Hacker News, Distilled

Nostalgia for the 1990s

• Many see the late ’80s–’90s as a peak era: strong job markets, relatively cheap housing, perceived safety, friendlier social norms, and a sense of excitement around technology.
• Cultural life is remembered as rich: influential music across rock, electronic, hip‑hop, “monster‑of‑the‑week” TV like The X‑Files, and abundant subcultures.
• Several recall optimism: belief that tech would improve lives, that democracy was stable, and that each generation would be better off.

Counterpoints & Who It Was Good For

• Others stress this “golden age” was highly conditional: better for straight, white, middle‑class people in stable countries.
• Anti‑gay bigotry, AIDS stigma, and opposition to interracial marriage were still common.
• Some countries had severe economic crises in the ’90s; for them the 2000s were better.
• Homelessness in the US was significant and may have peaked in that decade.

Politics, Economy, and Decline Narratives

• Some trace today’s problems to policy shifts from the ’90s onward: deregulation, tax cuts, war spending, and rising inequality.
• Others argue “third way” centrism moved politics rightward and normalized low taxes and high inequality.
• There is strong pessimism about post‑9/11 politics, Trump‑era polarization, and a sense that younger generations are worse off.

Technology, Internet, and Media

• The early commercial internet is remembered as open, quirky, and less surveilled; search “just worked” and wasn’t ad‑driven.
• Today’s web is seen as consolidated into a few platforms focused on ads and data extraction.
• Some note that despite vastly better tools and media availability now, people feel more nihilistic and overwhelmed.

The X-Files: Impact, Reboots, and Themes

• Widely praised as “right show at the right time,” blending government conspiracy, rural weirdness, and pre‑smartphone tech.
• Debate over whether it normalized conspiratorial thinking or simply reflected preexisting fringe culture.
• Strong skepticism that modern reboots can work, given today’s “post‑truth” environment and different fears; some suggest a new show would need to foreground corporate power and AI rather than secret-government alone.

Living “Like It’s the 90s” Today

• Some argue you can partially recreate the vibe (physical media, fewer screens, more local socializing).
• Others say that without the broader social context—pre‑social‑media norms, economic conditions, pre‑9/11 politics—it’s ultimately just nostalgia role‑play.

Legal responsibility and qualified immunity

• Many see this as systemic harassment enabled by tech and bad police practices.
• Several argue victims have little recourse due to qualified immunity; others note some states have limited it, so change is possible.
• Debate over who is primarily at fault: police entering/using bad data vs. Flock enabling and scaling misuse.

Role and culpability of Flock

• Some call for Flock to be shut down and its leadership criminally liable for facilitating rights violations.
• Others say Flock is unethical but not clearly illegal, and that the real abuse stems from government actors.
• Counterargument: Flock is acting as an “agent of the state” and should share constitutional burdens and liability.
• There’s concern that privatization launders responsibility between vendor and police.

AI and automation in law enforcement

• Strong faction: AI should never be used in law enforcement; any AI involvement should spoil a case and exonerate defendants.
• Others argue the core problem is unaccountable, lazy, or abusive policing; AI is just another tool being misused.
• Some foresee AI adoption as inevitable and focus on oversight, accountability, and process design instead of bans.

License plate design and data practices

• Root technical issue: conflating “O” and “0” (and other homoglyphs) in plate assignments and warrant databases.
• Some say listing all variants is “insane”; others say it’s rational given asymmetric officer-safety concerns.
• Automation via ALPRs turns a previously rare edge case into a large-scale recurring problem.

Due process, lists, and civil liberties

• The target’s inability to discover or correct the erroneous warrant hit is described as Kafkaesque.
• Calls for “anti-Kafka” laws: right to know why you’re on any kind of list, and a clear appeal/removal process.
• Concerns that third-party data (Flock, other surveillance vendors) is used to evade Fourth Amendment protections and enable broad, long-term tracking.

