Hacker News, Distilled

Scope, timelines, and “carefulness”

• Many argue the real lever isn’t “be more careful” but “do less”: clarify what won’t be built, cut low‑ROI features, and avoid overspecified requirements.
• Several anecdotes show big time savings when engineers speak directly with stakeholders, revealing that “big” asks can be solved by tiny changes.
• Some feel the industry has drifted from this discipline, with engineers pushed away from clients and forced to sacrifice quality instead of scope.

Risk, incidents, and postmortems

• Multiple comments emphasize explicit risk management: probability × impact, with clear choices between prevention, mitigation, and remediation.
• Some leaders deliberately choose “do nothing” after one‑off incidents, stressing that every safeguard has a cost and overreaction leads to slowdown.
• Others push for safety nets, faster rollback, and better monitoring rather than blanket “be more careful” directives.

Debate over the “carefulness knob” metaphor

• Supporters see it as a useful way to frame the tradeoff: too little care causes incidents that erase any speed gains; too much care slows delivery.
• Critics argue the metaphor is childish or misleading, preferring concrete discussions about specific risks, processes, and data.
• There is extended debate about the graph in the article (shape, slope, anchoring), with some warning that imprecise visuals invite unproductive bargaining in real meetings.

Process, bureaucracy, and over‑carefulness

• Many warn that each incident spawning a new check or approval leads to “grandpa’s keys”–style process accretion and bureaucratic misery.
• Some note cultural tendencies (e.g., heavy process, spreadsheets, manual approvals) that slow teams without clearly improving outcomes.
• Others counter that well‑chosen processes and guardrails can both reduce incidents and speed work, if regularly re‑evaluated and pruned.

Management, responsibility, and decision rights

• Strong view: don’t negotiate core quality/safety standards feature‑by‑feature; negotiate scope and priorities instead.
• Disagreement over who should own risk decisions: some say safety level is a management call; others stress engineers remain ethically responsible.
• Several comments criticize managers who push “go faster” without understanding tradeoffs; others defend competent management as essential to avoid building the wrong thing.

Automation and tooling

• Multiple comments stress that improving build times, tests, linters, and rollback tools is often a better use of “carefulness” than manual scrutiny.
• Emphasis that “things occasionally breaking” is acceptable; the goal is predictable risk with good recovery and clear SLAs, not zero incidents.

Power vs. Energy Units

• Large subthread debates the article’s use of “500 megawatts” for “total energy” from South African solar in 2019.
• Some argue this is technically wrong (confusing power and energy) and a common sign of poor understanding.
• Others counter that describing average annual output in MW is acceptable (energy per unit time), even if phrasing is awkward.
• Additional confusion arises over MW vs. MWp (peak capacity) and MWh; consensus is that clarity on units and context is often lacking in journalism and public discourse.

---

Scale and Economics of Solar

• Strong agreement that falling panel prices and simple installation at small scale are transformative, especially versus large, capital‑intensive plants (nuclear, coal, gas).
• Counterpoint: in the US, data show utility‑scale PV is growing faster than rooftop, driven by economies of scale and lower per‑Watt costs.
• Several note that in some countries (e.g., Australia vs US) rooftop costs diverge sharply due to permitting, labor, and regulatory differences.

---

Role of Storage and Grid Integration

• Many discuss solar’s intermittency and evening “duck curve” problems.
• Some claim solar beyond ~20% of the mix becomes much more expensive (3–5×) because of storage needs; others say storage costs are falling and diverse storage (batteries, pumped hydro, thermal) makes higher shares feasible.
• Examples: water chilling for cooling, sand or thermal storage for heating, and large pumped‑hydro paired with huge solar parks.
• Debate on whether “the grid is a battery”: conceptually true at system level but physically requires other generators or storage to balance.

---

Solar vs. Nuclear and Other Generation

• Ongoing argument over whether nuclear is indispensable or uneconomic relative to ever‑cheaper solar + storage.
• Pro‑nuclear points: long‑lived firm capacity, needed to limit gas build‑out and decarbonize heating; concern that renewables alone lock in natural gas for decades.
• Skeptical points: high capital cost, long timelines, cost overruns, proliferation/waste issues, dependence on complex regulation; doubts that small modular reactors will be cheap or fast.
• Some suggest a mixed portfolio (solar/wind for daytime/AC/EVs; nuclear or other firm sources for winter and baseload).

---

Policy, Regulation, and Trade

• Disagreement over calls for more subsidies/regulation when solar is already booming; others note existing growth is itself subsidy‑driven and that public support should now target storage and complementary firm power.
• Complaints about tariffs on cheap Chinese panels/EVs and about permitting, grid‑connection limits, and HOA bans that suppress rooftop adoption.
• Utilities are portrayed by some as blocking or capping distributed solar and capturing regulators.

---

Local vs. Utility-Scale and Resilience

• Some emphasize local solar for resilience in fragile grids (e.g., unreliable grids, rolling blackouts) and as a “backup that becomes primary.”
• Others stress that very local provision (home/municipal) can be more expensive and that broad, interconnected grids spread risk and cost more efficiently.
• Discussion extends this “local vs centralized” frame to food and water, with trade‑offs between resilience and price.

---

Global Deployment, Costs, and Coal

• Comments highlight China’s massive, largely subsidy‑free build‑out of solar and wind, dramatic price declines, and very large integrated solar + hydro/storage complexes.
• There is concern that overall electricity demand growth means more of every source is being built, including coal, especially in China and India.
• Some note “experience curves”: renewables keep getting cheaper as they scale, making it hard for slower‑learning technologies to compete.

---

Social, Political, and Market Frictions

• Reports of scammy residential solar sales in some US regions and fear of being left with nonperforming systems.
• Culture‑war framing: in some right‑leaning spaces, solar/wind are attacked as “woke” or tied to “green” politics; elsewhere, “degrowth” and nuclear shutdowns are criticized as self‑sabotaging.
• Overall, thread mixes strong optimism that cheap solar is reshaping power systems with concern over storage, grid upgrades, policy choices, and technology mix.

Scope of the concern: black plastic & flame retardants

• Linked study and NGO press release report brominated flame retardants and other contaminants in many black plastic food-contact items, often from recycled e‑waste (ABS, HIPS, PP).
• Several commenters stress a key nuance: most “dirty” recycled plastic is made black with carbon black to hide mixed waste; not all black plastic is bad, but almost all bad recycled plastic ends up black.
• Because consumers can’t tell virgin from recycled black plastic, many argue the only practical advice is: avoid black plastic that contacts hot food or drink, especially cheap utensils and takeout containers.

