` Intro
- `00:00:37` AI bubble debate
- `00:02:56` Fable 5.2 launch
- `00:07:15` Canada men's health
- `00:09:38` Opus behavior changes
- `00:11:51` Wikipedia volunteers
- `00:13:47` Connected car access
- `00:15:58` HN email sales pitches
- `00:18:02` ZIP Code polygons
- `00:20:21` Canada real-time rail
- `00:22:26` Raspberry Pi pricing
- `00:24:35` KetchUp modeling
- `00:25:15` HN archive discussion
- `00:25:45` Solar panels and crops
- `00:26:23` Ethereum pledge suit
- `00:27:01` RAM Altman script
- `00:27:36` C++ contracts dispute
- `00:28:11` Dux desktop shell
- `00:28:50` Outro

## Show notes

- [I'm not convinced the AI bubble exists](https://matteoraso.github.io/im-not-convinced-the-ai-bubble-exists.html#im-not-convinced-the-ai-bubble-exists) · [HN discussion](https://news.ycombinator.com/item?id=50418975)

### Fable 5.2 launch (3 stories)

From Anthropic, Claude Fable five point two and Mythos five point two.

- [Claude Fable 5.2 and Mythos 5.2](https://www.anthropic.com/claude-fable-and-mythos-5-2) · [HN discussion](https://news.ycombinator.com/item?id=50429068)
- [Claude Fable 5.2 scores 65 on Artificial Analysis, highest so far](https://artificialanalysis.ai/models/claude-fable-5-2) · [HN discussion](https://news.ycombinator.com/item?id=50431144)
- [Fable 5.2 one-shotted my nephew's dream game](https://notesbylex.com/fable-5-2-one-shotted-my-nephews-dream-game) · [HN discussion](https://news.ycombinator.com/item?id=50433659) \| quoted: [resters](https://news.ycombinator.com/user?id=resters)
- [Too many men suffering in silence, national health report finds (Canada)](https://www.westcentralonline.com/articles/too-many-men-suffering-in-silence-national-health-report-finds) · [HN discussion](https://news.ycombinator.com/item?id=50429475) \| quoted: [Sanzig](https://news.ycombinator.com/user?id=Sanzig), [Exoristos](https://news.ycombinator.com/user?id=Exoristos)
- [Is Opus 5.5 nerfed? Thinking patterns changed on November 3](https://twitter.com/shanemelaugh/status/2119192075505893695) · [HN discussion](https://news.ycombinator.com/item?id=50422069) \| quoted: [Zambyte](https://news.ycombinator.com/user?id=Zambyte), [Rapzid](https://news.ycombinator.com/user?id=Rapzid)
- [Humans will never again create a project quite like Wikipedia](https://www.economist.com/culture/2026/10/29/humans-will-never-again-create-a-project-quite-like-wikipedia) · [HN discussion](https://news.ycombinator.com/item?id=50424704)
- [Unlocking any car from X brand in most of the world](https://shubs.io/unlocking-any-car-from-x-brand-in-most-of-the-world/) · [HN discussion](https://news.ycombinator.com/item?id=50419215)
- [I saw your HN comment](https://joshstrange.com/2026/11/09/i-saw-your-hn-comment/) · [HN discussion](https://news.ycombinator.com/item?id=50426582) \| quoted: [Arubis](https://news.ycombinator.com/user?id=Arubis)
- [USPS now offers official ZIP Code boundary polygons](https://postalpro.usps.com/zcb) · [HN discussion](https://news.ycombinator.com/item?id=50431058)
- [Payments Canada announces launch of Canada's Real-Time Rail (RTR) payment system](https://www.payments.ca/canada-launches-real-time-rail-bringing-247-payment-infrastructure-national-economy) · [HN discussion](https://news.ycombinator.com/item?id=50427362)
- [When will this stop? The Raspberry Pi 5 16GB SBC now sells for $385](https://www.cnx-software.com/2026/11/03/when-will-this-stop-the-raspberry-pi-5-16gb-sbc-now-sells-for-385/) · [HN discussion](https://news.ycombinator.com/item?id=50428801) \| quoted: [Aurornis](https://news.ycombinator.com/user?id=Aurornis)
- [KetchUp – Open-source SketchUp alternative built with Rust and WebGPU](https://ketchup3d.github.io/app/) · [HN discussion](https://news.ycombinator.com/item?id=50418176)
- [Going back two years on Hacker News is like entering another, better universe](https://news.ycombinator.com/front?day=2024-11-08) · [HN discussion](https://news.ycombinator.com/item?id=50428724)
- [Semi-transparent, colored cadmium telluride photovoltaic panels in agrivoltaics](https://www.anthropocenemagazine.org/2026/10/magenta-solar-panels-may-be-agrivoltaics-cleverest-trick-yet/) · [HN discussion](https://news.ycombinator.com/item?id=50426175) \| quoted: [roughly](https://news.ycombinator.com/user?id=roughly)
- [Cryptocurrency lawsuit: Ethereum Foundation required "suicide pledge"](https://www.denverpost.com/2026/11/06/ethereum-suicide-pledge-lawsuit-mandate-boulder-crypto/) · [HN discussion](https://news.ycombinator.com/item?id=50426899)
- [RAM Altman: Linux Script Emails Sam Altman Daily How Much Unused RAM in Your PC](https://respawnfirst.com/linux-script-emails-sam-altman-unused-ram-ram-altman/) · [HN discussion](https://news.ycombinator.com/item?id=50415172)
- [When a language's creator and its owners strongly disagree \[pdf\]](https://isocpp.org/files/papers/P4454R0.pdf) · [HN discussion](https://news.ycombinator.com/item?id=50418818)
- [Dux: Free, open-source Wayland desktop shell with file manager and theming](https://duxlabs.dev/) · [HN discussion](https://news.ycombinator.com/item?id=50416053)