Meta and future trajectory

• Some see this as a glimpse of a surveillance-dystopia future (Flock, Palantir, Ring, social-credit-like scoring).
• Others note apparent “Flock apologists” and suspect astroturfing in online discussions.

Scope of AI Data Center Water Use

• Many commenters argue AI/data centers use far less water than popular claims (e.g., “10,000 gallons per photo”) suggest.
• Several compare estimated AI/data center use (tens of billions of gallons/year) to:
  • US residential outdoor watering (~9B gallons/day)
  • Golf courses (~500B gallons/year)
  • Agriculture (orders of magnitude higher, esp. alfalfa, nuts, corn for ethanol).
• Others counter that “billions of gallons” is still significant, especially where water is scarce.

Local vs Global Impacts and Siting

• Strong theme: global totals can be small while local impacts are severe.
• Examples raised: central Arizona alfalfa, California/Colorado River depletion, Loudoun County (VA), Mexican regions where data center water competes with farming.
• Some conclude siting should favor water-rich regions (Great Lakes, wetter climates) and/or graywater use.

Cooling Technologies and Tradeoffs

• Clarifications around:
  • Open-loop evaporative cooling (high water use, cheaper, more common where water is cheap).
  • Closed-loop and immersion: less direct water, more electricity; still often dump heat via cooling towers.
• Tradeoff highlighted: saving water usually increases power consumption.

Water Quality, Pollution, and Aquifers

• Multiple comments stress that it’s not just volume but:
  • Use of potable vs non-potable water.
  • Evaporation from stressed aquifers that recharge slowly.
  • Discharge containing biocides, corrosion inhibitors, and heavy metals.
• Some link to broader groundwater depletion and land subsidence concerns.

Pricing, Water Rights, and Policy

• Frequent argument: the core problem is badly designed water rights and underpriced industrial water.
• Suggestions:
  • Tiered or higher pricing for large users instead of outright bans.
  • Allow/encourage graywater and wastewater reuse.
  • Reform “use it or lose it” agricultural rights that incentivize waste.

Trust, Transparency, and Use of AI as Source

• Skepticism that hyperscalers hide water data (lawsuits, NDAs) undermines trust.
• Several criticize the article and other defenses for:
  • Relying on LLM estimates as “citations.”
  • Using favorable comparisons (e.g., beer) that embed value judgments.

Broader AI Debates Bleeding In

• Thread frequently veers into:
  • Wealth inequality and job loss fears vs claims AI boosts productivity and access to services.
  • Claims that water focus is a proxy for deeper opposition to AI or a “morality sink” that’s easy to message.
  • Disagreement over whether environmental critiques are good-faith or mainly anti-AI rhetoric.

Mechanism of the “gay jailbreak”

• Core idea: don’t directly ask for disallowed content (e.g., drug synthesis); instead, ask how a gay person (or similar identity) would describe it, often in a flamboyant, role-played style.
• Many commenters see this as a variant of older “role play” / “grandma” jailbreaks that reframe the request rather than a fundamentally new technique.
• Some note it’s essentially exploiting that guardrails are largely linguistic pattern-matching and can be sidestepped by indirection or obfuscation.

Is LGBTQ context actually special?

• One view: models are extra-eager to be supportive of protected groups, so refusing such a request feels “risky” to the alignment layer; political over‑correction becomes an attack surface.
• Counterview: identity is incidental; the real bypass comes from role-play framing, emotional backstory, and language markers (e.g., “Gen‑Z” slang).
• People report similar effects using other identities (e.g., Christians, senior engineers, “Karen” complainant), suggesting it’s not uniquely LGBTQ-driven.
• An experiment on an open model attributed the effect to language choice and role-play, not queer identity per se.