Where people worry this shows up

• Obvious: black plastic spatulas/turners, cheap utensil sets, dollar-store cookware, warped/melted tools.
• Less obvious: coffee makers (drip machines with black internals), kettle lids, thermos lids, Aeropress caps, drip-brew filter baskets, and black parts in water paths.
• Some suspect “food-grade” claims from brands may not fully control upstream resin sourcing; others think large brands likely specify and audit materials, but still see reputational risk.

Alternatives and practical cooking advice

• Strong push to use:
  • Stainless, carbon steel, cast iron, and enameled cookware instead of non-stick + plastic.
  • Metal spatulas on bare metal pans; wood or silicone on non-stick.
• Large subthread on cooking eggs and other “sticky” foods in stainless, cast iron, and carbon steel:
  • Techniques shared: preheating pans, “hot pan, cold oil,” appropriate oil quantity, temperature control (often with IR thermometers), letting food develop a crust before moving, deglazing for cleanup.
• Wood and silicone utensils are widely recommended, with caveats:
  • Wood can crack, harbor bacteria, or be glued/finished with unknown chemicals.
  • Silicone is seen as more inert than typical plastics but may still have additives and can absorb flavors.

Non-stick (PTFE/Teflon) debate

• One side: PTFE is inert below ~260–450°C, used in medical implants, and practical if not overheated or scratched; major risk is burning oil before PTFE decomposes.
• Other side: easy to overheat empty pans (especially on induction), polymer fume fever and bird deaths show real toxicity, PFAS manufacturing and leaching are serious concerns, coatings wear and are effectively disposable.
• Many have already replaced non-stick with stainless/carbon steel/cast iron on durability and safety grounds alone.

Risk framing, skepticism, and regulation

• Some see this as another alarmist “may cause cancer” story without clear real-world risk magnitudes; note replication issues and a single-researcher “crusade”.
• Others argue the precautionary principle applies because:
  • Avoiding black plastic utensils is cheap and simple.
  • Flame retardants and PFAS are plausibly harmful at very low doses and bioaccumulate.
• Broader point: individuals can’t continually “do their own research” on hundreds of products; commenters call for stronger regulation, mandatory testing, clear standards, and restrictions on harmful additives and single-use plastics.

Overall impression of the tech

• Many find the audio technically impressive and plausibly human, especially to non-native speakers.
• Several commenters describe a “holy shit” moment where their brain briefly accepted it as real conversation.
• Others emphasize it’s good but “not yet great,” especially around disfluencies (“um,” “uh”) and pacing.

Uncanny valley & “fake personality”

• A dominant reaction is discomfort: voices feel like over-enthusiastic podcasters, ad reads, or awkward people reading a script.
• Listeners dislike the exaggerated friendliness, faux excitement, and constant back-channeling (“oh yeah,” etc.), calling it grating and shallow.
• People say they’d find this style annoying even from humans; the issue is tone and persona, not just artificiality.
• Some report no uncanny valley, especially non-native speakers, but still don’t like the “talking over each other” podcast format.

Identity, style, and training data

• Commenters note the voices lack a coherent “person” behind them: mannerisms and vocabulary feel averaged from training data, not tied to a distinct identity.
• Accents are discussed (e.g., “British accent”), with recognition that lumping many regional accents together is imprecise.
• Several suspect training was skewed toward “professional audio” (ads, podcasts, audiobooks), leading to overfitted “podcaster banter.”
• The fake disfluencies feel mistimed and mechanical, which enhances the uncanny effect.

Use cases, tools, and adoption

• Proposed uses: low-budget voice acting, YouTube narration, “reaction-style” commentary, reading articles or documents.
• Some already use similar TTS tools (browser/OS features, commercial apps, cloud TTS APIs) to listen to blogs and papers.
• NotebookLM’s podcast-style summaries are reported as both engaging and, for others, depressing—seen as replacing careful reading with chatty overviews.

Societal and creative impact

• Concerns that AI-generated audio/music will flood platforms with low-effort content and “AI elevator music.”
• Worries that automating commercial creative work “eats the seed corn,” undermining the ecosystem that trains future human creatives.
• Others argue creative fields may eventually regrow as human-made work becomes a premium differentiator.

Display & Nano‑Texture / Matte Option

• Many are excited Apple reintroduced a matte‑like option (nano‑texture) on MacBook Pro for the first time in years.
• Concerns: nano‑texture’s susceptibility to damage, fingerprint/oil staining, and needing a special cloth. Some prefer simple workarounds (cloth over keyboard) or tempered glass protectors.
• Question whether nano‑texture will come to the MacBook Air; some suspect it may be used as an upsell on higher‑end models only.

RAM, Storage, and Pricing Strategy

• Strong approval for base RAM moving to 16 GB across M4 Macs and updated M2/M3 Airs, seen as overdue and improving longevity.
• Complaints shift to base 256 GB SSD on many models and high prices for internal storage upgrades versus much cheaper external SSDs.
• Memory laddering is criticized: some configs (e.g., base M4 Max) are capped at 36 GB, requiring a costly CPU upgrade just to access higher RAM tiers; 96 GB options are gone.
• Some argue 16 GB is enough for typical office/dev work; others insist laptops at these prices should start at far higher RAM or at least be user‑upgradable.

Performance, Benchmarks, and Upgrade Value

• Apple’s marketing comparisons to old Intel and M1 machines are seen as partly targeted at those users and partly as number‑inflation; real‑world M3→M4 gains are viewed as ~10–20% in many tasks.
• Single‑core performance of M4 is praised, but many M1/M1 Pro/Max owners say their machines still feel “fast enough” and see little reason to upgrade unless doing heavy builds, media work, or local AI.
• Several anecdotes: M1/M2 laptops remain quiet, cool, with excellent battery life even under dev workloads; fans rarely spin up.

LLMs, Unified Memory & AI Workloads

• M4 Max’s 128 GB unified memory and ~546 GB/s bandwidth are viewed as very attractive for local LLM inference; some already use M‑series desktops for this instead of renting cloud GPUs.
• Still, they’re far slower than datacenter GPUs for training; consensus is LLM inference on Macs is practical, full‑scale training is not.
• Debate over cost‑effectiveness: for occasional or privacy‑sensitive use, local makes sense; for heavy or frontier‑model use, cloud remains better.

