Hacker News, Distilled

Article & Site Reception

• Several readers found the piece compelling but criticized it as overly long, repetitive, and “burying the lede” compared to traditional inverted-pyramid news.
• Others defended it as appropriate long-form journalism, especially for those emotionally invested in Sierra.
• Multiple comments complained that Vice’s ad-heavy, jumpy layout made the article difficult to read, prompting use of ad blockers, reader modes, or PDF printing.

---

Sierra’s Legacy & Nostalgia

• Strong nostalgia for Sierra’s adventure games (King’s Quest, Police Quest, Space Quest, Quest for Glory, Leisure Suit Larry, Gabriel Knight, etc.).
• Many credit these games with shaping their childhoods, teaching logic, reading, or typing.
• Some are searching for modern equivalents for their kids, with a few puzzle/logic mobile games suggested, but few true “quest” successors.
• There’s debate over Sierra’s actual game quality: early innovation and charm vs. later recognition that many designs were hostile, unfair, and eclipsed by competitors.

---

Acquisition, Fraud, and Responsibility

• Core story: Sierra was acquired via stock by a parent company later exposed for large-scale accounting fraud, which destroyed shareholder value.
• One camp emphasizes that unprecedented fraud was the main cause; another argues greed, over-ambition, and poor due diligence (accepting stock, not demanding financial transparency) were decisive.
• Disagreement over whether the CEO’s fiduciary duty effectively forced acceptance of a high-premium offer vs. whether caution (e.g., insisting on cash) was a viable alternative.
• Several commenters see this as a cautionary tale about selling out a beloved, niche-strength company in pursuit of “more, more, more.”

---

Employee Stock, Compensation, and Risk

• Many employees lost large portions of their net worth; some who borrowed against options ended up bankrupt and deeply in debt.
• This fuels skepticism about stock-based compensation: some refuse such offers; others accept but value them at zero when evaluating total comp.
• A few point out that diversification practices were weaker then and that the company was already public, so equity risk wasn’t new.

---

Auditors, Governance & Financial Crime

• Discussion of how major auditors often appear in large frauds, with perceptions ranging from incompetence to complicity.
• Some detail how regulations like Sarbanes-Oxley changed control testing but may be unevenly implemented.
• Debates over punishment for financial crimes: proposals range from jail time linearly tied to fraud dollars to valuing financial harm in “lives” using government statistical life valuations; others argue money and life are categorically different and sentences must consider context, not just amounts.

---

Market Dynamics, Greed & Company Fate

• Several see Sierra’s fate as emblematic of markets rewarding short-term extraction over long-term product and brand quality.
• Others argue Sierra was already struggling with technology shifts, design stagnation, and location-driven hiring challenges; selling was seen as the only realistic path.
• Broader theme: going public and creating an “exit” aligns leadership with financial markets and can make preserving a boutique, fan-loved company structurally difficult.

Meta: HN mechanics & Substack UX

• Some discuss HN’s “vouch” feature for dead posts and karma thresholds.
• Others complain Substack uses aggressive subscription popups and ads.

Is an AI winter coming?

• One camp expects an AI winter or at least a sharp correction: returns from scaling LLMs seem to be diminishing, public expectations are overheated, and many “AI features” feel useless.
• Others argue “AI winter” is unlikely soon: current models keep improving, more compute is coming, and AI isn’t close to saturating real-world applications.

Scaling, data limits, and capabilities

• Disagreement on whether throwing more compute at current LLM architectures still yields major gains.
• Some see recent models as plateaued around GPT‑4 level, with smaller improvements (e.g., GPT‑4o, Claude 3) vs the GPT‑3.5→GPT‑4 jump.
• One view: the main bottleneck is high‑quality human text; most providers train on similar corpora.
• Others note improvements in efficiency (smaller models matching or beating GPT‑4) and progress in image/video generation and 3D, suggesting plenty of headroom.

AGI, reasoning, and interpretability

• Several posters are skeptical that bigger LLMs will yield true reasoning or AGI, arguing LLMs are sophisticated imitators of text, not thinkers.
• Others say we don’t fully understand how LLMs work internally (mechanistic interpretability is hard), so it’s premature to declare hard limits.
• Some see future systems as LLM “orchestrators” routing tasks to specialized models rather than monolithic AGI.

OpenAI vs Big Tech & business outlook

• One side: OpenAI still leads on benchmarks and user share; GPT‑4o being free strengthens that lead. Loss of some alignment staff is seen as overblown.
• Counterpoint: competitors (e.g., Google, Anthropic, Meta, Apple) have vast distribution, capital, and can win even with slightly weaker models integrated into phones, suites, and social platforms.
• Many expect a bubble: most AI startups will fail; a small fraction will be very profitable (e.g., high‑margin niche apps).

Adoption, usefulness, and hype

• Some see huge untapped enterprise value (middleware, unstructured data, voice interfaces).
• Others note earlier ML products failed without near‑perfect accuracy and question why less reliable, hallucinating LLMs will fare better.
• Many current integrations (email “prompt builders,” generic RAG buttons) are viewed as shallow and likely to be cut.

Work, creativity, and macro framing

• Concern about elimination of creative jobs; comparisons to mass‑produced goods replacing artisans.
• Some expect a continued market for human-made art; others argue revealed preferences favor cheap, mass AI output.
• Debate over GDP as a measure of “good” and whether AI-driven gains are socially beneficial or just enrich a few and accelerate environmental harm.
• Several hope that if a bust comes, VCs, not taxpayers or retail investors, bear most of the losses.

Environmental impact and ethics

• Many criticize disposables as unnecessary single-use plastic and chemical waste, especially when film is already resource-intensive.
• Others argue that focusing on kids’ small-scale hobbies dilutes climate activism; individual choices matter but shouldn’t overshadow systemic sources like packaging and electronics waste.
• Debate over personal responsibility: some say “yes, you should feel responsible” for even small choices; others see moralizing as counterproductive and note far larger waste streams.
• A recurring suggestion: if you want film, use existing gear (old family 35mm cameras, used point‑and‑shoots) instead of new disposable shells.