Effectiveness and current status

• Several commenters say they cannot reproduce the jailbreak on current major models; the original prompts are ~10 months old and likely patched.
• Others report partial or “lite” versions still working in some systems, especially when combined with obfuscation (encoding, foreign languages, poetry, etc.).
• Some argue the returned “dangerous” content was not very deep or practically useful; others stress that any bypass of stated safety policies is the relevant failure.

Guardrails, classifiers, and safety design

• Broad consensus that you cannot get a “purely safe” LLM from training data alone; harmful outputs can be derived from benign knowledge (chemistry, history, anatomy).
• Thus, many say you need separate safety layers (classifiers, keyword heuristics, secondary LLMs) on both inputs and outputs.
• Mention of real-world practices: fine‑tuned BERT‑style classifiers, content‑safety APIs, and runtime flags like “Trusted Access for Cyber.”
• Some note that stronger alignment can paradoxically widen the attack surface when attackers learn to “weaponize” the very norms (e.g., inclusivity) used for safety.

Broader themes and reactions

• Mix of amusement (“high‑tech social engineering,” “Bugs Bunny mindset”) and skepticism (“lazy and old” jailbreak, overinterpreted theory).
• Comparisons to human social engineering and legal questions around impersonation (e.g., claiming to be FBI to get restricted info from a model).
• Ongoing tension: some want fully uncensored models; others emphasize models are marketed to the general public (including children), making guardrails inevitable.

Perceived Quality and Use Cases

• Many report V4 Pro (and Flash) as “good enough” or comparable to Claude Opus 4.6 / GPT‑5.4 / Sonnet‑class for day‑to‑day coding, refactors, and prototyping.
• Some users successfully ran deep analyses of medium–large TypeScript or backend codebases for a few cents, saying they’d previously spent dollars to tens of dollars on frontier models for similar work.
• Others found V4 Pro clearly below GPT‑5.5 / Opus‑max for hard planning, design critique, or complex UX work, calling it more “Sonnet‑level” than frontier.
• Flash is often preferred for cheap, “stupid or speculative” tasks (summaries, stylistic cleanup, brute‑force refactors).

Pricing, Subsidies, and Token Efficiency

• Pricing is a major attraction: reports of ~150–200M tokens per $100 at list price, or hundreds of millions for a few dollars under current discounts.
• Several note the 75% promotional discount on V4 Pro; others emphasize it’s still cheap at full price, especially versus Anthropic/OpenAI subscriptions.
• Skeptics point out that V4 Pro and K2.6 often use many more “thinking” tokens than frontier models, so effective cost per solved task may be closer than raw token prices suggest.

Tooling, Harnesses, and Providers

• Used through DeepSeek’s own API, OpenRouter, Ollama Cloud, Bedrock, Azure AI, Tinfoil (enclave), Claude Code-compatible harnesses, VS Code integrations, pi.dev, opencode, and custom agents.
• Some harnesses mis-handle tool calling or expose raw “thinking traces,” which can look like the model is having a meltdown but are likely internal self‑corrections.
• High cache hit rates on the official API (claimed >99% in long sessions) can drastically reduce cost when working in one codebase.

Privacy, Data Use, and Open-Weights Tradeoffs

• Strong concern that the official DeepSeek API does not guarantee non‑training on user data, even for paying users.
• Some are fine with this as a “trade” for open weights and low prices; others prefer US/EU hosts with stricter legal regimes or confidential‑computing services.
• Open‑weights are praised for: (a) enabling alternative providers that “don’t phone home,” (b) allowing local/self‑hosted use, and (c) reducing lock‑in risk.

Alignment, Censorship, and Safety Filters

• Several contrast DeepSeek’s relative permissiveness (e.g., reverse engineering assistance) with OpenAI/Anthropic, where users report refusals and even account warnings.
• Others note censorship exists on all sides (e.g., Tiananmen‑type queries), but open weights allow “uncensoring” or moving to other hosts.