Connectivity & Wi‑Fi 7

• Lack of Wi‑Fi 7 on new Macs (while iPhone 16 has it) disappoints many, especially those with Wi‑Fi 7 routers wanting near‑2.5 Gbps wireless.
• Others argue Wi‑Fi 6E is sufficient for most laptop workloads; Wi‑Fi 7 advantages (throughput, preamble puncturing, MLO) are seen as “future‑proofing” rather than essential today.

OS, Privacy, and Alternatives

• Asahi Linux praised but currently supports only M1/M2; M3/M4 support may take time. Many treat macOS as host with Linux VMs instead.
• Apple’s privacy posture and on‑device/“private cloud” AI are lauded by some, but others view notarization checks, closed hardware, and App Store control as incompatible with true ownership and privacy.

Real-world use cases discussed

• The demo task (moving parts between shelves) matches real jobs in auto manufacturing and Amazon warehouses, where humans currently pick from dense storage systems.
• Some see more compelling use cases in hazardous work (e.g., racking high‑voltage breakers, arc‑flash risk) or last‑mile delivery in human-centric buildings (stairs, elevators).
• Others argue the demoed task is already “solved” by conveyors, stacker cranes, and specialized pick‑and‑place systems.

Humanoid vs specialized robots

• One camp: humanoid form is powerful because it can be dropped into existing human workflows with minimal facility changes; “like a human but cheaper” is an easier sell.
• Opposing camp: wheels, gantries, and specialized machines are simpler, faster, cheaper, and already widespread; humanoids are seen as overcomplicated “gimmicks” for real factories.
• Some foresee humanoids as an intermediate phase, with later transition to highly optimized, non-human forms once processes are redesigned.

Technical capabilities & limitations

• Atlas is now fully electric; previous hydraulic versions were bulkier but more explosively dynamic.
• Locomotion and balance are widely praised, including recovery from disturbances and complex torso/leg counter-rotations.
• Manipulation is seen as the harder frontier; the demo uses large, structured parts and fixtures, far from unstructured bin‑picking.
• ML is reportedly used for perception and localization, but commenters note the video reveals little about how quickly it can learn new tasks.
• Battery life and charging/swapping logistics are viewed as critical open questions.

Comparisons and authenticity concerns

• Atlas is favorably compared to Tesla’s humanoid demos (seen as earlier, more teleoperated) but some argue Tesla may win on factory deployment scale.
• Chinese Unitree robots are noted as cheaper and improving fast; BD still seen as ahead in locomotion and real deployments.
• Some suspect CGI or heavy “polish” in other companies’ videos; BD’s current video is generally viewed as real, though some think past BD marketing may have enhanced performances.

Economics, labor, and social reactions

• Pro‑automation arguments: robots don’t need breaks, don’t unionize, can work 24/7, and simplify HR and scheduling.
• Skeptics question whether current humanoids are cost‑competitive with cheap human labor and established automation, especially given maintenance and programming overhead.
• Ethical concerns arise around job loss, treatment of robots (violent durability demos), and broader societal impacts if humanoids replace low‑skill work.

Scope of the paper and related work

• Thread sees this paper as part of a broader line: probing internal activations to detect truth/falsehood and “know-what-you-know” calibration.
• Several related papers are cited that claim LLMs often have internal signals about correctness even when not expressed in outputs, though at least one of these is heavily criticized as overclaiming relative to its figures.
• The specific contribution here is framed as: truth-related information is concentrated in certain “critical tokens” and mid-layer activations, which can be used to better detect errors.

Can LLMs encode “truthfulness”?

• One camp: “truthfulness” is not a meaningful internal property for systems trained only on token correlations; they learn patterns, not truth.
• Counter-arguments:
  • Humans also mostly learn from language and social consensus, not direct experience.
  • Models can learn categories like “false” or “trivia” from textual patterns (“X is not Y”) and infer that some unseen statements likely belong to the “false” category.
  • Truth-like dimensions could emerge in embedding space much like sentiment or sarcasm.

Hallucinations: inherent vs mitigable

• Strong skeptics argue hallucinations are fundamental: every output is a sample from a distribution, so “fixing hallucinations” is conceptually wrong; at best you reduce error rates.
• Others see value in:
  • Taxonomizing hallucination types and causes.
  • Building detectors that flag low-confidence or likely-wrong answers for human review.
  • Using uncertainty measures (entropy, calibration probes) or resampling strategies to reduce harmful errors.

Comparison to humans and continual learning

• Repeated analogy: humans also hold false beliefs and update via social “swarms”; current LLMs are static weights, which limits their ability to converge toward consensus truth.
• Some advocate multi-LLM “swarms” and online learning; others note this is technically and operationally difficult today.

Philosophical and definitional disputes

• Long subthread on whether human reasoning is “non-statistical,” whether any non-physical “soul” is implied, and whether talk of “knowing” or “truth” for LLMs is meaningful.
• Some want more precise, less anthropomorphic language (“internal error signal” vs “knows it’s wrong”) to avoid confusion.

Skepticism about research quality and hype

• Several comments complain that papers and headlines overstate findings (“we found the gene for cancer” vibe).
• Concerns include weak correlations, poor out-of-distribution performance, and the risk of cherry-picking papers that fit a preferred narrative (either “LLMs know” or “LLMs will always hallucinate”).

CEO “responsibility” and accountability

• Many criticize the phrase “I take full responsibility” as empty rhetoric when the CEO keeps the job, compensation, and upside while workers lose livelihoods.
• Proposed “real” consequences: resignation, forfeiting bonuses/stock, large pay cuts, or sharing the fate of laid‑off staff (job loss, loss of health insurance, job search stress).
• Others argue layoffs do not automatically imply bad leadership: markets change, bets fail, and risk‑taking is part of the CEO’s job. Firing CEOs after every failed bet would discourage hiring and experimentation.
• There is disagreement over whether Dropbox’s issues reflect mismanagement (overly complex org, over‑investment in failed initiatives) or normal maturation of a business.

Layoff rationale, scale, and business context

• Multiple commenters note Dropbox is profitable but low‑growth, in a commoditized space (file sync/storage) facing intense competition from Google, Microsoft, Apple, etc.
• View 1: This is classic cost‑cutting to please investors (Rule of 40, margin expansion) and wage suppression, not survival. Prior layoffs (11%, 16%) suggest ongoing strategic problems.
• View 2: As growth plateaus, it’s rational to “right‑size,” flatten management, and treat Dropbox more like a mature cash‑generating asset; some ambitious bets (Paper, email, new products) apparently didn’t pan out.