Nostalgia, aesthetics, and experience

• Supporters say film “looks like memories,” forces you to slow down, and gives physical artifacts that feel more meaningful than cloud backups.
• Some treat film/Polaroid/Instax as a creative or even symbolic “protest” against digital ubiquity and infinite, low‑effort shooting.
• Skeptics counter that similar discipline and aesthetic can be achieved with digital, and that nostalgia is being commercialized.
• A few emphasize that what matters is the memory and context, not the medium itself.

Alternatives and gear

• Strong preference in the thread for:
  • Used 35mm SLRs and point‑and‑shoots over disposables.
  • Reloadable “toy” cameras or half‑frame models for the lo‑fi look with less waste.
  • Instax/Polaroid for instant prints, despite cost and rudimentary controls.
• Early‑2000s digital point‑and‑shoots and film‑simulation apps/presets mentioned as cheaper, less wasteful ways to get “imperfect” or “film‑like” images.

Cost, access, and industry trends

• Film and lab work are now expensive; labs have high rents and low volume, so per‑roll prices are high.
• Some lament that compact digital cameras are effectively gone below $1,000, making film relatively more attractive despite costs.
• Discussion of Fujifilm:
  • Strong digital profits driven by Instax and X100‑series with baked‑in “film simulations.”
  • Meanwhile, many Fuji film stocks have been discontinued or are scarce and costly.

Media and trend skepticism

• Several see the article as trend‑inflating or borderline advertorial (e.g., Fujifilm focus, buy links).
• Some question whether a few anecdotes really constitute a “comeback,” calling it lifestyle journalism for clicks.

Contract structure & equity mechanics

• Discussion centers on OpenAI’s exit paperwork tying retention of already-“vested” profit participation units (PPUs) to signing a broad release with lifetime non‑disparagement and NDA terms.
• PPUs are described as profit‑sharing instruments in a capped‑profit, private structure, not ordinary RSUs or stock. Liquidity comes only via discretionary tender offers or a distant profit event, giving the company strong leverage.
• Several commenters argue this is effectively “lending” employees equity whose realizable value depends on continued good behavior and management discretion.

Legality and enforceability debates

• Multiple commenters (including self‑identified lawyers) say conditioning already‑earned equity on new speech restrictions is likely illegal or unenforceable (lack of consideration, economic duress, conflict with NLRB guidance on severance non‑disparagement, “silenced no more” laws, etc.).
• Others reply that:
  • Employees probably agreed up front to sign a “general release” on exit, and the NDA/non‑disparagement is buried in that;
  • Clauses can be scare tactics—common, sometimes knowingly unenforceable, but effective because individuals rarely litigate.
• A later update is noted: OpenAI states it has never actually cancelled vested equity for refusing to sign, and leadership publicly calls the clause a mistake and says forms are being revised. Skepticism remains about intent and chilling effect.

Power imbalance and chilling effects

• Many emphasize that legality is secondary to practical reality: ex‑employees face huge financial stakes, costly litigation, and fear of career retaliation.
• Lifetime non‑disparagement and “NDA about the NDA” are seen as especially egregious, likely to suppress whistleblowing and corroboration around safety, data sources, or misconduct.

Reputation, ethics, and “Open” in OpenAI

• Commenters contrast these practices with OpenAI’s original non‑profit, “benefit humanity” / “open” branding and AI‑safety rhetoric, calling the shift cynical or deceptive.
• Some see this as normal big‑tech behavior; others argue that, given OpenAI’s claimed societal role and pursuit of AGI, the bar should be higher.

AI safety, AGI timelines, and governance

• Thread frequently links the gag clauses to broader worries: dissolution of the superalignment team, safety staff departures, and leadership publicly predicting near‑term AGI/ASI.
• Some are deeply alarmed that a firm claiming to “shoulder responsibility for humanity” is simultaneously muzzling insiders. Others are skeptical of AGI timelines and view “safety” as moat‑building PR.

Comparisons and practical responses

• Comparisons drawn to Wall Street, big tech, non‑competes, severance NDAs, Boeing whistleblower cases, and private‑company stock structures like SpaceX.
• Suggested responses: file NLRB complaints, consult California employment lawyers, refuse or carve up such clauses, or avoid taking opaque private “equity” as key compensation.

Feasibility of “Email 2.0” / MX2

• Many see email as already undergoing a slow, de facto “v2” via DMARC/DKIM/SPF and big-provider policies, not via a clean new spec.
• Several argue MX2-like changes could succeed only if Google/Microsoft/Apple/Yahoo push them and bias spam filters toward it; without that, it’s DOA.
• Others think large providers have weak incentives to simplify or reset email, since complexity and lock‑in benefit them.
• Skeptics compare MX2 to IPv6: technically better, but hard to deploy because the legacy system works “well enough.”

Spam, authentication, and power dynamics

• Consensus: spam is the major technical pain; everything else is secondary.
• Many doubt domain-based trust alone solves spam: domains are cheap, and reputation cartels just move from IP lists to domain lists.
• Some say deliverability is effectively a cartel; incumbents use complexity to retain control and sell “access to eyeballs.”
• Ideas floated: stricter SMTP-level rejections, richer status codes, sender‑authorization tokens, or “pickup only” or IM2000-style models; others respond with standard anti‑spam checklists showing likely failure modes.

HTML, formatting, and client behavior

• Strong disagreement about HTML:
  • Marketers and UX‑oriented commenters want standardized, richer HTML (or XML/JSON‑based formats), referencing existing but messy practice.
  • Others want HTML removed entirely or replaced with a constrained subset or lightweight markup; some propose text/markdown or similar.
• Several note decades of failed attempts to standardize HTML for email, citing conflicting interests (marketing, security, Outlook’s Word-based renderer).
• There’s skepticism that plain‑text or Markdown-only email would ever be adopted by marketers, which would hurt adoption of any new standard.

Encryption, identity, and tokens

• Some want any reboot to bake in modern E2EE (Signal-like, OMEMO-like, or better OpenPGP/S/MIME usability).
• Others argue new crypto alone doesn’t fix core problems (key distribution, usability, long-term storage vs. forward secrecy).
• Token- or capability-based models (only authorized senders can email you; revocable per-sender aliases) are discussed, but seen as complex and vulnerable to new abuse patterns.