Technical and Hardware Considerations

• V4 Pro reportedly uses new attention/KV‑cache tricks (HCA, mCH) for long context, drastically cutting FLOPs and cache needs relative to earlier DeepSeek versions.
• Flash can run locally with large RAM / multi‑GPU setups; one user runs 1M‑token context fully in GPU memory and sees >2× speedup vs V3.2.
• Running Pro on CPU alone is seen as possible but very slow; better suited to unattended/overnight runs.

Benchmarks, Competition, and Geopolitics

• The “pelican on a bike” test is widely dismissed as overfitted and uninformative; calls for more meaningful evals.
• Mixed views on how V4 compares to Kimi K2.6 and GLM‑5.1: some say Kimi/GLM outperform for coding; others see V4/Flash winning on specific tasks (e.g., spatial reasoning).
• Discussion touches on training sources (alleged distillation from US models), Nvidia embargoes, and Chinese hardware (Huawei) as strategic backdrop, but details remain unclear.

Nature of Spotify’s “Verified” Badges

• Many see the badges as anti-scam / anti-bot markers for human-operated accounts, not an AI-content label.
• Verification is at the artist level, so a human can be “verified” while releasing AI-generated tracks.
• Several argue verification should be per-track, since estates or verified artists could flood catalogs with AI-generated material under a human name.

Desire to Block or Label AI Content

• Strong recurring request for:
  • A global setting to hide or deprioritize AI-generated music.
  • Clear labeling or warning badges on AI tracks or “bot” artists.
• Others want the opposite: the ability to search for or stream only AI music, or a dedicated AI-only service.
• Some argue mandatory self-declaration (“AI or human”) should be required for uploads, but others note creators will lie if there’s any penalty.

Spotify’s Incentives and Business Model

• Broad distrust of Spotify’s intentions: claims it benefits from AI/content-farm music because:
  • Generic background playlists can be filled with cheaper or in-house content.
  • Fraudulent streams might still generate ad revenue in the short term.
• Counter-argument: Spotify is squeezed by labels; shifting to podcasts, audiobooks, and maybe AI is a way to escape major-label control, not just to underpay artists.

Fraud, Bots, and Fake Artists

• Reports of:
  • AI tracks and AI covers flooding recommendation playlists and search results.
  • Bot-driven fraudulent streams and possible money laundering (paying for streams of one’s own AI tracks).
  • “Fake artists” and stock/background music already being commissioned or sourced for mood playlists; AI seen as the next step.

Listener Experiences and Alternatives

• Some users say they rarely encounter AI on Spotify; others say discovery playlists are now heavily contaminated with obvious AI “slop.”
• Several have canceled Spotify in favor of Bandcamp, Tidal, Qobuz, radio, self-hosted libraries, or torrenting, citing both AI and low artist payouts.

Philosophical Debate Over AI Music

• One camp: art should be human; AI music is derivative, “soulless,” undermines culture, and exploits copyrighted training data.
• Other camp: if it sounds good, it’s valid; many use music as background noise and don’t care about the creator’s biology.
• Nuanced views:
  • AI is acceptable as a tool within human workflows, but “fully auto” slop is undesirable.
  • Emotional connection often depends on believing there’s a real person and story behind the work.

Generational and Future Outlook

• Some predict an “AI-native” generation that embraces AI creation and finds current resistance quaint.
• Others expect a split: AI-centric consumers vs. a hyper-authenticity movement that values live, human, and analog music.

Concept & Core Metric

• Site tracks private-jet takeoffs as a proxy for elites fleeing cities before an “Event.”
• Metric is calibrated so roughly one day per year exceeds the max level, which some see as making a “Level 5” essentially non-alarming.

Feasibility of Jet-Based Warning

• Many argue nuclear strike warning times (minutes) are too short to reach an airport, prep a jet, and depart.
• Others note the system targets earlier indicators (wars, crises) where rich insiders might move hours or days in advance, not during inbound missiles.
• Several point out that in an actual nuclear exchange, sheltering locally may be safer than trying to fly.