Severance, healthcare, and worker impact

• Package details (16 weeks plus tenure, Q4 vest, prorated bonus, keep devices) are seen by some as strong by US standards; others call it merely “okay,” especially given the difficult job market.
• Six months of paid COBRA is widely criticized as insufficient and extremely expensive for families; long job searches can outlast coverage.
• Thread branches into US vs. EU norms on severance, WARN Act mechanics, ACA vs. COBRA, and Medicaid eligibility.

Headcount, productivity, and org design

• Many are surprised Dropbox ever needed ~2,600 staff for “just file sync,” prompting debate on:
  • Compounding maintenance and feature work, scale/reliability needs, bespoke enterprise features.
  • Empire‑building managers, promotion incentives based on team size, and Parkinson’s Law.
  • Observations that companies often can cut 20%+ without obvious short‑term impact, but at the cost of morale, institutional knowledge, and long‑term innovation.

Product, pricing, and user sentiment

• Some praise Dropbox’s UX and reliability versus iCloud/OneDrive, and are willing to pay more; others left due to higher prices, feature bloat, and “invasive” clients.
• There is concern about AI‑related data use and toggles enabled by default, undermining trust.
• Several suggest Dropbox should focus on core sync, lower prices, and accept “maintenance mode” rather than chasing unfocused new products.

Corporate communication and culture

• Mixed reactions to the announcement’s tone:
  • Critiques of euphemisms (“Dropboxers,” “macro headwinds”), and desire for plainer honesty (“we tried things, they didn’t work; we’re cutting costs”).
  • Some think, relative to other tech layoffs, the message and implementation are about as respectful as this kind of event gets.

Overall reaction to Apple’s “Apple Intelligence” ads

• Many find the AI ads “bleak,” “dystopian,” and morally off‑putting, especially the ones where:
  • An employee uses AI to sound “professional.”
  • A woman uses AI to fake remembering someone’s name and to pretend she read a colleague’s email.
  • A spouse uses AI to cover up forgetting a birthday with a last‑second auto‑generated collage.
• Critics say the ads normalize and celebrate lying, laziness, and disrespect, framing deception as “genius” rather than something to feel uneasy about.
• Defenders argue the spots are clearly meant as light sitcom‑style humor, exaggerating familiar failings (forgetfulness, procrastination) and not intended as moral instruction.

Ethics, culture, and “infantilizing” tech

• Several commenters see the ads as emblematic of a wider trend: AI pitched as “let the computer think for you” instead of “think better with the computer.”
• Concern that this reinforces intellectual passivity, undermines personal responsibility, and accelerates a culture where nothing can be trusted (deepfakes, AI‑written messages).
• Some tie this to broader worries about narcissism, dishonesty, and loneliness in modern consumer culture.

Comparison with Ubuntu / Canonical ad

• The Ubuntu animation is widely panned as generic, jargon‑heavy, and emotionally flat; many say it doesn’t explain what Ubuntu is or why anyone should care.
• A few agree it’s less objectionable than Apple’s ads but also note it “tells” buzzwords (“secure,” “performant,” “certified”) instead of “showing” concrete benefits.
• Some argue the author’s praise of the Ubuntu ad undermines their credibility about Apple’s ads.

Apple marketing, past and present

• Multiple comments contrast the new AI campaign with earlier Apple ads that were aspirational, creative, or playful without encouraging deceit.
• Others say the Steve Jobs era of emotionally resonant, “Think Different”‑style messaging is gone; current Apple feels more cynical or confused about AI’s value.
• The fear‑based Apple Watch and crash‑detection campaigns are cited as another shift: “buy this or you/your kids might die,” which some defend as legitimate safety marketing and others see as tasteless.

Views on AI itself

• Many see consumer AI use cases as flimsy—mostly about cheating at work or outsourcing trivial tasks.
• Others report genuine usefulness in narrow, transparent roles (translation, transcription, accessibility, meeting prep) and argue the best AI is often invisible rather than the star of the ad.

Overall reaction to DCPA ban

• Many welcome the ban, especially given fetal toxicity and birth-defect risks.
• Strong frustration that EPA classified DCPA as a likely carcinogen decades ago yet only now issued an emergency stop-use order.
• Some worry that prohibiting sale may push remaining stock to weaker-regulation countries, citing historical precedents with other pesticides.

Glyphosate comparison and pesticide risk

• Some argue glyphosate should be next, citing repeated patterns: chemicals approved, widely used, then later restricted (DDT, BPA, PFCs, DCPA).
• Others say glyphosate is not comparable: toxicology data suggest relatively low acute toxicity, and classification as “probably carcinogenic” is still debated.
• Distinction raised between pure glyphosate and commercial formulations whose surfactants/adjuvants may be more toxic.
• A meta‑analysis linking glyphosate-based herbicides to non‑Hodgkin lymphoma is mentioned as “compelling” evidence, while others point to sources describing overall low risk.

Regulation: EPA, IARC, and timelines

• Clarification that EPA and IARC use different classification systems and do not bind each other.
• EPA’s “likely carcinogen” category is described as based mainly on animal and mechanistic evidence with limited or no clear human data.
• Some note the U.S. acts more reactively, reopening pesticide reviews roughly every 15 years; calls for more frequent and better-funded re-evaluations.
• Others ask why re-checking is periodic rather than triggered only by new data.

US vs EU regulatory philosophy

• One view: U.S. is more libertarian and reactive; Europe more precautionary and protective of citizens.
• Counterview: EU regulation often protects domestic agriculture/economic interests rather than people; skepticism that Europe is truly safer overall.
• Examples surface on both sides (pesticide bans, antibiotics access) without clear resolution.

Externalities and moral limits of pricing harm

• Debate on whether harms to fetuses/unborn children could be “priced in” as an externality vs being inherently moral/non‑monetizable.
• Some propose actuarial-style costing (medical costs, lost earnings, disability-adjusted life years).
• Others stress practical difficulties: attribution, slow or subtle harms, and unmeasurable losses (e.g., reduced IQ, lost potential).
• Several argue some harms justify outright bans rather than taxes or price adjustments.