Self‑hosting, centralization, and regulation

• Many lament that self-hosting is practically blocked by IP/domain reputation and opaque big‑provider filtering.
• Some fear MX2 would further centralize power in a few providers; others say email is already captured.
• Proposals include government‑blessed “secure email v2,” portability mandates for email addresses, or even government-run mail infrastructure.
• A minority suggests abandoning email entirely in favor of newer messaging systems; others defend email’s openness, federated nature, and Lindy‑like resilience.

Purpose and Positioning of the Tool

• Web app to build discounted cash flow (DCF) valuation models for stocks.
• Not marketed as a stock-price predictor; users are expected to input their own assumptions.
• “AI models” in the pro tier take a user’s qualitative view of a company and translate it into DCF inputs, plus export-to-Excel.

Modeling Approach and Accuracy Issues

• Defaults often just extrapolate last year’s metrics (e.g., revenue growth) five years forward.
• This yields absurd projections for extreme recent growth (e.g., NVDA, CRSP) and negative or tiny valuations for others (e.g., Boeing, Chipotle).
• Several commenters stress that DCF output is extremely sensitive to inputs; the math is trivial, assumptions are not.
• Suggestions: add bounds to auto-filled values, hide projected prices until users adjust assumptions, or initialize with breakeven / more conservative defaults, possibly show multiple contrasting models.

Reception: Enthusiasm vs Skepticism

• Positive: people like the simple UI for DCF, the ability to tweak parameters, and find it educational or entertaining.
• Skeptical: some argue that if such a tool yielded real alpha, it wouldn’t be sold cheaply; others doubt DCF’s usefulness for predicting equity prices, especially for tech and “story” stocks.
• Counterpoints note DCF is standard in fundamental analysis, especially for steady businesses, but mainly useful for understanding sensitivities, not for guaranteed outperformance.

UX, Performance, and Data

• Reports of mobile layout problems, clunky signup (inputs obscured), ticker-selection bugs, and lack of clear explanation of how projected price is calculated.
• Strong requests to view at least some models without registration.
• Site initially hit free API limits, causing errors; suggestions to cache results and move to paid tiers.
• Data comes from a specific financial API; commenters notice missing required attribution.

Feature Requests and Extensions

• Email/alert system when market price deviates materially from a user-defined fair value.
• Metric tooltips and basic “good vs bad” guidance.
• Support for regional/segment breakouts, correlation/backtesting against historical prices, and alternate valuation heuristics.

Legal and Compliance Concerns

• Some warn about potential regulatory risk (SEC/FINRA) if naive users treat outputs as advice, especially without clear disclaimers or identifiable ownership.
• Others downplay this, arguing obviously nonsensical outputs are self-disqualifying, but this is contested.

Macroeconomic & Structural Causes

• Many frame rent growth as basic supply–demand: household formation and migration outpace new housing, especially since housing starts lagged badly after the 2008 crisis and COVID.
• Inflation, higher construction costs (materials, labor, insurance, utilities), and higher interest rates all raise the cost of building and owning, which is passed on to renters.
• Others argue monetary policy and massive COVID-era stimulus “printed” money and inflated asset prices, including housing; critics counter that similar policies abroad and dollar strength complicate that story.

Supply Constraints: Zoning, NIMBY, and Local Politics

• Zoning, height limits, parking minimums, and neighborhood opposition are repeatedly cited as root causes restricting multifamily and denser housing.
• Local hearings and “weaponized bureaucracy” are described as slow, arbitrary, and dominated by older, already-housed voters resisting new development to protect home values.
• Some point out that even in places with looser land constraints (Midwest), prices have surged, suggesting broader financial and comparative-pricing dynamics.

Corporate, Investor, and Landlord Dynamics

• Debate over corporate investors: some see them as symptom (profiting from scarcity) rather than primary cause; others blame them for treating housing as a financial asset and hoarding units.
• Algorithmic rent-setting software is criticized as “collusion as a service,” allegedly enabling coordinated rent hikes.
• “Mom-and-pop” landlords report thin margins and rising costs; tenant advocates highlight increasing barriers and regulatory pushback in some regions (e.g., UK, California) that may shrink rental supply.

Regulation, Codes, and Safety

• Big subthread on building and fire codes: some argue overregulation and costly permitting add large overhead and block low-cost housing; others emphasize historical disasters and safety, viewing codes as essential.
• Several distinguish between reasonable core safety standards vs. ever-expanding, marginal requirements and bureaucratic delays.

Generational, Inequality, and Demographic Angles

• Strong sentiment that returns to capital have outpaced wages for decades, degrading living standards for younger cohorts despite technological luxuries.
• Older homeowners’ political power and reliance on housing wealth for retirement are seen as key reasons policy doesn’t prioritize lower rents.
• Immigration’s role is contested: some blame higher demand from newcomers; others argue the real issue is failure to build enough housing where people want to live.

Proposed Directions

• Common suggestions: upzone and streamline permitting, build far more housing (including social/affordable), tax vacant or speculative holdings, and support YIMBY-style local activism.
• Some advocate stricter limits or higher taxes on multiple/investor-owned properties; others warn this could further reduce rental supply.

Media framing & journalism style

• Many criticize the article’s headline (“I’m the new Oppenheimer”) as misleading clickbait: the line was uttered by an attendee, not Palantir leadership, and not clearly about AI at all.
• Others argue the headline is defensible within the piece’s subjective, “New Journalism” style and broader theme of AI warfare as analogous to nuclear weapons.
• Some find the reporter’s emotional tone (“life force sucked out”) overwrought and “virtue-signaling”; others say centering moral shock is appropriate at a conference normalizing industrialized killing.

Palantir, ethics, and the “someone worse will build it” argument

• Palantir is described as monetizing projects other tech firms avoid, charging a premium for controversial work (warfare, surveillance) and heavy consulting.
• The “if you refuse to build it, someone with fewer scruples will” rationale is sharply contested:
  • Critics compare it to dealing drugs and call it a self-serving excuse to feel righteous while doing harm.
  • Defenders invoke WWII and fear of Nazi nukes, but others note that historical dilemma was existentially different.
• Some say Palantir’s actual tech is unremarkable, IBM-like data tooling with lots of paid integration; others note it attracts strong talent and enforces a tough meritocracy.