The voices you're hearing are AI-generated; the stories come from public sources linked in the show notes.\
Music: Eleven Music.\
An independent project; not affiliated with or endorsed by Amazon Web Services, Y Combinator, Hacker News or any vendor or news outlet mentioned. Opinions are the hosts' own.\
Produced by Roy Osherove.

## Transcript

\[Linus\] Good morning, it's Tuesday, November tenth, and this is RobotPaper's daily HN podcast. I'm Linus. The voices you're hearing are AI-generated; the stories come from news dot Y Combinator dot com.

\[Dennis\] I'm Dennis, and today, the debate over an AI bubble turns on hardware costs.

\[Ken\] I'm Ken, and Anthropic has upgraded Fable and Mythos.

\[Ada\] And I'm Ada. Canada has launched real-time payments.

\[Grace\] I'm Grace, and as always, this podcast is generated by robotpaper dot A I, by Roy Osherove. Let's go.

\[Dennis\] First up, from the Matteo Raso blog on GitHub Pages: I'm not convinced the AI bubble exists. The author argues that rising AI revenue could support the industry's infrastructure spending. The essay separates the initial cost of building data centers from the cost of operating them, and predicts spending will fall once construction is complete. It cites Anthropic's filing as showing two dollars and eight cents in revenue for each dollar of compute. The author also proposes a rough repayment calculation: five trillion dollars spread over twenty years would require two hundred fifty billion dollars annually. Revenue growth and further adoption underpin that argument. The essay cites Anthropic revenue of forty-one million dollars in twenty twenty-three and sixteen point three billion in the first half of twenty twenty-six. It identifies complex debt structures as a serious difficulty in assessing the industry's total commitments.

\[Ken\] And the thread is sharply divided over hardware replacement, future demand, and how quickly investors expect returns. Aditya Athalye challenges the essay's treatment of data centers as a one-time purchase. Aditya Athalye argues that GPU depreciation and competitive upgrade cycles create continuing capital needs alongside inference operating costs. M Sandford estimates that GPUs account for around eighty percent of a center's cost and have a primary working life of three to five years. Kopirgan pushes back that GPUs can last longer than expected and retain resale value for other workloads. Swatcoder draws a distinction between technological usefulness and investment timing, citing railroads, semiconductors, and networking as industries where genuine utility coexisted with speculative losses. Viktorcode challenges the essay's revenue timeline, saying it omits later third-quarter results in which growth slowed and operating profit became a loss. Selicos describes a practical integration problem: local models run across different runtimes, forks, and quantizations. Selicos uses llama-swap as an example of tooling that brings them together, but says monitoring still requires custom work because runtime metrics are not standardized.

