Hacker News, Distilled

Basic setup & immediate reactions

• Scenario: Everyone must press red or blue. Red pressers always survive. If >50% press blue, everyone survives; otherwise blue pressers die.
• Many argue it’s trivial: everyone “should” press red; then all live and no risk is taken.
• Others say real humans aren’t fully rational, so some will press blue regardless (altruism, confusion, trolling, children, impaired people).

Rational self-interest vs altruism

• Red advocates:
  • Red guarantees survival; blue adds risk with no extra payoff beyond what universal red already yields.
  • Pressing blue is framed as needless self-endangerment or “suicidal empathy.”
• Blue advocates:
  • Expect a non-trivial blue minority; pressing red then knowingly contributes to their deaths.
  • Prefer to risk their own life rather than help create a world where self-sacrificing or cooperative people are wiped out.
  • Some explicitly say they would not want to live in a post-red world.

Game theory & collective action

• Several compare it (often unfavorably) to the prisoner’s dilemma: here the “cooperative” choice (blue) has no better outcome than universal red, and is individually dominated.
• Others note it’s more like a collective-action problem with a harsh threshold: every extra blue vote worsens outcomes until 50% is reached; then it suddenly becomes optimal.
• Debate over Nash equilibrium: if everyone is rational and knows everyone else is rational, “all red” is the equilibrium; but many stress this ignores irrational agents and children.

Framing effects and variants

• Alternate framings: magic guns that jam if >50% shoot; jumping off a cliff if enough others jump; overloaded ship; protest under dictatorship.
• Many note that making “do nothing” the safe option (e.g., don’t pull the trigger) changes behavior drastically vs “must pick a button.”
• Color choice and political connotations (red/blue parties) are seen as intentional and biasing.

Moral responsibility & blame

• Fierce disagreement over whether red pressers are “murderers” or simply declining to join others’ risky gambles.
• Some argue blue is akin to voluntary suicide; others say red is complicity in mass killing.
• Inclusion of babies, cognitively impaired people, and animals is used both to justify blue (protect the vulnerable) and red (don’t increase their risk).

Critiques of the thought experiment

• Some call it shallow “virtue signaling” bait or a “Shari’s Scissor” designed to polarize.
• Others dislike that it ignores long-term systemic context, reduces ethics to one binary choice, or assumes away real-world social structures.

GoDaddy’s reputation and why it’s still widely used

• Many commenters describe GoDaddy as one of the worst registrars: long history of technical screwups, dark patterns, predatory expiry/auction behavior, and controversial marketing and policy stances.
• Others note it remains the largest registrar, heavily advertising to non‑technical buyers; for many small businesses it’s the first (or only) name they recognize.
• Some enterprises also use it, partly for 24/7 phone support and managed hosting, and because migrating away later feels risky or costly.
• Several people say the company has been notoriously bad for decades and are baffled that anyone still signs up.

Assessment of this specific incident

• Most see the wrongful domain transfer without documentation as gross negligence, not a rare fluke.
• Likely cause discussed: internal employee error misreading a similar domain in a support email, then processes and support that failed to detect or promptly correct it.
• Particularly criticized: support slow‑rolling, denying responsibility, or falsely claiming correct documentation existed.

Security, risk, and business impact

• Commenters stress how critical domains are: loss affects email, login/2FA flows, banking, SaaS accounts, SEO, and brand trust.
• Registrar support is called out as a major attack surface: social engineering and lax procedures can bypass technical controls like MFA.
• Using email or SMS as primary 2FA is seen as fragile because users don’t truly control phone numbers or domains.

Alternatives and domain-management best practices

• Frequently recommended registrars: Porkbun, Dynadot, Namecheap, Gandi, Name.com, Dnsimple; for enterprises: MarkMonitor, CSC, or cloud vendors (AWS Route53, Cloudflare) with paid support.
• Each alternative has some criticism (e.g., support quality, pricing changes, policy choices, or engineering mistakes).
• Strong advice to separate roles: keep registrar, DNS, hosting, and email with different providers to reduce lock‑in and blast radius.
• Use registry lock and strong internal processes for high‑value domains; avoid registrar‑provided DNS when possible.

Legal and regulatory recourse

• Some urge arbitration or lawsuits for damages; others (including lawyers) doubt it’s worth the cost given arbitration clauses and limited provable loss.
• ICANN complaints are mentioned but seen as slow and more about ownership disputes than registrar malpractice.
• Ideas raised: stronger regulation of registrars, “infrastructure building codes,” and public complaint databases to pressure providers.

Broader systemic points

• Domain pricing leaves almost no margin, so support becomes a cost center and gets minimized, especially at scale.
• Centralized DNS/registrar dependence undermines “decentralization” of services like email or federated apps.
• A recurring theme: “friends don’t let friends use GoDaddy,” and many readers report migrating all remaining domains away.

Responsibility and root cause

• Many argue the incident is primarily the operator’s fault: they architected a system where a single agent (or person) could irrevocably wipe prod and its backups.
• Repeated points: bad access management (root-like tokens, prod creds in reachable files), lack of environment isolation, and a weak backup/restore strategy (months‑old external backup, provider “backups” on same volume).
• Others note partial blame on vendors but still insist post‑incident the team must own fixes like tightening IAM, backup testing, and architecture.

AI agents, safety, and guardrails

• Strong consensus: do not give LLM agents direct write/delete access to production. At most, let them propose changes or operate in sandboxes with human approval.
• Prompts like “NEVER GUESS” or “don’t do X” are seen as non‑safety; they’re just text influencing probabilities, not hard constraints.
• Recommended mitigations: least‑privilege tokens, wrappers exposing only safe operations, soft deletes and deletion protection, time‑delayed destructive actions, off‑provider/off‑account backups (3‑2‑1), and explicit human gates for high‑blast‑radius actions.

Introspection, “confessions,” and anthropomorphism

• Many criticize treating the model’s explanation as a “confession” or evidence of intent.
• View: LLMs can generate plausible post‑hoc narratives but have no stable internal state or accountable agency; you’re just getting more text conditioned on the log.
• Some compare this to human rationalization (e.g., split‑brain experiments), but most conclude you cannot rely on an LLM’s self‑explanation for safety or root‑cause analysis.

Railway and infrastructure design

• Several commenters call Railway’s model “unsafe by design”:
  • Tokens are effectively root; no operation/resource/environment scoping for the relevant token type.
  • Deleting a volume also deletes all its “backups,” and those backups live in the same blast radius.
• Others note that confirmations belong in UIs, not APIs; APIs should instead rely on strong permissions and, optionally, delayed or protected deletes.