AI targeting, agency, and civilian harm

• A key controversy: Palantir systems do not prevent “nominating” targets in civilian areas; the company defers to “end user” judgment.
• One side argues this is correct: battlefields are too dynamic for a map-based “civilian safe zone” AI, and any hard AI gatekeeping would itself be dangerous when ground intel contradicts the model.
• Others stress that these tools still shape decisions and can expand the scale and distance of killing while diffusing accountability.

Deterrence, MAD, and “war is peace”

• Strong debate over rhetoric like “scare adversaries to death”:
  • Some frame it as standard deterrence logic dating back to the Cold War.
  • Others insist this is not true MAD, which is nuclear-specific and designed to stabilize rather than terrorize.
• Broader philosophical split:
  • One camp sees war as an unfortunate but real part of human affairs; credible force and overwhelming capability are considered prerequisites for peace.
  • Another camp argues that normalizing war as legitimate policy guarantees permanent conflict, and that “war is peace”–style slogans serve militarists and the defense industry.

Propaganda, extremism, and the military‑industrial complex

• Several comments highlight the rise of openly exterminationist rhetoric (e.g., calls for “rivers of blood” in Gaza) as deliberate envelope-pushing to normalize hate and violence.
• Concerns are raised about the military‑industrial complex, “deep state” narratives, and vast US defense spending crowding out social investment.
• Some call for new security paradigms: mutual and intrinsic security, open-source “sensemaking” tools, and using technologies of abundance to reduce rather than manufacture scarcity and conflict.

Overview & Goals

• Bend is a high-level, mostly Python-syntax language targeting HVM2, which evaluates interaction nets on CPUs and GPUs.
• Main promise: “everything that can run in parallel, will,” with automatic parallelization of pure functional code, including closures and unrestricted recursion on GPUs.
• A future type layer “Kind2” is planned, analogous to TypeScript over JS, but more integrated and proof-capable.

Language Design: bend/fork and Purity

• Core construct bend is a structured recursion/loop that expands a computation tree; it is conceptually dual to fold (anamorphism).
• fork is a special built‑in tied to bend, representing recursive re‑invocation with new state; initial fork(seed) is implicit in bend x = seed.
• Variables are immutable; side effects are constrained, which aids automatic parallelization. Some find the Pythonic surface syntax helpful; others think it obscures the functional core or is confusing around bend’s implicit return semantics.

Performance & Benchmarks

• Claimed results: near‑linear scaling with cores on HVM2, especially on GPUs (e.g., large speedups vs single-thread Bend).
• Several commenters re-ran examples and found:
  • Single‑thread VM performance can be extremely slow or buggy on some platforms.
  • For a recursive sum example, optimized C/C++/Haskell/Julia can match or beat GPU Bend on commodity CPUs.
  • More complex, allocation-heavy tasks (e.g., bitonic sort) show more favorable scaling and can outperform GHC with multiple cores.
• The implementation focuses first on correctness of the parallel evaluator; codegen, tail‑call optimization, and loop lowering are acknowledged as “abysmal/early.”

Limitations & Missing Features

• Current numeric types are 24‑bit (u24/i24/f24) due to packing into 64‑bit interaction‑net nodes; 64‑bit boxed numbers and more numeric/vector types are promised “soon.”
• No native arrays yet; focus is on tree/graph structures. Max heap ~4GB of nodes.
• No tail-call optimization; some examples build huge call stacks.
• Some evaluation semantics (e.g., reducing both sides of conditionals) have correctness caveats in the paper.

Use Cases & Comparisons

• Enthusiasm for: compilers, type checkers, interpreters, evolutionary computation, signal processing, and “general GPU programming without CUDA.”
• Skeptics note: real HPC and ML workloads are tuned to arrays, caches, and BLAS-style kernels; Bend may struggle against specialized CUDA/JAX/Mojo/Futhark.
• Some see it more as a research/existence proof than production‑ready today.

Tooling, Metrics & Backends

• Discussion of how to measure parallelism (time vs “interactions/sec”). Some want FLOPS-like metrics; others argue interactions/sec is the natural unit for interaction nets.
• Suggestions for profiling tools and runtime stats to understand parallelization.
• Current backends: C, CUDA; SPIR‑V/WebGPU/OpenCL and broader GPU support are desired but not yet present.
• FFI exists internally but is not fully exposed; plans include integrating with external GPU kernels and adding textures/strings/arrays.

Community Reaction & Communication

• Many commenters are excited by unrestricted recursion and closures on GPUs and the clarity of the homepage/readme.
• Others criticize: over‑strong marketing phrases (“future of parallel computation,” “near‑ideal speedup”), lack of comparisons against strong baselines (e.g., JAX/Mojo), and confusing examples.
• There’s meta‑discussion about the harshness of early criticism vs the value of honest benchmarks and clear disclaimers; suggestions include moving performance caveats higher in docs and showing both “weak” and “strong” examples.

Overall value of GDPR

• Many EU-based developers and some business owners describe GDPR as a strong net positive:
  • Forces data minimization and better security practices.
  • Gives users rights to be informed, access, rectify, and delete their data.
  • Provides leverage to push back on unnecessary tracking and “data hoovering.”
• Others argue it is overrated or “worth nothing”:
  • See it as bureaucratic, vague, and a job-creation scheme for lawyers.
  • Claim consumer data is not substantially safer; collection methods just changed.
  • Note it’s sometimes used politically (e.g. blocking services like ChatGPT).

Cookie banners and tracking

• Large agreement that cookie/consent banners are annoying; many use blockers or automation to reject.
• Several clarify: GDPR doesn’t mention cookies; the banners mainly come from the older ePrivacy directive plus ad-industry “malicious compliance.”
• Banners are seen as:
  • A useful signal of which sites track aggressively and use dark patterns.
  • Often non-compliant (no “reject all,” opt-out hidden, illegitimate “legitimate interest”).
• Some want browser-level or API signals (DNT/GPC) to stand in for banners; others note industry refused to honor such signals and even used them for fingerprinting.