Transponders, ADS-B, and Flight Procedures

• Multiple commenters say ADS‑B/transponders on modern jets are usually on by default once powered; disabling them is non-standard.
• In a functioning airspace system, flying dark risks delays, conflicts with ATC, and possibly interception; even in crisis, collision-avoidance is a reason to keep them on.
• Some note most business jets use towered airports, though many airports overall are untowered.

Who Gets Early Warning?

• Debate over whether “the rich” as a class have special apocalyptic notice, versus a much narrower set (senior officials, defense contractors, political insiders).
• Skepticism that information would be uniformly shared across all jet owners.
• Discussion of prepper-style plans involving helicopters-to-jets-to-bunkers, often citing New Zealand; others doubt such bunkers’ long-term viability or political acceptance.

Alternative / Additional Signals

• Suggestions: monitor government and “doomsday” aircraft, military tanker/cargo patterns, encrypted radio traffic, and exercises.
• Proposals to track large contiguous failures in weather stations or network infrastructure, with jokes about fiber cuts being indistinguishable from apocalypse.
• Ideas to integrate prediction markets (e.g., Polymarket bets on wars or religious events), though payouts might be moot in true apocalypses.

Similar Projects & Data Coverage

• Past projects mentioned, like an “Apocalypse Feed” combining network pings, space weather, asteroid data, and news scraping; another site tracks >1M-fatality risks.
• Critique that current implementation is US-centric due to FAA registry; ADS‑B data could broaden coverage to non-US jets.

Limitations, Noise & Latency

• Several note fundamental issues: noisy, incomplete data and signal construction that lags actual events.
• Lowering latency would raise false positives; more diverse, lower-latency sources would be needed.
• Some argue mainstream news and visible geopolitical escalation will remain more reliable indicators.

Tone of Reactions

• Many find the project amusing, “cool,” or more useful than vague “monitoring the situation” dashboards.
• Others see it as basically ineffective for real safety decisions but entertaining as art, commentary, or a curiosity.

Perceived Scale of Abuse

• Many argue the “at least 14” figure is almost certainly an undercount, since it only includes cases that surfaced via media, were detected, and often prosecuted.
• Others push back that moving from “undercount” to “widespread” requires more data, not just inference from human nature or prior police misconduct.
• Some note the base rate problem: 14 confirmed cases feels numerically small, but others respond that any number >0 is unacceptable due to the harm involved.

Data, Evidence, and Methodology

• The underlying review relied on media reports, which by definition miss undisclosed or quietly resolved incidents.
• Debate centers on what conclusions are justified from such a dataset and how to argue for policy change without stronger quantitative evidence.

Flock Systems, Auditing, and FOIA

• One commenter describes local Flock audit logs becoming anonymized over time, making it harder to spot suspicious usage.
• Concerns that Flock and similar vendors are insulated from public-records laws; some jurisdictions have explicitly exempted their data.
• Suggestions include regulation to require detailed, identifiable audit logs and making such data FOIA-accessible via the government agencies that use it.

Civil Liberties and Legal Frameworks

• Several comments connect ALPR abuse to broader surveillance issues and the “third-party doctrine,” arguing it should be narrowed for digital data.
• Some propose redefining “reasonable expectation of privacy” to account for aggregate tracking rather than isolated observations.

Accountability of Police and Institutions

• Disagreement over whether this is just individual misconduct or a systemic/institutional failure.
• Ideas include: mandatory malpractice-style insurance for officers, reduced special legal protections, stronger oversight, and real penalties to deter abuse.
• Skeptics of market-based insurance models warn about corporate incentives and regulatory capture.

Role of Surveillance Technology

• Split views: some see cameras and ALPRs as valuable crime-prevention and property-protection tools; others emphasize their chilling effects and ease of abuse.
• Repeated theme: any powerful surveillance system without strict, enforced oversight will be misused.