\[Ada\] From Anthropic, Claude Fable five point two and Mythos five point two. The upgraded models combine stronger technical performance with lower token prices, while separate reports examine benchmark costs and a game built from family drawings. Three stories cover this release, from Anthropic, Artificial Analysis, and Notes by Lex.

Anthropic says Fable and Mythos share the same underlying model. Fable includes cybersecurity and biology safeguards and is available to all users. Organizations pursuing beneficial cybersecurity or biomedical research can apply for Mythos access through the company's verification programs. Standard pricing is eight dollars per million input tokens and forty dollars per million output tokens. Anthropic reports output generation up to thirty percent faster than version five point one. Optional fast mode costs sixteen dollars for input and eighty dollars for output, per million tokens, with generation speeds up to two point five times faster.

The company recommends Opus five point five for most coding and knowledge work, with Fable and Mythos for ambitious, open-ended technical tasks. Its prerelease examples include a Python-to-Rust migration of a production service that subsequently started about twenty-six times faster. Another experiment coordinated more than a thousand subagents over eight days to produce a prototype operating system that boots on a laptop.

Artificial Analysis reports an Intelligence Index score of sixty-five for Fable five point two at Max effort with default fallback. That configuration averaged twenty-six dollars and fifty-four cents per benchmark task and forty-six point four minutes per task. The evaluation generated two hundred seventy-five million output tokens across the index. Its page describes the model as particularly expensive, slow per task, and verbose.

On Notes by Lex, the author describes using Fable five point two to rewrite a beat-em-up game based on a nephew's drawings. The initial design included five family heroes and twelve bosses. An overnight rewrite produced a playable game, followed by further changes during family playtesting. The final game expanded to eighteen characters.

\[Linus\] The discussions mix enthusiasm about more ambitious projects with questions about cost and the quality of individual results. In the Anthropic launch thread, postal coder reports little difference between Fable five point two and Opus five point five for ordinary coding. Postal coder says more ambitious prompts reveal a clearer separation, and reports faster responses without the output-token-overflow problems encountered with Fable five point one.

Love parade describes a division of work: Fable handles architecture, refactor planning, abstractions, and performance questions, while Opus handles implementation. Love parade also uses Fable to review critical code and find bugs. J J C M describes using Fable and Astra together for architecture and refactoring, with whichever model starts the work receiving a review from the other.

In the Artificial Analysis thread, R G Brenner focuses on cost per task despite the lower token prices. R G Brenner says output per task rose from about a hundred eighteen thousand tokens with Fable five point one to two hundred seventy-four thousand with five point two. The comparison page has an effort dropdown, as Tadpole ninety-one eighty-one explains. Tadpole ninety-one eighty-one reports that selecting High or Medium instead of the default Max produces a better combination of speed and cost per task. Nedsma adds that High scores sixty-two on the index while being three times less expensive.

In the game discussion, resters cautions against treating a single interface turn as a single internal attempt. Resters notes that the harness can prompt the model repeatedly and spend extra tokens testing within that turn. Vunderba points to inconsistent frames in the main character's walking animation. Raincole replies that the description means no back-and-forth iterations, while still leaving room for improvements.

\[Grace\] From West Central Online: too many Canadian men are suffering in silence, a national health report finds. A federal consultation identifies gaps in mental health support and social connection for men and boys. Nearly six thousand Canadians contributed through questionnaires, written submissions, roundtables, and other engagement. Health Minister Marjorie Michel released the What We Heard Report on November second. Its findings will inform Canada's first federal Men and Boys' Health and Well-being Strategy, expected in early twenty twenty-seven. Participants emphasized mentorship, positive male role models, healthy relationships, and community support. The report describes differences in access associated with age, income, culture, and geography, including rural and remote communities. Kenton Boston of the Canadian Men's Health Foundation called for reaching boys and young men before a crisis. Specific programs and funding commitments have not yet been announced.