Language: “unborn babies” vs “fetuses”

• Disagreement over terminology: some see “unborn babies” as politically loaded, typically used by anti‑abortion advocates.
• Others argue it’s a straightforward humanizing term with historical analogs and that focusing on wording here distracts from pesticide safety.
• Concern that language shapes public attitudes about abortion and fetal status, so neutral terms like “fetus” are preferred by some.

Repetition of GLP‑1 Threads on HN

• Several commenters are fatigued by frequent GLP‑1 posts, noting the same questions repeat (weight‑loss mechanism, side effects, addictiveness, dosage, evolution).
• Others argue repetition is natural as new readers arrive and forums lack good knowledge aggregation.

Medical Effects and Evidence

• Commenters highlight trial data showing meaningful cardiovascular benefits that appear before major weight loss.
• Multiple firsthand reports describe large, rapid weight loss and improved blood sugar control (including avoidance of bariatric surgery).
• Some note GLP‑1s have been in clinical use for diabetes for ~15–20 years and are analogues of a natural hormone with extended half‑life.

Safety, Addiction, and Long‑Term Risks

• One side insists GLP‑1s are not amphetamines, not known to be addictive, and have relatively mild side‑effect profiles, with severe events rare.
• Skeptics invoke the opioid crisis as a warning against early claims of non‑addictiveness and emphasize unknown long‑term effects and permanent dependence on a drug.
• Side effects listed include gastrointestinal issues, gallstones, pancreatitis, possible thyroid risk, and concerns about post‑discontinuation weight regain.
• Some argue that compared with known harms of untreated obesity and diabetes, the risk–benefit still strongly favors GLP‑1s.

Willpower, Morality, and “Cheating”

• Strong disagreement over whether obesity mainly reflects lack of willpower versus complex biology, environment, and mental health.
• Several compare blaming obese people to telling depressed or psychotic patients to “just stop” their symptoms.
• Others maintain that diet and exercise can achieve the same or better results, and that drugs are being used to treat behavior rather than the body.

Cosmetic vs Medical Use and Social Pressure

• Debate over whether prescribing GLP‑1s for “cosmetic” weight loss is acceptable, given that excess weight correlates with many health risks.
• One long comment worries that easy weight loss will intensify body‑image pressure and stigmatize people who remain larger, including children.
• Others counter that positive self‑image doesn’t negate objective obesity risks and that individuals should be free to use effective drugs.

Societal and Economic Implications

• Some predict GLP‑1s could rival antibiotics or birth control in societal impact if they dramatically reduce obesity rates and related healthcare burden.
• Others see hype parallels with AI, noting earlier breakthroughs like antibiotics, insulin, and chemotherapy as more transformative so far.
• There is speculation about food‑industry impacts: reduced appetite might lower junk‑food consumption, but also creates a lucrative new pharma market.
• Concerns are raised about media and pharma incentives, and whether coverage (e.g., in The Economist) is overly promotional or quasi‑sponsored.

Technical and Philosophical Side Notes

• Arguments over the phrase “among the most important” (seen by some as empty marketing, others as a reasonable hedge).
• Disagreement on BMI’s value: some call it crude but useful at population scale; others stress better measures and contextual clinical markers.
• A few comments note the philosophical oddity of using drugs to suppress a fundamental drive like hunger, though others say this isn’t meaningfully comparable to other “wants.”

Attribution of Emissions from Investments

• Disagreement over whether an investor’s portfolio emissions should be counted as “their” pollution.
• Critics argue selling shares doesn’t reduce real-world emissions and see the framing as misleading.
• Others counter that capital allocation determines which companies can raise money and grow; investing in fossil-heavy firms perpetuates pollution and legitimizes it for others.
• Some argue large fortunes could be redirected to renewables and reshape incentives for incumbents.

Policy Tools: Taxes, Tariffs, and Offsets

• Strong support from several commenters for carbon taxes and border tariffs tied to equivalent pricing on emissions.
• Debate over whether tax revenue should be redistributed to citizens, fund decarbonization, or pay for carbon removal.
• Concern that carbon taxes can be regressive, especially for heating and basic energy; proposals include pairing them with wealth taxes and efficiency programs.
• Offsets are widely viewed as ineffective or scam-prone; regulated sequestration and direct air capture are seen as more credible.

Aviation, Rockets, and Billionaire Lifestyles

• Some emphasize that aviation is under 2% of global emissions; private jets of a few individuals are arithmetically negligible.
• Others argue “every bit counts” and that elite excess is morally and politically important, even if numerically small.
• Space launches are noted as intensive per event but seen by some as less important than electrifying ground transport and cleaning the grid.

Systemic vs Individual Responsibility

• Many see focus on personal “carbon footprints” as a diversion from systemic change and major industrial sectors (energy production, heavy industry, shipping).
• Arguments that capitalism “needs fossils” versus counterclaims that it just needs energy and could have been powered by nuclear/renewables.

Oxfam’s Framing and Credibility

• Several commenters distrust Oxfam, citing past reports they consider cherry-picked or ideologically driven.
• Others defend the general thrust of highlighting inequality and high-emitter responsibility, even if specific numbers are debated.

Equity, Power, and Climate Outcomes

• Frustration that ordinary people are pushed to conserve while the ultra-rich fly private and own yachts.
• Some argue rich individuals have done substantial good via EVs and solar but still should face stricter constraints or higher progressive carbon pricing.

Rice cooker mechanisms & physics

• Several comments expand on how modern rice cookers work beyond the article:
  • Many use a Curie-point alloy and magnet: the pot sits on a metal puck that loses magnetism just above 100 °C, releasing a spring and switching from “cook” to “keep warm.”
  • Others (and the original design described in the article) use bimetallic switches and “bang‑bang” control based on the latent heat of water: temperature only rises above boiling once liquid water is gone.
  • Indirect-heating “double-boiler” style cookers (common in Taiwan) boil away water in an outer reservoir; most current consumer models don’t use this approach.
• Comparisons are drawn to thermostats, electric kettles, dryers, and old turn signals using similar bimetallic or thermal tricks.

Instant Pots, pressure cookers, and rice cookers

• Some argue an electric pressure cooker (e.g., Instant Pot) makes rice as well or better, faster, and replaces a single‑purpose device.
• Others strongly prefer dedicated rice cookers:
  • Simpler, lighter, one‑button operation, easy cleaning.
  • Better texture, consistency, and long “keep warm” capability; can be used multiple times a day.
  • High‑end “fuzzy logic” / pressure rice cookers in Asia are praised for quality, though some see “fuzzy” as partly a marketing buzzword for standard control algorithms.
• Practical constraints noted:
  • Instant Pots need heat‑up/pressure/release time and can’t cook by “water fully absorbed” directly.
  • If the pressure cooker is busy with other dishes, a separate rice cooker is still useful.