Authenticity, tone, and broader lessons

• Many find it ironic and off‑putting that the postmortem itself appears AI‑generated and heavily blames vendors and the agent while showing little self‑critique.
• Some suspect engagement‑bait or marketing, though others think the failure is plausible given current “vibe coding” practices.
• Broader takeaway: agents amplify existing operational weaknesses; boring, well‑understood infra and disciplined processes matter more, not less, in the “agent era.”

Technical feasibility & alternatives

• Ceramics are already common in electronics (capacitors, resistors, some PCBs); this project uses low‑fire clay instead, mainly for energy/ethical reasons.
• Some question whether air‑dry clay or 3D‑printed substrates could work; others note air‑dry clay lacks thermal robustness and that 3D‑printed plastics deform under modest heat loads.
• Clay PCBs are compared to thick‑film hybrid and ceramic boards from earlier decades, and to older phenolic-paper PCBs and breadboards.
• Suggestions include CNC machining before firing, embedding copper wire then refiring, or using wood and copper tape for minimal‑material builds.
• Several note point‑to‑point “free‑air” wiring or wire‑wrap would suffice for such a simple circuit, but others stress PCBs (even in clay) lower assembly error and skill needs.

Energy, emissions, and “sustainability”

• Debate over whether firing with wood is “better” than using grid electricity:
  • Some argue wood is CO₂‑neutral and appropriate at small scale, others stress particulate and NOx pollution and local air quality.
  • Counterpoint: the power grid and chip fabs already exist; marginal individual choices have limited impact, and electricity can be from renewables.
• Several argue that early, small‑scale experiments should not be dismissed for inefficiency; system‑level change often starts with “inefficient” prototypes.
• Skeptics question whether PCB substrates are a meaningful environmental problem compared to the rest of the electronics supply chain, and whether fragile clay boards would really reduce lifecycle impact.

Art project vs engineering / research

• Many see it explicitly as an art/educational project, not an industrial proposal, and defend the artistic, playful language and constraints.
• Others are put off by what they view as overblown or academic framing (“serious research”, “ethical hardware”), or as eco‑virtue signaling.
• Some worry such symbolic “green” projects stall after a first viral success and never tackle harder engineering challenges (vias, durability, humidity, vibration).

Feminist framing and politics

• Major subthread debates why the project labels itself “feminist hacking.”
• Some interpret this as about examining tech, labor, and material supply chains through a feminist and intersectional lens; others see the label as confusing, overbroad, or likely to provoke trolling.
• There is back‑and‑forth over definitions of feminism, its relation to capitalism, and whether this framing clarifies or obscures the project’s goals.

Reception and experiences

• Several commenters express strong enthusiasm for the creativity, “stonepunk” aesthetic, and hands‑on workshop value.
• At least one participant reports attending a workshop, enjoying the process, and finding the resulting clay artifacts pleasing and distinctive.

New GitHub Issue Popup Behavior

• Change: clicking some issue links now opens a right-side overlay panel instead of navigating to a full page, especially for cross‑repo links and certain views (Projects, dashboards).
• Behavior is inconsistent across repos/pages, adding to confusion.
• Some users initially mistook it for a bug due to nonstandard behavior.

User Reactions to the Change

• Strong negative majority:
  • Breaks basic browser expectations (click = navigate, back button semantics).
  • Overlay is cramped, distracting, and interferes with reading and navigation.
  • Compared unfavorably to Jira, GitLab, and Azure DevOps patterns users already dislike.
• A minority liked the feature:
  • Helpful for quick inspection without losing context.
  • Seen as similar to “glance”/split-view patterns and more modern list/detail UX.

GitHub Response and Performance Rationale

• A GitHub insider explains it was driven by performance:
  • Cross‑repo issue loads are significantly slower due to a heavy, Rails-rendered header and hybrid React/Rails architecture.
  • Overlay avoided reloading that header and matched behavior used in other GitHub UI surfaces.
• After backlash, GitHub rolled the change back and acknowledged it “missed the mark.”
• Some call this an embarrassing workaround for backend performance issues instead of fixing the root cause.

Broader Critique of UX & Product Management

• Many see this as part of a long trend of UX regressions, bugs, and “enshitification”:
  • GitHub allegedly ignores long‑standing bugs and community feedback.
  • Design changes are perceived as change-for-its-own-sake or to justify roles.
• Heavy criticism of modern UX:
  • Seen as driven by fads, marketing, and metrics (especially “time on site”) rather than actual task efficiency.
  • A/B testing is criticized for measuring interaction, not preference or satisfaction.
• Some defend UX as inherently hard and under‑respected; others argue generalist developers should own UX to avoid unnecessary churn.

Big-Tech Incentives, Complexity, and Workarounds

• Large companies’ UIs (GitHub, LinkedIn, Intercom, online shops) are described as fragmented, siloed, and over‑featured, with weak governance.
• Internal KPIs and revenue focus deprioritize basic usability once dominance is achieved.
• Technical side: complaints about SPA heaviness, stale UI state, and degraded performance over time.
• Users point to browser split‑view features and extensions/userscripts (e.g., Refined GitHub) as ways to reclaim control and undo unwanted UX changes.

Localization / UX Tangent

• Several comments complain about forced automatic translation of the OpenAI page and other apps, with no obvious way to switch back to English.
• People argue users should always be able to override language, independent of headers or IP-based geolocation.

Flaws in SWE-bench Verified

• OpenAI’s audit claims a large fraction of frequently-failed tasks have flawed or overly strict tests (e.g., requiring specific function names or implementations not specified in the task).
• Commenters calculate this implies roughly one in six tasks is problematic, which many see as “extraordinarily” high.
• Others note this is still acceptable if the benchmark’s main job is ranking models relative to each other, not providing an absolute quality measure.

Contamination and Goalpost Moving

• Many highlight contamination: models can reproduce exact bug fixes or problem statements, proving benchmark items are in training data.
• Some see OpenAI’s move away from SWE-bench Verified as “moving the goalposts”; others say that’s inevitable and healthy once a benchmark is saturated or compromised.
• There is skepticism about all major labs avoiding benchmark leakage, given strong marketing incentives.

Interpreting High Scores (Opus, Mythos, etc.)