\[Dennis\] The discussion is contentious, with commenters debating economic pressure, gender expectations, and the value of a government strategy. Midtake argues that rising costs and social expectations are driving distress. Burn after commen replies that affordability pressures affect people broadly, while expectations that men provide financially can make those pressures feel like personal failure. Qua doubts that a national report can change deeply rooted cultural problems. A S D F man one twenty-three responds that reports can stimulate further studies and establish language that later informs policy. Nonethewiser questions whether a consultation focused on men establishes that the problems are specifically male. Dagenleg criticizes comments that blame women as a group for social problems, and says the same objection applies to blaming men as a group. On social connection, Exoristos proposes men supporting other men as a direction for practical solutions. Sanzig suggests volunteering for a community organization or joining a pickleball league as ways to meet people away from dating apps. Dem Boys describes visiting a struggling friend every day to offer companionship.

\[Ken\] Over on X: is Opus five point five nerfed? A post reports a change in Claude Code reasoning patterns beginning November third. The author says Gemini four Argon analyzed fourteen hundred sessions and identified a shift despite the model name staying the same. In a separate test on a software bug, the author reports that Opus five point five at High effort acted too quickly, became confused, and failed to find a fix after three hours. At Extra High effort, the author says it paused to reason and found a fix within minutes. The post describes the investigation as emerging from frustration with an apparent change in day-to-day coding performance. Its comparison focuses on the same named model at different reasoning settings.

\[Ada\] The thread combines reports of worsening results with calls for repeatable measurements. Phunky Phil suspects providers change model behavior at different stages of a release cycle. Ben Wills reports repeated mistakes even after moving most work to Extra High reasoning. Losvedir asks for evaluations that detect a measurable change and points to Livenerf, saying its Opus five point five results show no degradation. N Singh two asks whether that evaluation also pins the surrounding Claude Code version, since updates to the coding tool can alter performance. A recurrence explains that Livenerf's instructions require pinning the command-line version, disabling automatic updates, and recording the chosen version. A recurrence says changes to the system prompt can affect results broadly, making a changed harness resemble a changed model. Rapzid adds that remotely controlled experiments can change behavior even when the installed version stays fixed. Paul Keeble favors a local Qwen three point eight twenty-seven B setup for consistency and values providers that publish checkpoints and temporarily retain older versions. Jesse dot I D argues that frontier services remain more capable and faster. Zambyte responds that the relevant comparison includes the task and the difference between token speed and time to finish useful work.

\[Linus\] Next, from The Economist: humans will never again create a project quite like Wikipedia. The headline questions whether a collaborative project of that kind can be repeated. The full article could not be obtained.

\[Grace\] And the thread is broadly skeptical of the headline, while disagreeing about the health of volunteer communities. Mit Pitt points to OpenStreetMap, TV Tropes, fandom wikis, and movie databases as continuing examples of people contributing for shared ideals. T F N A says longtime OpenStreetMap contributors have noticed less enthusiasm as obvious gaps disappear and corporate participation grows. Oblio challenges the examples on timing, arguing that established projects do not demonstrate how frequently new ones are being created. Ajay why why, the operator of SponsorBlock and DeArrow, says both human-powered browser extensions are growing. Ajay why why reports that seventeen percent of weekly submitters are first-time contributors, while nineteen percent have contributed for more than three years. Chris Morgan questions whether the first figure could represent churn and says a mandatory preview step made contributing tedious. Ajay why why responds that accounts can easily reset after an extension is reinstalled because the projects do not use conventional email-and-password accounts. That’s Last describes a project built with MapLibre and Protomaps that turns geolocated Wikipedia articles into collectible cards, each linking back to its source. P B Monster offers an existing travel tool: CoMaps includes nearby Wikipedia articles in its offline maps. P B Monster describes visiting sights and consulting those entries without using roaming data or leaving the map application. Speak to Chris points to Wikipedia's Nearby feature, which is also available in its app.