Cooking methods and convenience

• Many share alternative methods: simple stovetop techniques, heavy claypots, microwaving, “pasta‑style” boiling with excess water, and “pot‑in‑pot” pressure cooking.
• The main value of rice cookers is framed as “asynchronous cooking”: set‑and‑forget, reliable results, and warm rice available all day, especially in households eating rice with most meals.

Health concerns: arsenic and rice

• Some references warn about arsenic in rice and recommend soaking and/or boiling in excess water then discarding it.
• Others consider this overblown or impractical, noting:
  • Arsenic levels vary by source and are also present in brown rice bran and soils generally.
  • Billions eat rice daily and countries with rice‑heavy diets often have high life expectancy.
• Consensus: thorough rinsing and, optionally, cooking in surplus water reduces exposure; evidence and recommendations are mixed.

Critiques of the article and broader context

• Multiple comments say the article glosses over modern designs and contains factual inaccuracies (e.g., about Chinese cooker styles).
• There is discussion that contributions of housewives and domestic inventors are often under‑credited, but some push back that examples do exist, and the phrase “not often” is hard to quantify.

Safety in Rust: memory vs I/O vs leaks

• Commenters stress the issue is not memory unsafety or UB.
• Rust’s notion of “safety” covers memory safety (no corruption, UB, data races), not resource leaks.
• I/O safety RFC is about preventing unauthorized handle access, not guaranteeing clean shutdowns.
• Several people say the article conflates “I/O safety” with “not leaking FDs,” which is a regular bug, not “unsafety” in Rust’s usual sense.

Async Rust model and cancellation

• Rust Futures assume: progress only when polled, and cancellation by dropping the future.
• This works well with epoll-style readiness, where spurious wakeups are cheap.
• With io_uring and Windows-style completion APIs, operations may complete in the kernel after a future is “cancelled,” creating tricky races.
• Lack of “async drop” makes non-blocking cleanup on cancellation difficult.

io_uring integration challenges

• The core bug discussed is FD leaks (and possibly dropped connections) when accept futures are cancelled while an io_uring accept completes.
• Some argue this is a known, fundamental race in any cancellation model: cancellation can always lose to completion.
• Others say a correct executor can still manage this by tracking completions and either cancelling or cleaning up resources when tasks lose interest.

Is this a Rust problem or a library problem?

• Many commenters frame it as a flawed or incomplete io_uring runtime API (e.g., monoio), not a flaw in “Async Rust” itself.
• Examples are given of alternative designs that:
  • store completed-but-unconsumed accepts and yield them later, or
  • register cancellation callbacks that run when completions arrive.
• There is disagreement whether Async Rust and io_uring are fundamentally mismatched or just require careful reactor design.

Broader views on async, APIs, and abstractions

• Some see this as another example of async Rust being easy to misuse and argue many codebases should avoid async entirely.
• Others defend async/await and Rust’s model as powerful but unfinished and demanding careful engineering.
• A side discussion notes that each OS async mechanism shift (select → epoll → io_uring) tends to invalidate large parts of existing abstractions; generic “one-size-fits-all” async layers are hard to design without leaks in the abstraction.

Scope of the NYC Change

• Jaywalking (crossing outside crosswalks or against signals) is now legal in NYC.
• Pedestrians may cross anywhere and against signals but must yield to vehicles with right of way.
• At marked/unmarked crosswalks with a walk signal, pedestrians still have right of way.
• Many note this largely matches longstanding NYC practice; main effect is removing a tool for selective enforcement.

---

Jaywalking Laws & Enforcement

• Multiple comments stress jaywalking violations have been used as pretexts to stop, search, and fine people, disproportionately affecting Black and brown residents (“walking while black”).
• Some describe courtroom examples where judges criticized such pretextual stops.
• Others argue that simply changing the law won’t end harassment; police can still improvise justifications.

---

Safety & Street Design

• Strong debate over where it’s actually safer to cross:
  • One side: mid‑block crossings are safer because there are fewer conflict directions and drivers are looking straight ahead.
  • Other side: intersections are safer due to lower speeds and expectation of crossings; some claim (contested) that most pedestrian deaths involve “jaywalking.”
• Counter‑evidence in the thread: most deaths occur on high‑speed arterials, often at night and without sidewalks, not mainly classic dense‑city mid‑block crossings.
• Many emphasize “being right” legally is irrelevant if you’re dead; pedestrians should assume drivers may be inattentive.
• Modern vehicle design (SUVs, thick A‑pillars, higher hoods) is cited as reducing visibility for drivers, especially at intersections.

---

International & Cultural Comparisons

• Wide range of norms:
  • Very strict compliance with signals in parts of Germany, Poland, Austria, Finland, Tokyo/Vienna.
  • Lax/“decorative” treatment of signals in southern Europe, Amsterdam, much of the UK.
  • Some countries criminalize crossing on red or within X meters of a crosswalk; others treat crossings almost entirely as a matter of judgment.
• Several note campaigns in Germany and elsewhere that framed red‑light crossing as endangering children.

---

Pedestrians vs Drivers & Urban Priorities

• Big philosophical split:
  • Car‑centric view: strict rules for pedestrians are needed because fast, heavy vehicles make free crossing inherently dangerous.
  • Pedestrian‑first view: streets existed before cars; jaywalking laws are “regulatory capture” by the auto lobby and encourage hostile, car‑dominated design.
• Some argue that mixing modes with lower speeds and mutual caution (as seen in some European and non‑US cities) can be safer than strict segregation.
• Concern raised that legalizing jaywalking could lead to more physical barriers (fences, railings) if conflicts increase.

Deep-Space Communication & Latency

• Commenters emphasize the extreme latency: ~45-hour round trip for commands.
• Discussion clarifies you can improve throughput (more data per unit time) via relays or better modulation/coding, but you cannot beat speed-of-light delay.
• Adding intermediate satellites would not reduce latency; extra hops add delay unless the main problem is extremely low data rate rather than light-time.
• Quantum entanglement as FTL communication is dismissed as incompatible with known physics.