• Tension arises because some models report ~90%+ scores on a benchmark OpenAI now calls unreliable.
• Possible explanations discussed: contamination, benchmark-specific optimization (“benchmaxxing”), better “guessing” of repo-specific styles, or issues with OpenAI’s own audit.
• Several note you can’t distinguish recall from reasoning at very high scores; a model at 93% vs 90% might just memorize more.

Broader Benchmark Critiques

• Benchmarks are seen as narrow, easily gamed, and quickly outdated once public; comparisons to ImageNet mislabels, database “benchmarketing,” and SPEC.
• Many advocate private or “blind” benchmarks, rotating or Olympiad-style test sets, or domain-specific hidden suites.
• Others argue static pass/fail coding tests miss what matters: integration into real workflows, long-context robustness, and agentic behavior.

Alternatives and New Directions

• Mentioned alternatives include SWE-bench Pro and follow-ons, ARC-AGI, game-based tests (Zork, StarCraft/Go), agent benchmarks, and third-party suites (e.g., coding/problem-solving evals).
• Some report stagnation in real-world coding quality despite large benchmark gains, suggesting overfitting to tests rather than genuine capability growth.

Philosophy: Nihilism, Absurdism, Meaning

• Several comments distinguish existentialism (“life has no inherent meaning; we create it”) from absurdism (searching for meaning despite its futility) and nihilism (giving up on meaning altogether).
• There’s debate over whether the OP’s frame is really existentialist, with Camus’s absurdism presented as a specific reaction to it.
• Some argue that life as a “farce” or “ride” is close to Buddhism or classic absurdism; others note that farce actually presupposes meaning and seriousness to contrast against.
• Viktor Frankl’s view is cited: meaning is often found via tasks, caring for others, or dignified suffering, even in extreme conditions.

Privilege, Suffering, and Who Gets to Say “Life Is a Farce”

• One camp argues that calling life meaningless is a luxury of the rich, secure, and idle; people in coal mines, wars, or deep precarity are too busy surviving.
• Others strongly dispute this, saying suffering and existential questioning are universal, not class-bound; you can find joy among the poor and deep misery among the wealthy.
• Some emphasize that housing/health security is hugely important, and minimizing this is unrealistic, especially in systems with expensive healthcare.
• There’s back-and-forth over whether historical or poor people pondered absurdism, with one side calling it a modern, elite preoccupation and the other pointing to religious and literary precedents.

Comedy, Coping, and Odenkirk’s Perspective

• Many see bleak, grounded worldviews as common among comedians; comedy is framed as both coping mechanism and way to expose everyday absurdity.
• Commenters link this to hard lives, depression, and “working your way up” in comedy, not just to later wealth.
• Some note that Odenkirk’s remarks in the interview come after health scares, aging, and shifting purpose (children grown, sketch no longer his domain), not from detached comfort.

Reaction to the Article and NYT Framing

• Multiple comments say the headline overstates the “meaningless farce” angle; in the transcript, the interviewer pushes that phrasing and Odenkirk mostly just agrees.
• Several argue the piece is being misread as pure nihilism when it’s more about midlife disorientation plus a commitment to “keep trying.”

Meta: AI, Style, and Online Discourse

• A subthread debates whether a particular polished comment was LLM-generated, with some calling AI-written contributions rude if not labeled, and others noting the writing style is common in human text too.

Project status & hardware support

• Asahi Linux 7.x continues to progress on Apple Silicon, with M1/M2 considered quite usable and M3 support approaching an “alpha-like” state (PCIe, keyboard/trackpad, RTC, NVMe now working).
• M4 support does not exist yet; users asking about M4 are pointed to a feature matrix that currently lists no support.
• External display support is still limited (e.g., some reports of HDMI-only on certain M2 models; no news on mainlining external display support).
• Some users report Asahi as a daily driver on M1/M2, others cite high sleep power drain and repeated post-update boot breakages, especially for headless or always-on use.

Technical achievements & quirks

• Notable win: reverse‑engineered audio chip programming to enable additional headphone sampling rates (44.1/88.2/176.4/192 kHz) beyond what macOS uses; upstreamed to mainline and backported.
• Debate over whether these extra rates are practically important, given modern resampling quality and minimal CPU cost.
• Installer is mostly Python, which some see as reasonable for an installer script.

Apple’s incentives, openness, and control

• Large subthread on why Apple doesn’t help: theories include service revenue focus, ecosystem lock‑in, fear of support burden, desire to control the entire stack, IP/NDAs, and avoiding obligations or negative press around partial openness.
• Counter‑arguments: Apple could gain hardware sales to Linux users who won’t buy into the ecosystem anyway, and could contribute minimal “documentation-level” support without offering Genius Bar help.
• Some suggest Apple tacitly benefits from Asahi (e.g., as proof Macs aren’t fully closed to regulators).

Linux vs macOS experience

• Opinions split: some say macOS “just works” with best-in-class hardware, power management, and ecosystem integration; others find Linux more polished for their workflows, especially on well-supported hardware (e.g., Framework, ThinkPads, older Intel Macs).
• Recurrent themes: Linux hardware compatibility is highly device-dependent; macOS window management and customizability frustrate some; Linux audio/graphics stacks are seen as either “finally fine” or still fragile, depending on experience.

Upstreaming, longevity, and risk

• Discussion around Asahi not being fully “mainlined”: some worry a niche reverse‑engineering effort may stall before reaching near‑complete polish.
• Others stress Asahi is aggressively upstreaming to the kernel and userland (Mesa, etc.), with Fedora Asahi Remix aiming to converge toward a vanilla distro.
• Concerns remain that Apple could technically lock the platform in future, making major distros hesitant to invest heavily.

LLMs and contribution policy

• Question about using LLMs to accelerate work: Asahi’s published policy explicitly forbids LLM‑assisted submissions, to avoid “slop” and low‑quality PR floods.
• Some agree LLMs are poorly suited to this kind of intricate reverse engineering; others note that, even if useful, token cost and review overhead are non‑trivial.

Community tone and past HN “ban”

• A visible part of the thread revisits a past anti‑HN banner on the project site reacting to trolling and perceived harassment.
• Debate centers on whether this was an understandable reaction to online abuse or an overreaction, with broader arguments about internet culture, victim‑blaming, and how much public projects should engage with hostile communities.
• For a few users, this history negatively colors their trust in or willingness to adopt the project, regardless of technical merit.