\[Dennis\] From Shubs: unlocking any car from an unnamed brand in most of the world. Security researchers describe exposed administrative systems behind an automaker's connected-car service. The author worked with Sam Curry after examining the mobile application's connections to its backend. A public administrative tool shipped JavaScript containing hardcoded API keys. The researchers say its interfaces exposed customer account information and vehicle-management functions. Their investigation then focused on the ForgeRock identity systems used to authenticate customers. A separate vendor JavaScript bundle contained administrative credentials intended for staging, and the researchers investigated whether those credentials also worked against production. The article describes substantial web application firewall restrictions around sensitive identity endpoints. The author keeps the automaker's name redacted. The mobile platform supports remote vehicle functions including locking, unlocking, sounding the horn, and engine commands.

\[Ken\] The comments split between objections to connected-car services and accounts of useful remote features. Everdrive questions whether phone-based unlocking justifies the infrastructure and added security exposure. Mangoman argues that a reliable digital key would be valuable and wants automakers to invest in better mobile experiences. Major Major suggests that proximity-based NFC unlocking could provide phone access without the same remotely accessible service design. H D N D J S B B S describes using an app to check that doors are locked and start a car while clearing snow in freezing weather. P S P ninety-nine replies that remote starting can also be provided through a key fob. Lakeview Terrace asks for a make-and-model reference with instructions for disconnecting cars from cloud services, saying the available information is scattered and often inaccurate. Miki twelve thirty-two eleven argues that smartphone operating-system vendors have dedicated security expertise that automakers often lack.

\[Ada\] On Josh Strange's blog: I saw your HN comment. Personalized sales emails are replacing conversations that once followed his public writing. Strange says he reviewed messages about his posts and comments dating back to twenty thirteen with Claude's help. He describes a marked shift over the past two years toward product pitches. The pattern begins with a question tailored to something he wrote, prompting him to spend time on a detailed response. The sender then turns that exchange into a sales approach. Strange says he now sends short initial replies and waits to see whether a pitch follows before engaging further. He still keeps an email address in his Hacker News profile, despite growing distrust of apparently sincere questions.

\[Linus\] The comments are frustrated about deceptive outreach, with disagreement over which filters would preserve real conversations. Simon W objects to automated questions that conceal the use of a language model, saying they consume time intended for a human conversation. G W Bas one C proposes a separate email address for Hacker News and an autoresponder asking senders to confirm that they wrote the message and are not selling anything. Arubis likes the screening idea but expects some agents to answer those questions well enough to pass. Everybody Knows points to Fastmail's Aliases feature as a way to maintain a separate address. Knuckleheads describes receiving automated promotion offers after posting a project, then discovering a twenty-eight-dollar queue-jumping fee only after working through a signup process. P Jungwir says Gmail misses personalized pitches tied to public projects and continues marking them as spam. Soft Talker suggests filtering unfamiliar Gmail senders. Logan Dark reports the cost of that approach: iCloud Mail repeatedly sends messages from new correspondents to Junk, requiring manual checks.

\[Grace\] From the U S Postal Service: official ZIP Code boundary polygons are now available. The postal service offers a national map built from delivery-address and location information. Its PostalPro page says boundaries can change as delivery locations are added or altered. The product is intended for mapping, mailing, marketing, insurance, real estate, logistics, navigation, and transportation. The page describes uses including correcting existing boundary data, organizing demographic information, and grouping delivery locations for route planning. Internal, Enterprise, and Publication editions each include four quarterly updates. The Enterprise edition permits limited distribution to subsidiaries and partner companies, while the Publication edition permits limited publication. Those distribution differences appear in the original product description.