Article credibility and numbers

• Many commenters view the linked piece as low‑quality and likely AI‑generated, with unsourced or fabricated figures.
• A separate, paywalled report is cited as the real source; it mentions ~11% of backend code updates from AI agents and gives no concrete cost numbers.
• Back‑of‑envelope math suggests the supposed AI spend would be a tiny fraction of Uber’s multi‑billion‑dollar R&D budget; people argue the interesting question is outcomes, not raw spend.

R&D scope and “why does Uber need this much code?”

• Some question why a ride‑hailing company needs huge engineering and AI budgets.
• Others outline the operational complexity: global regulations, payments, fraud, logistics, data pipelines, experimentation infra, reliability, localization, etc., plus other business lines (Eats, freight, internal tools).

Token costs, pricing, and “tokenmaxxing”

• Several describe how easy it is to burn hundreds or thousands of dollars via long contexts, agents spawning sub‑agents, autoresearch loops, and LLM‑driven CI/CD.
• Contrast between heavily subsidized personal/teams subscriptions and expensive enterprise/API pricing (including Anthropic’s employee‑count cliff) is a recurring theme.
• “Tokenmaxxing” and internal leaderboards or minimum‑usage KPIs push people to maximize token use regardless of value.

Mandated AI usage and perverse incentives

• In some orgs, LLM usage is part of performance reviews; low usage can hurt ratings.
• This drives behaviors like using AI for trivial tasks (git commands, formatting, reading emails/logs) or running long, barely supervised agent loops.
• Commenters repeatedly invoke Goodhart’s law: once AI/token usage is a target, it gets gamed.

Productivity, code quality, and technical debt

• Enthusiastic reports:
  • Claims of 5–10× faster feature delivery, massive code generation (hundreds of thousands of LOC/week), and rapid progress on ambitious projects when humans do strong upfront design, testing and tight supervision.
  • Senior engineers use LLMs to offload “mechanical” work and focus on architecture, experimentation, and analysis.
• Skeptical reports:
  • AI‑generated code often lower quality, more verbose, and harder to maintain; organizations historically underprice technical debt and are now amplifying it.
  • Over months, codebases can bloat, architecture fragments across agent runs, and understanding of the system erodes; some expect net negative productivity long‑term.
  • Non‑SWE staff using LLMs to self‑serve can create brittle, insecure systems that engineering later has to own.

Measurement, ROI, and governance

• Many emphasize that software productivity is intrinsically hard to measure; attributing revenue jumps to AI is speculative.
• Consensus: high AI spend is not evidence of value; what matters is whether it replaces engineer time or accelerates valuable features without wrecking quality.
• Several argue AI tools need the same governance as cloud: per‑team budgets, caps, alerts, model selection discipline, and eventually “token budgets” per task/story.

Work visas, green cards, and costs

• H‑1B is a nonimmigrant work visa; EB‑3 is an employment‑based green card category, so not directly comparable.
• Typical H‑1B total cost (legal + gov fees) cited as roughly $5k–$10k, depending on firm size and premium processing.
• New $100k H‑1B fee applies in many cases where the worker is abroad or can’t change status in the U.S.; most employers now avoid such filings. Some expect litigation may overturn this; others just note its current practical chilling effect.
• H‑1B beyond 6 years is possible if green card steps are underway; otherwise people often pivot to O‑1.

Options for students, juniors, and founders

• For early‑career people from abroad: main routes discussed are H‑1B (lottery), F‑1 (study, then OPT), J‑1 (training), E‑2 (for some nationalities, including Serbia), and L‑1 via multinational transfer.
• U.S. master’s/PhD is framed as a strong though costly path to employment and later sponsorship.
• For founders: E‑2, L‑1, O‑1, and country‑specific visas (E‑3, TN, H‑1B1) are common; O‑1 for founders is still viable but seen as getting noticeably tougher, sometimes converging toward EB‑1A standards.