Management, incentives, and short‑termism

• Many see the core problem as managerial and ideological: optimizing for quarterly profit, headcount reduction, and “efficiency” instead of long‑term capability and resilience.
• Cutting “slack” (time for mentoring, R&D, maintenance) is argued to hollow out organizations: tacit knowledge isn’t transferred, only superficial documentation and automation remain.
• Some tie this to shareholder capitalism, private equity behavior, and concentrated corporate power; others argue this is just basic cost minimization and consumer preference for lower prices.

AI coding tools: productivity vs de‑skilling

• Supporters report concrete productivity and profitability gains when combining LLMs with strong tests and existing engineering discipline (e.g., TDD, good integration tests).
• Critics argue LLMs encourage “vibecoding”: lots of plausible but fragile code, ballooning review costs, shallow understanding, and eventual loss of deep skills.
• Several note that writing code is now faster than reviewing it; maintainers feel buried in “LLM slop” and foresee projects restricting PRs.
• There’s concern that LLM use erodes the ability to reason without them, especially among juniors and non‑experts who can’t reliably tell when the AI is wrong.

Institutional knowledge and the junior pipeline

• Strong agreement that institutional and tacit knowledge (how systems actually break, subtle domain rules) cannot be fully replaced by docs or tools.
• Reduced junior hiring plus pressure to “let AI do it” is seen as setting up a future shortage of people able to debug, refactor, or rebuild complex systems.
• Others counter that this pattern (skills atrophy, then get painfully rebuilt when needed) is historically normal and can’t be fully avoided.

Manufacturing, defense, and globalization analogies

• Some think the article overstates “the West forgot how to make things,” noting that US/EU manufacturing output remains high but more automated and specialized.
• Others point to COVID supply shocks, ammo shortages, and lost manufacturing lines as real examples of capacity and know‑how atrophy.
• Debate over NATO, Russia, China, and offshoring: some see a strategic hollowing‑out; others say outsourcing and JIT were rational under peacetime assumptions.

Meta: AI‑authored writing and “slop”

• Multiple commenters suspect the article itself is AI‑assisted, and express fatigue with recognizable LLM rhetorical patterns and filler.
• This is framed as symptomatic of a broader “slopification” of text and code: surface‑polished output, weak substance, and more work for humans to sift signal from noise.

Meta / Thread Context

• Original site was overloaded; readers share archive links.
• Several participants note that HN often debates without many people reading source material closely.

Surveillance, Record‑Keeping, and Consent

• Strong disagreement over whether digital ID systems are inherently “surveillance” or just record‑keeping.
• One side: any compulsory ID use for web access is mass surveillance; the core problem is the existence of centralized personal data, not just accuracy.
• Other side: governments and companies already keep necessary records; legality, purpose, and consent distinguish surveillance from legitimate data use.
• Dispute over whether democratic legislation counts as “consent” from citizens.

Child Protection and Age Verification

• Many argue age checks for online content are largely ineffective: kids will use VPNs, pirate sites, or borrowed credentials.
• Others argue partial effectiveness still helps parents and society set norms around children’s exposure to social media and porn.
• Concern that “protecting children” is a pretext for broader identity requirements and control.

Digital IDs: Inevitable or Dangerous?

• Some see digital IDs as inevitable and convenient, already used for taxes and government services; focus should be on legal limits (e.g., preventing “de‑personing” by revoking IDs).
• Others see a slippery slope to total surveillance and “social credit”–like outcomes: mandatory ID for banking, social media, SIMs, payments, even buying food.
• Experiences differ: some EU countries have long‑standing eID with limited scope; skeptics warn that scope can expand via regulation and corporate incentives.

Technical Design: ZKPs, Wallets, and Platform Lock‑In

• EU age‑verification docs explicitly mention zero‑knowledge proofs, but several commenters highlight practical gaps:
  • Current schemes may still allow linking via revocation and reuse patterns.
  • Hardware in phones doesn’t support advanced anonymous credentials (e.g., BBS+), so systems rely on rotating signatures tied to secure elements.
• Serious concern about reliance on Google/Apple ecosystems (e.g., Google Play Integrity, attestation on outdated Android), potentially excluding alternative OSes and violating EU competition rules.
• Debate over whether remote attestation and centralized databases create major breach/abuse risks.

State vs Corporate Power; Bots and Disinformation

• Some justify stronger identity controls to counter bots, foreign influence, and AI‑generated disinformation.
• Others argue this just shifts power from opaque platforms to governments without solving algorithms, censorship, or propaganda.
• Disagreement whether foreign “bot farms” are a major factor or an exaggerated excuse for political failures.

Resistance and Alternatives

• Proposals mentioned: web‑of‑trust systems to fight bots without tying to real‑world IDs; decentralized protocols; continuing to use cash and physical tokens.
• A minority plan to block or avoid any site requiring ID/age verification, though others note many essential services already require ID.

Scope of the Issue

• Multiple users report the Headspace app silently appearing or reappearing on their iPhones, often daily.
• Common pattern: users had installed Headspace (or its related app) in the past, deleted it, and now see it reinstalled.
• Devices affected include recent iPhones, various iOS versions (including latest stable and beta), non‑US stores, personal devices with no MDM.
• Some report the icon as “grayed out” / waiting for Wi‑Fi, suggesting a pending automatic install.

User Settings & Environment

• Many confirm:
  • Automatic app downloads and updates are turned off.
  • No Family Sharing or shared Apple ID scenario.
  • No jailbreak, no TestFlight involvement, no corporate MDM.
  • “Offload unused apps” is often disabled.
• Some had previously used an Apple Watch or macOS with companion apps, but this doesn’t explain all cases.

Hypotheses About the Cause

• Most participants lean toward an Apple/App Store/iOS bug rather than intentional behavior by the app developer.
• Leading theories:
  • App Store server‑side misconfiguration or a test configuration accidentally pushed to production, possibly starting around a specific Thursday.
  • Some interaction with restore-from-backup logic or “purchased apps that should be on device.”
  • Notification or reminder system triggering reinstall, especially if old scheduled notifications remain after deletion.
  • A bug in “automatic downloads off” or a regression introduced in a recent iOS release.
• Less likely / more speculative ideas:
  • Exploit of a hidden backdoor or mandated mechanism.
  • Carrier- or country-level manipulation (generally dismissed or unsupported by links cited).
  • iCloud Drive or document-sync artifacts forcing app reinstalls.