Legal clarity, implementation, and enforcement

• Confusion around “should” vs “shall” is traced to people reading the preamble rather than binding articles.
• Implementers report real ambiguity:
  • What counts as “deletion” when data exists in WALs, backups, data lakes, sketches, or hashes.
  • How to honor “right to be forgotten” while remembering opt-outs.
  • How to authenticate data-subject requests reliably.
• Some want clearer, binding guidance or certification schemes.
• Regulators are said to focus on cooperation and gradual compliance, but enforcement is seen as too slow and uneven.

Impact on businesses and small players

• Startups and non-EU sites often block EU users rather than risk non-compliance.
• Others counter that if you can’t safely handle PII, you shouldn’t collect it; many small EU firms do comply.
• Misconception: GDPR does not strictly require EU-only servers; it allows data export with adequate protections.

Broader effects

• WHOIS redaction is blamed by some on GDPR; others say it’s mostly about spam and that business contact info isn’t protected in the same way.
• Some see GDPR as making personal data “toxic waste,” intentionally raising the cost of storing it to change incentives.

Regulation, “Market Access,” and Export Concerns

• “Market access” is widely interpreted as: if NZ uses certain antibiotics/vaccines/sterilization methods, major export markets (esp. EU, organic segments) may reject the honey.
• Several comments say NZ also has strong biosecurity and conservative approval processes for agricultural pharmaceuticals.
• Some argue the premium reputation of NZ/Manuka “clean” or organic honey is economically significant, so strict rules are rational.

Antibiotics vs Vaccines vs Spores

• Clarification that American foulbrood (AFB) is bacterial; traditional treatment is antibiotics, but spores are extremely resistant and not reliably eliminated that way.
• There is confusion in the article between “vaccine” and “antibiotic”; commenters stress they are very different.
• The AFB vaccine is noted as extremely new (first approval in the US in 2023) and not a cure for already infected hives. Availability in NZ is unclear.

Burning Hives vs Alternative Sterilization

• Many say burning infected hives and equipment is standard practice worldwide because spores survive most treatments.
• Others mention alternatives used or trialed elsewhere: autoclaving, irradiation, fumigation, heat/torch treatment, soda lye, artificial-swarm methods.
• Some claim EU rules often still require destruction of equipment contacting AFB; others describe more labor‑intensive salvage regimes.
• Several commenters find the article vague or inconsistent about what sterilization options are actually legal and practical in NZ.

Scale, Management, and Spread

• People question how AFB reached thousands of hives before intervention.
• Explanation: large commercial operations frequently move frames and boxes between colonies, making traceability impossible once honey supers are mixed; if you can’t prove which boxes are clean, everything is presumed contaminated.
• Concerns are raised that high-density, large-scale operations inherently increase disease risk and may reduce genetic diversity.

Insurance, Compensation, and Incentives

• Strong debate on insurance: some say this is exactly what agricultural insurance is for; others note bee policies found online may not clearly cover disease or compulsory destruction.
• Criticism that the beekeeper may have underinsured and is now seeking socialized losses.
• Counter‑argument: without financial backstops, beekeepers have incentive to under‑report AFB, risking wider spread.
• Some suggest a government-backed fund or insurance scheme, given bees’ importance to agriculture.

Overall Reaction

• Many find the idea “cool” and conceptually fascinating, especially seeing non‑Euclidean cartoon worlds reconstructed.
• Others are disappointed by the visual quality: artifacts, haze, jittery geometry, and poor alignment when moving away from original camera views.

Perceived Use Cases

• Possible uses suggested:
  • Pre‑production / camera‑movement visualization and rough “scratch tracks” for animators.
  • Tweening aid or stereoscopic / 3D conversions of cartoons.
  • VR/AR “re-watch” experiences of old shows, walking around scenes.
  • Rapid prototyping for licensed 3D games based on 2D shows.
  • Reconstruction from historical imagery or old paintings (though some argue small datasets are better handled manually).
• Many doubt strong real-world demand, especially since 3D tools and pipelines already exist and are mature.

Technical Approach & Limits

• Recognized as using Gaussian splatting; some note it resembles an evolution of NeRFs and old tools like Photosynth.
• Praised for estimating camera positions; criticized for weak geometry and temporal coherence, and for hallucinating extra objects.
• One commenter argues Gaussian splats + spherical harmonics are a poor fit for inconsistent drawings and few viewpoints, and lack of photometric calibration worsens blending.
• Output is effectively a point‑cloud/splat representation, not a clean mesh; this limits direct use in VR/production without heavy cleanup.

Artistic & Conceptual Critiques

• Strong pushback on the premise that inconsistencies exist mainly because humans “can’t” draw consistent 3D.
• Multiple comments stress that 2D and even high‑end 3D animation deliberately distort geometry, perspective, and scale for composition and emotional effect.
• Therefore, chasing a single “true” 3D reconstruction is seen as misunderstanding the medium; often no coherent 3D ground truth exists.

Industry Context & Economics

• Many animated shows already integrate 3D (vehicles, buildings, complex shots) via toon shading; if a studio wants 3D, it typically just models it.
• Skepticism that 2D→3D reconstruction will meaningfully improve workflows or be cheaper than standard 3D asset creation.

Performance & UX

• Site criticized for heavy RAM use, many looping/autoplay videos, and poor mobile behavior (stutters, crashes, images not enlarging).

What Changing π in Doom Actually Does

• Changing π mainly breaks graphics and movement, leading to warped FOV, sliding motions, texture popping, and eventual unplayability.
• Effects come from Doom’s heavy use of radians in movement and rendering; lookup tables appear to assume a fixed π, and extreme changes cause out-of-bounds accesses and crashes.
• Some wished for smaller incremental changes to see how space “deviates” rather than jumping straight to wildly distorted values.

Is This Really “Non-Euclidean”?

• Several argue this isn’t true non-Euclidean geometry, just “messing with constants” causing glitches.
• Others accept a looser use of “non-Euclidean” for any space that violates ordinary geometric intuitions, including portal-based worlds.
• One perspective notes that Euclidean geometry plus portals breaks several Euclidean axioms, making it “non-Euclidean” in a broad sense, though not a formal geometry.

Game Engines, Portals, and Non-Euclidean Spaces

• Discussion compares Doom’s BSP-based, mostly-2.5D engine to portal-based engines (e.g., Build engine, Marathon, Descent).
• Doom originally disallowed overlapping sectors; later source ports add portals and tricks enabling room-over-room and “impossible” spaces (e.g., MyHouse.wad).
• Portal-style rendering is contrasted with true non-Euclidean manifolds; collision and spatial reasoning through portals are noted as harder than rendering.