Voyager’s Radios, Bands, and Deep Space Network

• NASA reactivated Voyager 1’s lower-power S‑band transmitter, unused since 1981, as a backup after an issue with the main X‑band path.
• Some were surprised NASA wasn’t sure S‑band would still be detectable; replies note DSN sensitivity has improved (larger/better antennas, arrays).
• Clarifications: X‑band’s advantage is higher antenna gain at both ends, not the frequency itself. S‑band has a wider beam and is more forgiving of pointing.
• DSN can array multiple dishes at a site and supports interferometry for precise tracking.

Power Source and Mission Lifetime

• Voyagers use RTGs, not solar panels or batteries. Power declines as plutonium decays, forcing gradual shutdown of instruments.
• The S‑band option may extend communications slightly at lower power, though eventually not enough energy will remain even for the radio.
• Discussion touches on RTG design limits, isotope half-life, and post–Cold War isotope availability; longer life mostly means more fuel or different isotopes.

Security and “Hacking” Voyager

• Protocols appear to have no encryption or strong authentication; in principle anyone could send commands.
• Practically, only DSN-scale antennas and specialized equipment can reach it, and the scientific/strategic value of hijacking is seen as negligible.

Engineering Durability vs Consumer Products

• Many contrast Voyager’s 47-year reliability with short-lived appliances and electronics.
• There’s debate over planned obsolescence vs value engineering and consumer price sensitivity.
• Some note survivorship bias in nostalgia for “old, durable” hardware and argue modern tech can be very reliable but is often cost-cut.

Software, Documentation, and Long-Term Projects

• Voyager code and documentation are scattered across decades of media; maintaining it is described as a “wizardly” effort.
• Thread contrasts this with fragile modern software stacks (e.g., JS builds breaking after only a few years) and calls for better documentation, stability, and maintenance culture.

Current Scientific Value & Cultural Impact

• Voyager still provides unique data on the heliopause and interstellar medium (density, plasma “sounds”), plus long-baseline trajectory measurements.
• Some question its marginal scientific value now; others strongly defend its inspirational role and the uniqueness of direct interstellar measurements.
• Multiple recommendations for the documentary “It’s Quieter in the Twilight,” highlighting the aging team keeping Voyager alive.

Overall reception of the article

• Many commenters praised the writing style as clear, technical, and humorous without being gimmicky.
• Some appreciated that it was long and detailed without mentioning AI.
• A minority found the tone grating or disliked the style.
• Several readers noted irony that a deep post on time doesn’t visibly show its own publication date; the author later said dates exist in metadata but aren’t rendered.

---

Weird and nonstandard time zones / practices

• Beyond Lord Howe, commenters listed many “weird” zones: Chatham Islands (+45 min), Australian Central Western (Eucla, +8:45, informal +9:45), Kathmandu (+5:45), Dublin historical offsets, Moscow’s +02:30:17 in 1900, and UTC+14 Pacific islands.
• Some zones are unofficial but widely used (e.g., Eucla, local “reservation” zones in the US).
• Ethiopia and parts of East Africa use a 12‑hour clock anchored to dawn and dusk; locals effectively offset official time by six hours and use a different calendar with 13 months.
• Historical oddities: Sandringham time, double DST in wartime UK, old Riyadh rules tied to sunset, Palestinian and Israeli DST rules creating overlapping “current times” in the same place, and Antarctica/Troll with “winter DST.”

---

DST prevalence and controversy

• Disagreement over whether “most of the world” uses DST; several point out that by both geography and population, most do not.
• Arguments for DST: more evening light when people are awake, energy savings in eras dominated by lighting, alignment with work and school patterns.
• Arguments against: health and safety impacts of clock changes, complexity for software and scheduling, and minimal modern energy benefit.
• Some regions have moved to permanent “summer time” (e.g., Turkey), while others have abolished DST entirely (e.g., Brazil, parts of the US).

---

Leap seconds and precision time

• Many consider leap seconds “trivia” for most programmers because systems smear them and hide details.
• Others report real-world impact: time-series systems, distributed industrial data, financial markets, and satellite data can be sensitive to sub‑second errors.
• There are known leap-second bugs; international bodies are moving toward abolishing leap seconds, decoupling UTC from Earth rotation.
• Libraries and languages that tried to model leap seconds precisely ended up with substantial complexity and unintuitive behavior.

---

Implementing time zones in software

• tzdb is praised for its depth and historical scholarship, but its complexity is daunting.
• The format encodes past changes and rule-based future transitions; DST rules are expressed in local time to avoid messy UTC-relative edge cases.
• Real systems hit numerous edge cases: ZIP‑to‑timezone mapping in the US, reservations and counties with different rules than surrounding states, inconsistent or last-minute DST announcements (e.g., Palestine), and legacy devices with frozen tzdata.
• Cron, recurring events, and cross-zone scheduling show how subtle bugs appear when zones change definition or DST shifts.

---

Should we abolish time zones?

• A recurring thread argues for a single global time (UTC everywhere), with local cultures just choosing different working hours.
• Supporters say this would remove timezone math, DST issues, and ambiguities when events span multiple regions.
• Opponents argue:
  • People reason about local solar time and “day” as a sleep–wake cycle; shifting the date boundary into the middle of the waking day would be confusing.
  • You still need to know when people are awake/working; time zones encode that implicitly.
  • Global time would make local narratives, holidays, and legal “effective dates” harder to reason about.
• Some suggest that even if the idea is intellectually neat, political and cultural inertia make it unrealistic.

---

Historical and cultural calendar quirks

• Commenters discuss Roman, Japanese, Hebrew, Islamic, Ethiopian, French Revolutionary, Shire (Tolkien), and ROC calendars, and systems where days start at sunset or noon or have variable-length hours.
• Many note that “normal” timekeeping is historically contingent; almost every culture has had its own peculiar system that’s awkward to encode in standard software.

Safety and Toxicity Concerns

• Many are uneasy about cooking in rosin, especially given its association with turpentine and industrial use.
• MSDS data is cited: rosin has a relatively high LD50 and is FDA-approved as a food additive, but can irritate skin, eyes, and respiratory tract; inhaling fumes is a concern.
• Some argue “not clearly toxic” is too low a safety bar for something not produced as food-grade; possible contamination (e.g., heavy metals, additives) is raised.
• Several point out the main acute risk is burns and flammability rather than chemical poisoning.