Reactions, Privacy, and Trust

• Many find it unsettling that an app can appear despite explicit user settings, reinforcing a sense that the device isn’t fully under user control.
• Comparisons are made to the forced U2 album rollout and to Apple’s growing tech debt and polish regressions.
• Some examine tracking SDKs in the app and raise concerns about analytics and health data, though others note such SDKs are typical.

Suggested Actions

• Call Apple Support and escalate; potentially provide sysdiagnose logs.
• Try toggling notification settings, auto-download settings, and observing behavior offline to narrow the trigger.

Amyloid Hypothesis, Fraud, and Gatekeeping

• Many comments blame decades of focus on β‑amyloid plaques (especially Abeta 42) for crowding out alternative ideas.
• Allegations of image fabrication and “foundational” fraud are seen as having distorted funding, publication, and diagnostic criteria (“plaque cartel”).
• Others argue the amyloid model is likely incomplete rather than outright wrong: plaques correlate with pathology, some mutations clearly increase amyloid, and early anti‑amyloid drugs may slow progression if used pre‑symptomatically.
• Gatekeeping by a small expert elite is blamed for grant control, peer‑review hostility to dissenting work, and “monoculture” in research agendas, even without bad intent.
• Some push back that experts are often right and that the scientific system, while flawed, isn’t fundamentally “broken.”

Inherent Difficulty and Heterogeneity

• The brain is hard to study (no easy biopsy, limited imaging), Alzheimer’s progresses slowly, and early stages are hard to detect.
• Definitive diagnosis still requires autopsy; many people have plaques without symptoms and vice versa.
• Several commenters stress Alzheimer’s is not a single disease but a heterogeneous set of age‑related neurodegenerative conditions.
• Comparisons are made to cancers and schizophrenia: many overlapping mechanisms, systems-level dysregulation rather than a single cause.

Alternative and Complementary Theories

• Viral/infectious hypotheses (notably herpesviruses, shingles) and epidemiologic links with vaccines and antivirals are discussed as promising but not definitive.
• Metabolic views frame Alzheimer’s as “type‑3 diabetes” or glucose transport failure in the brain; energy shortage, endocrine decline, and cardiovascular issues are emphasized.
• Other proposed contributors: gum disease and oral bacteria, gut microbiome, chronic inflammation, trauma and high ACE scores, sleep and glymphatic failure, lifestyle and social isolation, and genetics (APP, presenilin, APOE).
• Low‑dose lithium and various supplements are mentioned, with mixed personal reports and acknowledgment that evidence is early or unclear.

Funding, Pharma, and Regulation

• Alzheimer’s is described as relatively well funded compared to conditions like ME/CFS or long COVID; disease funding overall is seen as misaligned with burden and quality‑of‑life loss.
• Some criticize pharma as structurally biased toward long‑term treatments; others respond that companies will pursue any truly effective therapy, including cures.
• The FDA is alternately criticized as too risk‑averse in general and too permissive specifically in Alzheimer’s (approving expensive drugs with weak clinical benefit and serious side effects).

Prevention, Aging, and Outlook

• A major theme is that dementia risk is strongly age‑linked; commenters argue aging biology itself is underfunded relative to Alzheimer’s.
• Multi-factor prevention (exercise, diet, social engagement, sleep, vascular health) is highlighted; one Lancet estimate of ~45% preventable cases is cited via a commercial prevention program.
• Overall sentiment: fraud and gatekeeping likely delayed progress, but the core problem remains that Alzheimer’s reflects a deeply complex, multi‑factor failure of aging brain systems.

Role of NSF and National Science Board (NSB)

• NSF is described as a core US basic-research funder (~$9.9B/yr), supporting ~25% of federally backed academic basic research and acting as the primary federal source in math, CS, social sciences, etc.
• It funds >10k grants/year, rare-disease genomics, foundational tech (e.g., internet backbone, GPS, AI‑relevant work), and programs like SBIR for small tech businesses.
• The NSB governs NSF; staggered 6‑year terms were meant to span administrations and preserve relative independence from day‑to‑day politics.

Motives and implications of firing the NSB

• Many see this as an attempt to replace independent experts with political loyalists, eliminating a check on the administration’s control of research priorities and spending.
• Some frame it as part of a broader authoritarian strategy (including Project 2025) to neutralize independent institutions and consolidate power.
• Others argue it’s consistent with a pattern: short‑term political and personal gain at the expense of long‑term US scientific and geopolitical strength.

Comparisons to China and global competition

• Several comments say China aggressively recruits top scientists and now treats scientific input more seriously in policy than the US, with fewer “culture-war” attacks on science.
• Others push back, citing Soviet‑style political interference (Lysenkoism) as a warning and doubting that Chinese advisory bodies meaningfully challenge political leaders.
• Multiple commenters report or speculate about significant US scientific “brain drain” to China, EU, India, and elsewhere.

Impact on research, innovation, and programs

• Academics and entrepreneurs fear delayed or politicized grants, more instability, and lost projects (e.g., SBIR/Phase II examples).
• Several note that destroying institutions and trust is far easier than rebuilding; frequent partisan reversals make long‑horizon research and careers riskier.

Legal and constitutional debate

• One side argues that, as executive officers, NSB members can be removed under default presidential authority when the statute is silent.
• Others argue the founding law does not grant removal power, so mass firing is likely unlawful and emblematic of broader disregard for constitutional norms; even if courts later reverse, the damage is already done.

Broader political and cultural themes

• Many frame this as part of US decline, anti‑intellectualism, and a “cultural revolution” against experts, with voters, elites, and propaganda all blamed.
• A minority defends the move or downplays its impact, citing mistrust of “experts” post‑COVID or arguing the US system has long been failing and Trump is mainly a symptom.
• Several note visible polarization and suspect bot/troll brigading in the thread.

Overall sentiment

• Many commenters see rising public hostility toward AI, driven less by the tech itself than by how it’s being deployed and marketed.
• Some argue “the public hates AI” is overstated: outside tech/arts circles many people are indifferent or casually positive, using it like a better search tool.
• Others say people now routinely use “AI” as a pejorative (slop, fake, low quality) even when content isn’t AI-generated.