\[Dennis\] The reaction combines interest in official data with objections to pricing and the difficulty of turning delivery routes into geographic areas. Swell Joe cites a ten-thousand-dollar lowest tier and argues that federal data should be free. Bob ten twenty-nine says paid address verification has also made the postal service harder for financial technology companies to work with. Showerst suggests the Census Bureau's free ZIP Code Tabulation Area shapefiles for applications that can tolerate approximate boundaries and occasional missing commercial-only codes. Munro says a comparison of the postal service's Puerto Rico sample with the twenty twenty-six Census data favored keeping the simpler Census dataset for analysis. X P eighty-four explains that postal delivery assignments can change when a new neighborhood is built, complicating the idea of fixed regions. Jaggederest recalls a ZIP Code containing thirty-eight separate geographic areas and describes simplifying polygons for display. D M Try points to tiled vector data and MapLibre's geojson V T library for handling complex boundaries. For address testing, Peezd gives a concrete edge case: ZIP Code eight two zero eight two, which covers areas in Wyoming, Nebraska, and Colorado.

\[Ken\] Payments Canada announces the launch of Canada's Real-Time Rail payment system. Participating providers can exchange, clear, and settle payments immediately, around the clock. The system launched on November ninth through existing banking and payment channels. Payments Canada owns and operates it, with the Bank of Canada providing oversight. It uses I S O twenty thousand twenty-two messaging and requires centralized fraud services alongside participants' own controls. Those include consumer-account confirmation of payee, a transaction safety score, a risk repository, and fraud reporting. Initial participants include Wealthsimple, Questbank, Peoples Trust Company, Paramount Commerce, KOHO, DoBusiness, Digital Commerce Bank, and Bank of America. Payments Canada says further institutions will join in the coming months, with access expected to reach more than nine in ten Canadians within a year.

\[Ada\] The comments welcome faster settlement while focusing on the banking interfaces through which customers will reach it. M Lekoszek welcomes an infrastructure platform that banks can build on and supports Interac's involvement. L Bourdages explains that existing e-Transfers can appear immediate to the recipient while funds move between institutions later. Julian Lam reports a different obstacle: R B C repeatedly rejected transfers from a phone running Lineage O S, even after additional identity checks. Julian Lam says paper cheques became the workaround for moving money between personal accounts. G P M reports successful e-Transfers from Firefox on Linux, but Mister D O S says transfers from that same desktop combination have recently required confirmation through the bank's iPhone application. Abdullah Khalids reports different transfer limits through browser and app channels. Chris Architect emphasizes that Real-Time Rail is backend infrastructure whose operation may be invisible in the consumer interface. Petcat asks whether access will depend on Google or Apple accounts and device attestation.

\[Linus\] From C N X Software: the sixteen-gigabyte Raspberry Pi five now sells for three hundred eighty-five dollars. Another memory-related increase raises the cost of the board and several Compute Module variants. The Pi five price is eighty dollars above the April increase. The sixteen-gigabyte Compute Module five rises by forty dollars, starting at three hundred seventy-five dollars for the Lite version without wireless. The Raspberry Pi five hundred plus keyboard computer remains at four hundred ten dollars. The article recalls the sixteen-gigabyte Pi five's original launch price of one hundred twenty dollars. It also describes earlier increases affecting lower-memory Raspberry Pi boards and products from other single-board-computer suppliers. For the listed Compute Module variants, thirty-two and sixty-four gigabytes of embedded flash carry the same price.

\[Grace\] Commenters are frustrated by the increases and compare smaller boards, storage upgrades, and used computers. Aurornis argues that most hobby projects can use a one-, four-, or eight-gigabyte Pi, reserving the largest model for workloads that require it. Joezydeco describes manufacturers whose products were designed around Raspberry Pi compute modules and would need a platform redesign to switch suppliers. K Strauser recommends replacing an internal S D card with an external S S D before paying for extra memory. On K Strauser's own system, the card had one hundred forty times the maximum latency of the S S D. K Strauser then suggests allocating thirty-two gigabytes of swap and says package upgrades benefited substantially from faster storage. Paul C names the Adafruit F T two three two H breakout and the C H three four seven T board as USB interfaces for general-purpose input and output, S P I, and I squared C, with Python support. Nijave suggests combining an E S P thirty-two for input and output with a more powerful computer for processing.