Related Games and Media

• Many games are cited as better demonstrations of non-Euclidean or perception-bending spaces: Antichamber, HyperRogue, Hyperbolica, Superliminal, Viewfinder, Manifold Garden, and various portal-based titles.
• Classic examples from other engines include Duke Nukem 3D’s “Lunatic Fringe”, Marathon’s “5D space”, and earlier work like Descent.

Mathematics and Geometry Discussion

• Some commenters clarify that in differential geometry, π itself does not change; curvature affects circle circumference but not the derivative that defines π.
• The talk’s premise is contrasted with more rigorous non-Euclidean constructions (hyperbolic, spherical, Nil geometry, etc.).

Bugs, Constants, and Programming Lessons

• Doom famously uses a slightly wrong 10-digit approximation of π, traced to a misremembered digit.
• Broader lesson: hardcoding constants (e.g., seconds in a day) is error-prone; several real-world codebases are cited as having such typos.

Overall Reception

• Some viewers find the talk fun, playful, and a good curiosity-inducing demo.
• Others dismiss it as clickbait or trivial “garbage in, garbage out” behavior, preferring deeper exploration of actual non-Euclidean geometry.

Overall view of C type syntax

• Many find C’s declaration syntax “horrible”, especially for complex pointer/array/function types.
• Others argue it’s actually simple and elegant once understood: “declaration follows usage” and there is almost no separate “type syntax”.
• There’s broad agreement that extreme declarations (e.g. nested pointers to arrays of function pointers) are unreadable and rarely justified in real code.

How to read/write declarations

• Recommended mental model: treat a declaration as an expression centered on the identifier, then apply operators (*, [], ()) with their usual precedence.
• Several posters say the “clockwise/reverse spiral rule” is misleading or outright wrong; better to understand the actual grammar.
• A strong theme: break complex types into named pieces using typedef/aliases; if a structure is used multiple places, it deserves its own name.

Function pointers, arrays, and pointers-to-arrays

• Function pointers are widely cited as the hardest part; many admit still looking up the syntax after years.
• Common pattern: first typedef the function type, then use Func *p instead of raw int (*p)(int).
• Pointers to arrays are considered both rare and a readability smell; array-decay semantics are called one of C’s biggest design mistakes.

Comparisons to other languages

• Rust: some say Rust syntax is harder to read and has accreted complexity; others find Rust’s tree-like type syntax much clearer than C’s mixed prefix/postfix style.
• Go: its C-inspired declarations are debated; many wish it had name: type with colons, noting this would help both readability and parsing, especially with generics.
• Ada, Pascal, Zig, D, C#, Nim, Odin, C# spans, and C++ templates/aliases are mentioned as having more algebraic or wordy type syntax that often reads more clearly.

Grammar, parsing, and language design

• C’s context-sensitive grammar and lexer hack are criticized as objectively poor design, though some say practical parsers are still simple.
• Multiple variable declarations with commas (especially with pointers) are widely viewed as a historical mistake; many advocate “one variable per line”.

Tools and workarounds

• cdecl (and its web version) is repeatedly recommended for translating between C “gibberish” and English.
• Other resources and habits: online function pointer guides, writing small helper macros (_Ptr, _Array), and avoiding clever one-liners in favor of clarity.

Technical notes on “wet string” ADSL

• Follow‑up blog post is mentioned for more detailed measurements.
• Touching the wet string is said to be mostly an electrical issue: added impedance and capacitance can disrupt the signal.
• If the pair also carries analog phone service, touching during ringing can give a noticeable shock, even though normal voltages are low.

Real‑world DSL and infrastructure quality

• Multiple stories of ADSL/VDSL lines heavily affected by water ingress: speeds collapsing during rain, trunk lines with degraded paper insulation, and telcos refusing to properly fix bad pairs.
• Some users report very poor ADSL speeds even today (a few Mbps), especially in rural or older areas.
• In contrast, others describe excellent experiences where small ISPs persistently push the incumbent operator to repair lines, resulting in quick fixes.

Copper vs. fiber, DOCSIS, and G.fast

• ADSL/VDSL are described as both impressive and problematic: they squeeze high speeds out of decaying copper, but that delays fiber rollout.
• Several comments blame DOCSIS and copper‑based approaches for making incumbents complacent, while praising FTTH builds (including rural co‑ops) where they exist.
• Debate over G.fast: one claim that only a single country has a significant deployment is challenged with examples from Germany, UK, and other operators; consensus is that deployments exist but are limited and being overtaken by FTTP upgrades.
• Policy decisions in multiple countries are criticized for choosing more copper/coax in the past instead of early nationwide fiber.

Improvised and historical media

• References to 100 Mbps Ethernet over barbed wire, and early DIY telephone networks over barbed‑wire fences and party lines.
• Anecdotes about “IP over tin cans and wet string” as a student project: resonances in the medium, trade‑offs between using the resonant frequency vs. staying away from it for more complex modulation.
• Comparisons to powerline networking, carrier pigeons with storage media, and the classic “station wagon full of tapes” as high‑bandwidth but high‑latency links.

Multi‑pair wiring and bonding

• Many homes have at least two copper pairs; historically used for second lines, fax, modems, or redundancy.
• Bonded ADSL can work, but is fragile when one pair degrades or when crosstalk is high; lab tests with coiled flat cable differed significantly from real‑world uncoiled deployments.

Fallback and repair considerations

• Question raised about using scavenged copper as wartime fallback; fiber is noted as patchable but requiring specialized fusion splicers rather than simple twisting.

Project scale, cost, and “waste”

• Many see Zwentendorf as an extreme example of wasted labor and money; others argue it’s minor compared to wars, fossil fuel subsidies, or failed megaprojects like the Superconducting Super Collider.
• One estimate: 5 billion Austrian Schillings (360M EUR then; ~560M EUR today) – compared to minutes of Super Bowl ad time.
• Some suggest failed megaprojects often hide technical or corruption issues; others view this case mainly as political.