Perceived harms and externalities

• Job loss and deskilling: fear of mass layoffs, weaker worker bargaining power, and erosion of meaningful, fulfilling work. Some devs feel forced to use AI, with dissent treated as a career risk.
• Economic inequality: perception that AI concentrates gains among a small set of corporations and investors while everyday life gets harder (housing, healthcare, wages).
• Environmental and infrastructure costs: large datacenters driving up electricity prices, straining grids, water use, local emissions caps, land use, and even housing construction in some regions.
• Cultural and informational damage: explosion of low-quality “slop,” fake videos, AI spam in media, and harder-to-detect fraud and manipulation.
• Surveillance and control: concern about AI-enhanced monitoring, automated HR decisions, and “social credit”-like systems used by states and corporations.

Industry behavior and messaging

• Many blame AI leaders’ own rhetoric: loudly predicting job “bloodbaths” and existential risks while racing to sell the tech to governments and corporations.
• Surveys presented by boosters (e.g., “93% at an AI conference are excited”) are mocked as sample-biased and socially pressured.
• There’s resentment over training on copyrighted works without consent, and perceived hypocrisy when companies object to others training on their outputs.

Jobs, productivity, and UBI

• Some propose taxing AI and funding UBI or welfare expansions to share productivity gains; critics say the numbers don’t add up, especially at current AI revenue levels.
• Debate over whether UBI would just entrench a two-tier society with minimal subsistence versus genuinely replacing lost careers and status.
• Skepticism that meaningful redistribution will happen given current political and corporate incentives.

Usefulness and limits of current AI

• Many programmers say LLMs are impressive but unreliable “mediocre assistants” that require extensive verification and generate tech debt.
• Others report large personal productivity gains and don’t want to code without them.
• Some highlight genuinely beneficial uses (e.g., in medicine, research), but others call these mostly aspirational compared to current visible downsides.

Broader context and resistance

• AI backlash is seen as part of wider anger about inequality, precarious work, and unaccountable elites.
• There is debate over how to respond: stronger regulation, slowing or banning “frontier” research, non-violent mass protest, and—more controversially—whether political violence has historically been effective.

Adaptation and the gut microbiome

• Many frequent bean‑eaters report that gas drops sharply after a few weeks or months of regular consumption; some say it returns when they stop and reintroduce beans or new types.
• Several participants attribute this to microbiome shifts: more efficient, less gas‑producing FODMAP‑eating bacteria and gas‑consuming microbes outcompeting others.
• Others are skeptical, echoing the article’s doubt that “eating more beans” should reduce gas, though countered by both anecdotes and a cited study suggesting most people adapt.
• Long‑term antibiotic use is mentioned as having increased gas, with only partial recovery over years.

Cooking, soaking, and processing methods

• Classic soak‑and‑rinse is heavily discussed. Some say long soaking with multiple water changes greatly reduces gas; others cite the article’s test that found little effect and question its methodology.
• Variants include: soaking in boiling water, soaking with baking soda, or using alkaline solutions (bicarbonate, lime), typically discarding the liquid afterward. Some report big gas reduction; others mainly note texture changes.
• Sprouting/germination is proposed as an untested but “obvious” approach; several describe simple home sprouting methods.
• Peeling bean skins is cited from African traditions as effective but labor‑intensive.

Additives, spices, and fermentation

• Asafoetida (hing), black mustard, fenugreek, cumin, and caraway are frequently mentioned folk remedies, mainly supported by anecdote.
• Fermentation approaches (lacto‑fermentation, koji, tempeh, miso/Aspergillus) are claimed to help by breaking down indigestible components, with trade‑offs in flavor and use.
• Acidic additions (apple cider vinegar after meals, heating onions with lime juice) are reported by some to reduce symptoms.

Enzyme supplements and other interventions

• Commercial alpha‑galactosidase and digestive enzyme blends (e.g., “Bean‑zyme” type products) are repeatedly cited as very effective for some.
• An unusual anecdote describes years of gas resolving after a targeted abdominal massage; others mention yoga‑like organ‑massaging poses but treat the story with curiosity and skepticism.

Individual variability and perception

• Several note that many people have no bean‑related gas at all, and that cultural jokes and rhymes may exaggerate beans’ typical effects.
• Others experience severe pain and bloating, sometimes more from inulin‑rich foods, lentils, stone fruits, or dairy than from beans themselves.
• A technical subthread discusses hydrogen vs. methane production, methanogenic archaea, and the possibility that “gentler” gas release or increased methane could reduce noticeable volume, though mechanisms remain unclear.

Attitudes and humor

• Some argue bean gas is inevitable and should be accepted; others view mitigation as a legitimate “hack.”
• Fart humor, workplace norms, and cultural attitudes around flatulence appear throughout the thread.

Ground-Source Heating/Cooling & District Systems

• Commenters describe shallow geothermal mainly as an HVAC efficiency play, not power generation: e.g., pumping ~64°F water from depth for cooling, then reinjecting it warmer.
• Idea: neighborhood-scale “geo-utility” or district energy networks; some real implementations exist (e.g., a MA town network, a Texas development, Nordic schools, Finnish single-family homes, parts of Poland/UK).
• Others argue district heating/chilled water is usually uneconomical for dispersed single-family homes, but can work in new developments or in regions with legacy district systems.

Air vs Ground-Source Heat Pumps

• Ground-source is praised for higher efficiency (COP) and using stable ground temperatures, but high upfront costs (especially vertical drilling) and space/septic conflicts limit adoption.
• Several note air-source heat pumps have improved dramatically and are now viable down to very low outdoor temperatures, with better cost-effectiveness in many markets.
• Disagreement over climates where heat pumps “don’t work”; others say they work well even in mild climates like San Francisco and in cold regions.

Enhanced Geothermal Systems (EGS) & Fervo

• Article’s “breakthrough” is identified as EGS: using fracking-style horizontal drilling and stimulation to create reservoirs in hot rock where natural hydrothermal systems don’t exist.
• Some see this as a significant extension of geothermal’s geographic potential; others call it marketing hype, noting EGS concepts and companies have existed for ~15–20 years.
• Fervo’s approach is discussed: engineered reservoirs in granitic rock, plug-and-perf fracturing, improved drill bits; pilot-scale power exists, but long-term economics, flow rates, and thermal decline are still open questions.
• Skepticism about hype is reinforced by mentions of earlier uneconomic geothermal plants and Wikipedia COI warnings (about the article page, not necessarily the company).

Economics, Drilling, and Water

• Drilling remains the dominant cost and technical risk; super-deep/plasma/microwave drilling is viewed as early-stage.
• Some note geothermal LCOE can be favorable versus nuclear but depends heavily on geology, drilling costs, and scaling.
• Modern EGS is usually closed-loop on the geothermal side; water use is mostly for initial stimulation, though cooling systems (wet vs dry) can still consume water.