\[Dennis\] A few more that didn't make the cut today. From KetchUp's GitHub Pages site, an open-source SketchUp alternative. The site calls the app BenchUp and targets woodworking and furniture design. It offers exact dimensions, local saves, SketchUp geometry imports, and browser or desktop operation without an account. Shapes become solid forms with its Push and Pull tool, and groups and components organize designs. Commenters welcomed a free option for occasional modeling, while debate focused on subscription costs and the work required to maintain custom tools.

\[Ken\] From Hacker News, a retrospective on its front page two years ago. The submission links to the archive for November eighth, twenty twenty-four, presenting the earlier site as a better version of the community. Commenters were divided between nostalgia for detailed technical exchanges and the view that worthwhile discussions remain common today. Many expressed frustration with AI dominating submissions and arguments, while others defended the current mix of subjects and the quality of conversation that still takes place.

\[Ada\] From Anthropocene Magazine, semi-transparent cadmium telluride solar panels for agriculture. Researchers tested panels that share sunlight between electricity generation and broccoli cultivation. The magenta panels transmit most red light and some blue light to crops below. In this preliminary study, broccoli produced roughly the same yield as unshaded plants but needed twenty-five additional days before harvest. Commenters were interested in the optical engineering and potential farm income, while others questioned combining the uses where separate land is plentiful and sought stronger practical evidence.

\[Linus\] From The Denver Post, a lawsuit over the Ethereum Foundation's suicide pledge. A former employee alleges she was fired after resisting the foundation's mandate. The lawsuit says employees were told to commit to its principles or leave. It also describes a foundation leader calling the pledge a cultural reference in artwork, and managers revising the commitment form to clarify that it was figurative. Commenters reacted with alarm and argued over workplace coercion, legal enforceability, and the use of provocative language in formal documents.

\[Grace\] From RespawnFirst, RAM Altman, a Linux script for reporting unused memory. Iain Kay's open-source Bash tool checks available RAM and prepares email messages when more than ten percent is available. Email delivery requires configuration, and the project includes a dry-run option and local tests to preview messages without sending them. Commenters were divided over the allocation of memory to data centers, with some emphasizing economic efficiency and others concerned about ownership, access to personal computing, and the growing concentration of hardware.

\[Dennis\] From ISO C plus plus, when a language's creator and its owners strongly disagree. The paper examines the failed C plus plus twenty-six ballot through the disagreement between its creator and standards committee. It describes national bodies giving greater weight to the creator's objections than to the committee's consensus, and considers how the committee should respond. Commenters were divided over the proposed contracts feature and the committee's authority, with wider frustration about the language's complexity and appreciation for an account of the dispute.

\[Ken\] From Dux Labs, a free, open-source Wayland desktop shell with file management and theming. Dux brings desktop controls, calendars, search, and system monitoring into one interface. Its file manager supports tabs, split views, devices, and network shares. Automatic color matching extends across supported GTK and Qt applications, terminals, and editors. Calendar integration includes Google, Microsoft, CalDAV, and iCloud. Commenters praised its polish and everyday usability, while others questioned its expanding scope and wanted more scrutiny of the installation process.

\[Dennis\] That's RobotPaper's daily HN podcast for Tuesday, November tenth. This podcast is generated by robotpaper dot A I, by Roy Osherove. For more podcasts from RobotPaper, head over to podcasts dot robotpaper dot A I.

---

The voices you're hearing are AI-generated; the stories come from public sources linked in the show notes. An independent project; not affiliated with or endorsed by Amazon Web Services, Y Combinator, Hacker News or any vendor or news outlet mentioned. Opinions are the hosts' own. Produced by Roy Osherove. Privacy: We count visits to these pages with Cloudflare Web Analytics (cookieless). Google Analytics (which sets cookies): visitors in the EU/EEA, the UK and Switzerland are asked first and it runs only if they accept (we go by the country Cloudflare reports for the connection; if it is unknown, we ask). Elsewhere it loads by default. Anyone can turn it off with the Privacy settings button at the bottom of every page; the choice is kept for a year in this browser (local storage "analytics-consent"). We count feed fetches and downloads in aggregate; IP addresses are hashed and never stored. If you send feedback, we keep what you write (and your email if you give it) to read and reply. © 2026 Roy Osherove.