Austrian politics, referendum, and anti‑nuclear sentiment

• Thread highlights the 1978 referendum that narrowly rejected starting up the plant; it became a milestone for direct democracy and environmental activism.
• Background: a reformist government backed the plant strongly; opposition parties exploited the vote to weaken it.
• Later accidents (especially Chernobyl) solidified public opposition and ended any restart plans.
• There’s distrust in government competence and honesty around nuclear safety and war‑time nuclear risks.

Grid integration, imports, and “irony” debate

• Some call it ironic that Austria bans domestic nuclear but imports power from nearby reactors; others insist Austria “relies on the grid,” not on nuclear per se.
• Disagreement over how critical nuclear is to European grid stability: one side says nuclear outages nearly “killed” the grid; the other cites official reviews calling the system resilient.

Nuclear vs renewables, climate, and externalities

• Strong split:
  • Pro‑nuclear side: argues nuclear is unique for large, low‑carbon, dispatchable power; critiques land use and wildlife impacts of wind, waste from solar/wind, and gas dependency.
  • Anti‑nuclear / pro‑renewables side: claims nuclear is too slow, expensive, and risky; grid plus hydro/pumped storage enables renewables; base‑load “scare” is framed as FUD.
• Some argue that environmental activism that blocked nuclear worsened CO₂ outcomes; others reject that framing as extreme or unfair.

Comparisons to other nuclear projects

• Zwentendorf is compared to unused or troubled plants (Shoreham, WNP‑3, SNR‑300, Kalkar, Angra III, Olkiluoto III) as examples of political risk, cost overruns, and public fear.
• Discussion touches on breeder and sodium‑cooled designs and their unrealized potential.

Reuse of the site

• The plant is now used for tours, training, festivals, film/TV, and solar generation on its roofs and grounds, described as an eerie but impressive “time capsule.”

Geopolitics and influence

• Some speculate about Soviet/Russian or fossil‑fuel interests shaping Western anti‑nuclear opinion; others say local history (Chernobyl, gas reliance, wartime fears) is the main driver.

Scope of the Problem: Why the U.S. “Can’t Build”

• Many argue the U.S. has swung from overly aggressive top‑down building to a system with so many veto points that even modest projects stall for years.
• Permitting, environmental review, lawsuits, and overlapping jurisdictions increase time and cost, often without clearly improving outcomes.
• Others say this is a tradeoff for strong individual rights and local control; the issue is overcorrection and poor design of protections, not protections per se.

Urban Renewal, Highways, and Displacement

• Several comments highlight highways and “renewal” projects that destroyed minority and low‑income neighborhoods (e.g., Cross‑Bronx, Seattle Chinatown, SF Western Addition, Baltimore’s “Highway to Nowhere,” St. Paul’s Rondo).
• Counter‑argument: routes often followed existing major roads, cheap land, and areas where buyouts cost less; minorities were there because of prior segregation and urban decline.
• Disagreement over whether this is active “targeting” or structural economics with racist effects either way.

Cars vs. Rail and Urban Form

• Many argue highways physically split cities, entrenched car dependence, and privileged those who can afford cars.
• Strong pro‑rail camp: rail needs less space, has much higher capacity, and could have replaced many urban highways if the U.S. had invested similarly.
• Skeptics cite U.S. size, lower average density, and cultural preference for single‑family homes and cars; others counter that relevant density is in metro corridors, where rail would work well.

Housing Politics, NIMBY, and Landlords/Homeowners

• Homeowners and landlords are frequently criticized as rent‑seekers: blocking new housing, manipulating zoning, benefiting from restricted supply and property‑tax regimes (e.g., Prop 13).
• Some say that’s a small subset of owners; core driver is demand far exceeding legal supply in high‑growth metros.
• Processes meant to protect vulnerable communities are described as routinely hijacked by affluent NIMBYs to stop infill and multifamily housing.
• Low‑income housing is seen as both necessary and often poorly implemented, with “missing middle” options (small homes, townhouses, modest apartments) underbuilt.

Institutions, Capital, and Culture

• 2008 crash is cited as making builders and lenders cautious; capital and skilled labor never fully rebounded before COVID, then hit high rates.
• Others stress that in truly liberalized local markets (some Texas cities, Minneapolis, Raleigh) supply has responded to demand.
• Some frame U.S. stagnation as cultural: complacency, fragmented politics, and car‑centric ideology; others see it as an inevitable byproduct of democracy and property rights compared to more authoritarian models (e.g., China).

Overview of “100 Exercises to Learn Rust”

• Many commenters find the exercises well-structured, practical, and “no‑nonsense,” with potential to be a top recommendation for Rust onboarding.
• The material is seen as covering most core Rust features and bringing learners to somewhere between beginner and intermediate, especially if they become comfortable with lifetimes.

Comparison with Rustlings and Other Resources

• Several people compare it to Rustlings:
  • Rustlings gives files with TODOs; you fix them until tests pass.
  • Some say Rustlings assumes Rust knowledge; others report learning Rust from scratch with it, using linked book sections.
  • The new exercises are perceived as more guided from first principles for those who already know another language.
• Other suggested resources: the official Rust “get started” page, Advent of Code for practice, various Rust tips collections, flashcards based on the Rust book.

Usage, Tooling, and Minor Confusions

• The wr CLI is just a Rust binary installed via cargo, which some newer users initially confuse with a platform-specific package manager.
• Multiple people trip over the “previous exercise” wording on the first syntax page; the actual “exercise 0” lives in the GitHub repo under 01_intro/00_welcome.

Rust’s Verbosity, Ownership, and Web Development

• Some see examples (e.g., setters, .into(), ownership/borrowing) as overly verbose and unsuited to typical web/API work.
• Counterpoints:
  • Rust is not always that verbose in practice; setters/builders are not dominant patterns.
  • Verbosity buys explicit control over mutability, ownership, and memory, especially compared to C/C++.
  • For some, Rust feels expressive and even less boilerplate than Go in resource-heavy code.
• Debate over whether Rust scales well as a “high-level” language:
  • Critics prefer C#/C++ for hiding low-level concerns.
  • Supporters argue Rust is highly expressive, with strong abstractions and predictable performance.