Role in the Energy Mix

• 150 GW is framed as ~10% of US generation: significant but not dominant.
• Several envision a system where cheap solar/wind plus batteries supply most energy, with a smaller share from 24/7 sources like geothermal and nuclear.
• General sentiment: geothermal is promising but unlikely to be a silver bullet; “we need all types.”

Model capabilities and tiers

• Multiple commenters note that the free ChatGPT tier (gpt‑5.4‑mini) feels heavily constrained and more hallucination‑prone, while paid “thinking” models (e.g., 5.5 Pro) can spend 20–80 minutes on a response and are qualitatively different products.
• Longer “thinking” is described as expensive inference-time compute, hence gated behind higher‑priced plans and API usage.
• Some report that Gemini reaches ~90–95% of ChatGPT Pro’s solution quality with far fewer tokens and much less thinking time.

Nature of the proof and verification

• The raw proof from the model was described (in the article and thread) as messy and hard to parse; experts were needed to extract and shorten the core idea.
• Several participants stress that formal verification (e.g., in Lean) is much harder than writing an informal proof, and that non‑experts cannot reliably check either the English proof or its formalization.
• Others point out that human papers also require significant expert time to verify and often have opaque notation and gaps.

Intelligence, creativity, and “just text prediction”

• One camp argues that solving a previously open Erdős problem with a novel technique is strong evidence of real intelligence and creativity, even if produced by a statistical next‑token model.
• Another camp insists these models remain “just text generators,” comparing them to calculators or automated theorem provers doing large search, and invoking shifting definitions of intelligence.
• There is debate over whether applying a known formula in a new context counts as creativity; many argue that cross‑domain recombination is precisely what a lot of human “creative” work is.

Brute force, reasoning, and prompts

• Some attribute success to a kind of powerful “brute force educated guessing” over a huge learned corpus; others reject the brute‑force characterization, emphasizing visible hypothesis‑driven reasoning.
• Prompt phrasing (e.g., “don’t search the internet,” “non‑trivial, creative and novel proofs”) is suspected to significantly shape the model’s search behavior; prompt sensitivity is widely acknowledged.
• Because outputs are stochastic, it’s unclear whether earlier models could have solved the problem but didn’t under prior prompts.

Cost, access, and democratization

• Commenters worry about the token and energy cost of such long runs and whether only well‑funded actors will benefit as models scale.
• Others counter that going from needing a top specialist to needing a motivated amateur plus a $100–$200/month tool is already a major democratization, though global affordability remains contentious.

Impact on mathematics and tooling

• Many see LLMs as promising “weird collaborators” that can propose unconventional approaches and cross‑apply techniques between subfields.
• There is interest in math‑specific “harnesses” that combine LLMs with tools like Python, Sage, and Lean, and in systematically running new models against curated lists of unsolved “dry lab” problems.
• Some caution that prior “AI solved an Erdős problem” claims have later reduced to rediscoveries, so formal and community verification remains essential.

CPU options: Intel vs AMD

• Questions around Intel vs AMD focus on battery life, performance (coding/compiling, local LLMs), and memory architecture.
• Some claim Intel’s new Panther Lake chips have notably better battery life, CPU/GPU performance, and integrated graphics than current AMD mobile parts, plus faster memory with LPCAMM2 (but no unified memory like Apple).
• Others note AMD’s latest gen is mostly a sideways step versus the strong 7000 series, partly blamed on NPU die area.
• Conflicting claims about AMD: one comment says no RAM suspend and no USB4; several others correct this, stating current AMD boards support suspend and USB4 (40 Gbps on 2 of 4 ports).
• On AMD, some mention the lack of low‑power cores making it more power‑hungry; on Intel, all four ports support USB4/Thunderbolt 4 and are interchangeable.

Modularity, upgrades & long‑term ownership

• Many see Framework’s modularity as the main value: screens, batteries, input covers, hinges, Wi‑Fi, SSDs, and mainboards can be swapped across generations.
• Several users describe “Ship of Theseus” laptops largely rebuilt from newer parts, highlighting real backward compatibility.
• Debate over economics:
  • Pro‑Framework: pay a premium once; later just buy upgrade kits, cheaper and more sustainable than full replacements.
  • Skeptical: used or on‑sale laptops can cost less than a Framework mainboard; selling your old Mac/PC and buying new can be cheaper.

Pricing and comparisons (Windows laptops & MacBook Pro)

• Many view Framework as overpriced versus heavily discounted gaming/Windows laptops and even MacBook Pros at list price.
• Others counter that Apple’s scale and service‑driven model suppress Mac hardware prices, while Framework must pass current RAM/SSD costs directly through.
• Some argue that once you need higher RAM/SSD, Framework can be cheaper than a similarly specced Mac.
• The lack of economies of scale and modular/repairable design are cited as inherent cost drivers.

Linux, OSes & openness

• Ubuntu preorders reportedly outpace Windows, reinforcing the “Linux‑first” positioning.
• Users discuss running a wide range of OSes (various Linux distros, BSDs, potentially others); storage expansion cards enable easy dual‑boot/“Windows on a card” setups.
• Some criticize Framework for not providing open firmware/coreboot yet, saying this conflicts with its “laptops you own” branding and prevents certifications like QubesOS’s. Others reply that coreboot is non‑trivial on modern platforms and a stretch goal.

Design, Mac comparisons & target audience

• Strong debate over the MacBook‑like look:
  • Critics dislike “clone” aesthetics and would prefer ThinkPad‑ or Latitude‑style designs and more visual diversity.
  • Supporters argue MacBooks set the bar for build, screen, trackpad, and acoustics, and many people explicitly want “Mac‑grade hardware that runs Linux/Windows.”
• Some prioritize non‑gloss, matte displays and note that alternative OEMs frequently change finish or drop desired configs.

Hardware details & quality

• Expansion card mechanism on older Framework 13s divides opinion: some find cards too loose (pull out with snug USB cables), others say they’re extremely tight and require tools. The Pro reportedly redesigns the latch, with questions about lock‑out for accidental removal.
• Concerns exist about Framework quality control relative to big OEMs; others report long‑term positive experiences with multiple generational upgrades.
• Interest in ARM/RISC‑V mainboards and ECC support is noted, but current third‑party non‑x86 boards reportedly have poor power characteristics for laptops.