Flavor, Texture, and Mechanism

• Supporters claim rosin potatoes have uniquely intense flavor and texture due to sealing in volatiles and steaming in their own moisture.
• Skeptics note the article doesn’t clearly explain how rosin improves flavor, and question why escaping steam wouldn’t also carry off aroma.
• Debate over physics:
  • Some insist the potato interior can’t exceed ~100°C while water remains.
  • Others suggest the rosin environment may inhibit vapor escape enough to alter behavior, but evidence is unclear.
• Multiple people suggest equivalent effects might be achievable via: deep-frying whole potatoes, confit in oil/fat, sous-vide plus finishing, or foil-wrapped baking.

Historical, Cultural, and Practical Aspects

• Rosin potatoes are described as a niche Southern/SE US tradition, historically linked to pine forests and rosin/turpentine industry, possibly offering a way to cook or semi-preserve potatoes near rosin kettles.
• Some recall restaurant versions as the best potatoes they’ve had; others are uninterested due to complexity, danger, and inedible skin.

Alternative Potato Methods

• Suggested “interesting but safer” options: Syracuse salt potatoes, British-style roast potatoes (often parboiled and cooked in goose/duck fat), stock-boiled-then-roasted potatoes, microwave “sexy potatoes,” deep-fried whole potatoes, clay/salt/sand crust baking, hāngī-style pit cooking.

Terminology and Side Debates

• Long tangent on “baked” vs “roast” potatoes and general roast vs bake definitions (temperature, flame, basting, peeling, cutting), with no consensus.
• Broader reflection that many historical techniques are more about constraints and ritual than superior results; modern controlled cooking and fats are seen by several as clearly better.

What “25% of code from AI” likely means

• Many argue this mostly reflects advanced autocomplete and boilerplate generation, not autonomous feature work.
• Google’s monorepo and heavy boilerplate (protos, configs, API surface changes, tests) are seen as ideal for AI-assisted large-scale refactors and rote edits.
• Some suspect the metric may also include long‑existing automated refactoring tools and codegen now rebranded as “AI.”
• Several question how the 25% was measured (keystrokes, characters, PRs, lines?) and view it as investor‑oriented marketing.

Productivity and workflow effects

• Enthusiastic users say LLMs are a major help for:
  • Boilerplate, glue code, simple scripts, SQL, Terraform, config files.
  • Unit test scaffolding and repetitive test variants.
  • Quickly recalling APIs or patterns in unfamiliar stacks.
• Others report marginal or negative net gains: time saved typing is lost debugging subtle errors or hallucinated APIs.
• AI is often compared to “supercharged snippets” or “tab completion on steroids,” most useful when the human already understands the solution.

Code quality, complexity, and technical debt

• Strong concern that fast generation of “leaf” or trivial code will worsen bloat and tech debt, especially when repeated logic should be abstracted instead.
• Critics note LLMs confidently produce subtly wrong code; without strong tests and review this can accumulate hidden bugs.
• Some counter that humans already write lots of bad code; if AI output is always supervised and tested, it can still be a net win.

Measurement and metrics debates

• Thread questions whether Google can “accurately and meaningfully” measure software productivity, citing Goodhart’s law and metric gaming.
• Suggested aggregate metrics: time from log inspection to first change, DORA metrics, business outcomes (revenue, reliability) rather than lines of code.

Inside Google’s tooling (per commenters claiming to work there)

• Internal AI is integrated into IDEs, using Gemini adapted to the monorepo; described as high‑quality autocomplete more than autonomous coding.
• Claims of safety processes: monitoring, provenance tracking, adversarial testing, A/B experiments showing productivity gains across languages and seniority levels.

Broader concerns and sentiment

• Worries about: erosion of junior roles and career ladders, long‑term training data pollution, slowing innovation in languages/frameworks, and Google product “enshittification.”
• Others see LLMs as a real but bounded step change, similar in impact to past tools (compilers, refactoring IDEs), not imminent AGI replacing senior engineers.

Emotional response & value of botsin.space

• Many express sadness; botsin.space was a long‑running source of joy, creativity, and experimentation with bots.
• People thank the operator for hosting, appreciate the long sunset period, and discuss finding new homes for their bots.
• Some note that, despite technical usefulness, it never gained strong network effects beyond being “just some server in space.”

Running federated instances: cost, complexity, burnout

• Operating Mastodon/Lemmy instances is portrayed as hard: time, money, moderation, and legal obligations (DMCA, privacy laws, possible subpoenas).
• One commenter initially overstated subpoena frequency, later corrected to far lower rates; others note subpoenas usually don’t mean the operator is “in trouble.”
• Cloud costs (especially storage and bandwidth) and Mastodon’s Rails stack are seen as heavy; alternatives like Pleroma/Akkoma/GotoSocial/Honk are cited as lighter.
• Suggestions to charge users or move to cheaper hosts (e.g., Hetzner) meet pushback: that turns a hobby into a business with billing, fraud, tax, and support burdens.
• Some say expiring old posts or self‑hosting at home can control costs.

Federation, centralization, and user power

• Several argue federation doesn’t fully fix power imbalances: users are still at mercy of admins, and most won’t self‑host.
• Others stress the benefit of being able to choose providers, akin to email, even if most people pick a few large instances.
• Mastodon is criticized for poor portability: instance moves often lose post history, and federation quirks make migration and search frustrating.
• Defederation “wars” and UX hurdles (choosing a server, broken trending/search) are seen as barriers to mainstream adoption.

Protocol & architecture debates (ActivityPub, ATProto, Matrix, etc.)

• ActivityPub’s heaviness is attributed more to specific implementations (e.g., Mastodon/Rails) than the protocol itself; some servers run efficiently.
• Performance debates span languages (Ruby vs Rust/Scala/etc.) and architectures (push vs pull, fan‑out, media retention).
• Comparisons:
  • ActivityPub/Mastodon: many small “Twitters,” federation by servers.
  • ATProto/Bluesky: portable identities via DIDs and domains; identity decoupled from hosting; seen by some as closer to email/web’s durability model.
  • Matrix: criticized for slow federation and uneven server implementations; some users retreat to XMPP.
  • Nostr and P2P ideas are discussed but seen as hard on mobile and for large‑scale querying.

Long‑term sustainability & the “death of sites”

• Commenters expect more shutdowns in the fediverse: volunteer‑run services can’t guarantee permanence.
• Some accept this as normal internet churn; others find frequent migrations and “islands” of community discouraging.
• Archive/“deathwatch” efforts are linked as a way to track and preserve dying services.