Learning Paths and Koans-Style Exercises

• The exercises are framed as “koans”-style; similar approaches exist for other languages (Go, Python, C) via sites like Go by Example, Exercism, and Python koans.
• Some learners struggle with any koan-style material for Rust and prefer building their own projects or following project-based books.

Careers and Ecosystem

• Rust jobs are perceived as relatively scarce, especially for those without strong experience.
• Community job threads and “who’s hiring” posts are mentioned as potential avenues, but no clear consensus on job prospects.

Authenticity of the Apple badge

• Most commenters conclude the badge is fake, citing:
  • A known early employee saying the photo is not the person named on the badge.
  • The badge’s date and design inconsistencies vs. another known early badge (#8 vs. #10 issue dates, different fonts, dimensions, camera quality).
  • The lamination looks artificially “sandpapered”; the inner photo/paper appears worn while the outer plastic and badge hole show almost no use.
  • The “map” on the back is traced from a known online floor plan.
• A minority argue some specific “clues” (e.g., issue-date ordering) are weak by themselves and internet detective work can be overconfident, but they still generally accept the badge is fake overall.

Forged German Red Cross invoice & cultural tells

• German speakers dismantle the invoice:
  • Wrong date format (slashes instead of dots).
  • Currency notation “DEM” and “3.000” DM price seen as implausible and/or misused.
  • Missing legally required information, VAT, and standard tax phrases.
  • Awkward or AI-like translations (e.g., “Wir danken Ihnen für Ihr Unternehmen/Geschäft”) that don’t fit a German invoice.
  • Use of “ZIP” vs. “PLZ”, odd embossing (EU-style stars), US Letter–like paper size.
• Many see the invoice as fabricated solely to create plausible deniability (“I bought it from the Red Cross on eBay in 2001”) rather than genuine provenance.

Typewriters, printing, and typography

• Long subthread debating whether the text quality matches an IBM Selectric:
  • Some say the clean, perfectly aligned text is too good for a 1970s typewriter.
  • Others counter that electric typewriters with carbon ribbons can produce very crisp output; misalignment depends on machine condition, paper, and ribbon type.
  • Links and anecdotes show real Selectric output often has visible alignment quirks; the badge’s text may be “too perfect.”

eBay, fraud, and collector behavior

• Many say eBay is rife with fakes across categories (memorabilia, software, phones, coins); enforcement is weak.
• Some describe buyers as “not powerless” due to chargebacks, while others note policy shifts and account bans make this risky.
• Several argue the story’s real lesson is: assume high-value memorabilia is fake unless backed by robust, checkable provenance.

Certificates of authenticity and trust

• Discussion on how to trust certificates:
  • You must trust the issuing body; that trust is ultimately social, legal, and reputational (“shared hallucination/intersubjectivity”).
  • Commenters compare this to CAs, rating agencies, and currency: the cryptography or paperwork only helps once a root of trust exists.
  • Some note this creates opportunities for meta-fraud (fake certificates, or certifying other certifiers).

Ethics, law, and “victimless” claims

• One line of argument: publicly calling someone a forger risks defamation; if the evidence is strong, better to go to authorities.
• Others respond that truth is a defense and the red flags are overwhelming; sharing analysis protects future buyers.
• A few frame the sale as near “victimless”: wealthy, low-due-diligence collectors pay for feelings more than facts.
• Counterpoint: it’s still fraud; buyers deserve accurate information and platforms should remove such sellers.

Forgery craft, motivation, and escalation

• Several are impressed by the effort (badge, invoice, sob-story, multiple “collectible” listings) but see clear patterns of fakery, including other dubious items (concert tickets, “signed” letters).
• Commenters note Cunningham’s Law: this analysis may help future forgers improve, but also helps detectors know what to scrutinize.
• Some suspect the forger enjoys the craft itself; others see it as a straightforward grift that pays until caught.

Broader side threads: typewriters, focus, and tools

• Extensive tangents on:
  • Why some people still use typewriters (focus, analog constraint, aesthetics).
  • Single-purpose “digital typewriters” vs. distraction-prone computers.
  • OCR quality (good on clean type, harder on receipts) and modern handwriting recognition.
• These tie back indirectly: understanding period tools and media is key to spotting anachronisms in forgeries.

Scope of Slack’s AI Training

• Slack’s policy says it uses “Customer Data” (messages, files, usage) to train “global models” for features like search ranking, channel recommendation, autocomplete, and emoji suggestions.
• Separate docs say customer data is not used to train large language models (LLMs) for “Slack AI”; those LLMs are hosted in-house and not updated with customer data.
• Some participants call this a “nothingburger” typical of long‑standing ML features; others argue the wording is vague and full of loopholes.

Opt‑Out vs Opt‑In and Friction

• Strong consensus that using private customer data for model training should be opt‑in, not opt‑out.
• The required opt‑out mechanism (admin must email support with a specific subject line) is seen as deliberately high‑friction and easy to miss, akin to burying notice “in a locked filing cabinet.”
• Multiple people share the exact email text they used and confirm Slack’s canned confirmation response.

Privacy, Security, and Legal Concerns

• Many see this as a serious risk for companies handling sensitive or regulated data (finance, healthcare, legal, IP‑heavy startups).
• Questions raised about:
  • GDPR / “right to be forgotten” and whether models can practically “unlearn” specific users’ data.
  • Whether “data will not leak across workspaces” is technically enforceable, especially even for non‑LLM classifiers and ranking models.
  • The difference between “Slack can’t access content” vs “employees won’t,” with skepticism about the word “can’t.”
• Some expect large enterprise legal departments to push back or demand redlines; others think small and mid‑size customers will largely ignore it.

Trust, Ethics, and Business Model

• Strong sentiment that being a paying B2B customer should preclude being treated as free training data.
• Many argue this erodes trust and will drive some customers to alternatives or to self‑hosted, end‑to‑end‑encrypted tools.
• A minority defend participation as “helping build a better product,” while critics counter that users should be compensated or at least explicitly consent.

Alternatives and Responses

• Numerous suggestions to move to or consider Matrix/Element, Zulip, Mattermost, Rocket.Chat, Campfire, Nextcloud Talk, or even Signal; mixed views on Teams and Discord (both also distrusted).
• Some propose “poisoning” training data with junk; others call for regulation, boycotts, or legal challenges.