Preorders, shortages & audience fit

• Some are wary of preordering before independent reviews; others cite hardware shortages, high component prices, and trust built over several generations as reasons to commit early.
• Overall, Framework is seen as ideal for users who: care about Linux support, repairability, matte screens, and modularity; accept higher up‑front cost and some rough edges over cheaper, less repairable mass‑market laptops.

Iran’s Deterrence and Regional Power Balance

• Several commenters argue Iran/IRGC has achieved real deterrence: it survived US strikes, hit multiple US bases and allies, and showed willingness to absorb major damage to its own infrastructure.
• Others counter that sanctions and blockade have gutted IRGC revenue (claims of ~90% income loss), making its position brittle and dependent on mercenaries.
• There is sharp disagreement over who is “blockading” the Strait of Hormuz, whether the US stance is militarily sustainable, and whether the global economy can tolerate a major oil shortfall for long.
• Some see Iran’s precise, low‑casualty strikes on US infrastructure as a deliberate signal: “we can hit what we want, when we want.”

US Military Vulnerabilities and Strategy

• Multiple comments say Iran’s ability to damage US forward bases exposes a core weakness of the US expeditionary model: reliance on vulnerable land bases for sortie generation.
• Claims that US used a very large share of its high‑end standoff weapons and interceptors; continuing at that burn rate is seen as strategically unsustainable.
• Damage to key enablers (AWACS, radars, tankers, runways) and pushing carriers further out supposedly breaks the sortie math and forces heavier use of expensive munitions.
• Some argue the US response devolved into threats against civilian infrastructure because the counter‑force air campaign hit diminishing returns.

Broader Geopolitics: Gulf Allies, Taiwan, and China

• One thread says Gulf states were attacked mainly because they host US bases, and notes that US behavior suggests Israel is its only truly protected ally.
• Several commenters claim US performance will alarm Japan, South Korea, the Philippines, and Australia, and predict no hot war over Taiwan—possibly a gradual Hong Kong–style accommodation.
• Others reject this: Hong Kong’s fate has hardened Taiwanese resistance to unification, and Taiwan is described as a capable “middle power.”
• A long subthread argues China’s missile reach, semi‑supply‑chain leverage, and potential to strike US energy infrastructure give it far more escalation options than Iran, undermining any US blockade strategy in East Asia.

US Policy, Transparency, and Legitimacy

• Many are skeptical of official narratives: references to downplayed casualties, mishandled incident explanations, and suppression of commercial satellite imagery of damaged bases.
• The article’s description of extensive, costly damage with few US deaths is seen by some as contradicting claims of an unqualified US success.
• Commenters tie current events to a longer history of US support for Saddam against Iran, coups, and past atrocities, arguing that US moral standing in the region is weak.

Regime Change, Responsibility, and Ethics

• Strategic debate: some say a blockade plus covert ops and arming civilians would have been smarter than high‑profile decapitation strikes; others say the US should never have gone to war at all.
• Several insist that overthrowing Iran’s regime by force would require large‑scale boots on the ground and would resemble or exceed Iraq in scale and difficulty.
• There is disagreement over whether Iran’s conventional military (Artesh) could smoothly replace the IRGC, and whether IRGC is primarily ideological or profit‑driven.
• Strong dispute over whether the US bears responsibility to “fix” Iran, versus having already done enough damage and lacking legitimacy or capability to reshape the country.

Information Warfare, Drones, and Bases

• Observations that quadcopter‑style drones are operating with near impunity around regional bases, with speculation about fiber‑optic control and gaps in current air defense.
• Some argue US bases were treated as expendable and largely emptied before attacks, trading infrastructure for lives.
• A side discussion contrasts Iran’s deeply buried military infrastructure with US surface bases, debating feasibility and scalability of underground basing.

Media Access and Archiving

• Several users complain about paywalls and dynamic site blocks; there is discussion on ethically archiving paywalled journalism for long‑term access versus real‑time piracy that might hurt news outlets.

Cloud vs Local Models and Cost

• Debate over paying for cloud AI subscriptions vs investing in local hardware.
• Some see $20/month coding assistants as trivial and multipurpose; others find $200/month tiers prohibitive.
• Local setups (Mac Studio, high‑end AMD/RTX PCs) praised for control and avoiding “nerfed” cloud services, but criticized as costly, noisy, and slower than top hosted models.
• Open models via local runtimes or hosted gateways (e.g., OpenRouter) seen by some as good-enough, by others as “glorified autocomplete” unless run on very strong hardware.

Use Cases: Reviving and Building Personal Projects

• Many report resurrecting old or abandoned projects: games, note apps, editors, networking setups, home servers, old mods, and internal tools.
• Coding assistants help re-understand old code, modernize dependencies, plan features, and generate missing UI, tests, and glue code.
• Game dev examples include Godot, Bevy, and custom engines, with LLMs assisting in procedural content, simulation logic, and tools rather than full game creation.
• Numerous “personal tools” built that would be uneconomical to buy or sell: niche GUI utilities, search libraries, admin panels, save editors, automations.

Perceived Benefits

• Major gains cited in:
  • Getting past “half-finished wall” and re-entry overhead.
  • Offloading boilerplate, refactors, and plumbing to focus on design and domain logic.
  • Exploring alternative architectures cheaply via mass refactoring.
  • Making non-coding hobbies more enjoyable by quickly building supporting software.

Limitations and Frictions

• Local models often too slow/weak for complex agentic workflows; hardware requirements high.
• Models struggle with visual/scene assets, engine-specific formats, and staying consistent across sessions.
• Productive use typically requires careful setup: CI, isolation, reproducible commands, headless modes, and explicit testing hooks.

Skepticism, Quality, and “Slop”

• Critics see a surge of shallow, low-quality “AI slop,” little learning, and reduced personal pride or attachment.
• Some argue effort used to filter out bad ideas; LLMs now make it too easy to pursue every whim.
• Concerns about astroturfing/marketing: posts that read like corporate propaganda draw suspicion.
• Others counter that for personal tools, maintainability and elegance matter less than solving one’s own problem quickly.

Skills, Learning, and Future Outlook

• Disagreement on “deskilling”: some fear losing problem-solving practice; others argue skills can be (re)learned and that AI simply shifts effort toward architecture, scoping, and integration.
• Several expect a future where bespoke “self-source” apps are routinely generated to spec, flooding the world with ultra-niche software and lowering the economic value of such projects.