Trends in adjudications and processing

• Reported increases in RFEs/denials for EB‑1A/EB‑1B/NIW and O‑1; EB‑1A especially hard for non‑academic, non‑research profiles.
• Marriage‑based green cards are described as comparatively “quick and easy,” often around 6 months.
• N‑400 naturalization processing has lengthened from prior 6‑month targets to ~9–12 months or more, varying by field office.
• Some employment‑based (EB‑3) categories recently advanced quickly; expectation is eventual slowdown or retrogression.

PERM and labor‑market tension

• PERM is widely criticized as burdensome and artificial for both employers and applicants.
• Officially, employers must recruit in good faith and either hire qualified U.S. workers or terminate the PERM and wait ~6 months.
• Multiple commenters argue real‑world practice often undermines this spirit (e.g., obscure ads, ritualized rejections), and debate whether that constitutes legal “abuse.”
• Others stress legal standards focus on totality of evidence, not just technical box‑ticking; several examples of enforcement actions against tech companies are cited.
• Layoffs now frequently pause or derail PERM, especially in big tech; some see company layoffs, not regulation, as the main obstacle.

Students, OPT, and work authorization

• F‑1 students must have explicit work authorization (CPT/OPT) to be paid, regardless of contractor vs employee classification or whether the payer is U.S. or foreign.
• Unpaid “volunteering” in roles that are normally paid can still count as unauthorized employment.
• STEM OPT + self‑employment is described as legally tricky; one commenter suggests significant travel risk because denial of re‑entry has little recourse.
• J‑1 postdocs can spend >30 days abroad, but this triggers SEVIS alerts that require sponsor action; universities may treat extended absences as administratively complex.

Travel, advance parole, and green card maintenance

• Advance parole travel is still generally workable but viewed as more stressful; attorneys recommend pre‑trip counseling to prepare for CBP questioning.
• Risk of being denied entry on AP has “increased” but remains low absent other issues (e.g., criminal history).
• Long absences can lead to green card abandonment; there are limited exceptions (e.g., uncontrollable events like COVID travel restrictions). Returning resident visas and reentry permits are discussed as tools.
• Holders of consular‑issued immigrant visas who delay moving 6–9 months post‑entry are told that’s typically fine, but getting a reentry permit is advised if stays outside the U.S. may be extended.

Changing jobs after an employment‑based green card

• No formal rule requires staying with the sponsoring employer for six months after receiving a green card.
• What matters legally is intent: both sides must have intended ongoing employment at the time of approval.
• In practice, changing jobs soon after approval is described as rarely causing naturalization issues, especially given current job‑portability rules, though brief legal consultation is recommended.

Other visa categories and niche issues

• TN for Canadians/Mexicans is still considered relatively easy when the role matches listed occupations and the degree is clearly related; prior TN history neither strongly helps nor hurts.
• L‑1 “individual” petitions are described as consistently tough; blanket L‑1s at consulates tend to be easier. Policy now more tightly ties where one can apply to citizenship or residence country.
• L‑2S spouses are work‑authorized incident to status and typically don’t need an EAD; consular appointment timing governs how fast they can start.
• E‑1/E‑2 traders and investors: nationality of ownership, not place of incorporation, is key. Detailed case‑by‑case analysis is stressed.
• ITAR‑regulated industries do hire foreign nationals; roles may be restricted but employment is not impossible.

Remote work, compliance, and risk perception

• Some see a pullback from cross‑border remote hiring, often attributed to misclassification and compliance fears, but others only share anecdotal impressions.
• One view is that third‑party platforms may exaggerate legal risk to sell “employer of record” services; another stresses this is outside pure immigration law.

Ethics, enforcement, and politics

• There is a long sub‑thread on whether startups and large tech firms routinely “game” immigration rules, and how outside counsel should respond.
• Some participants push for much stricter consequences for visa and PERM abuse (including linking corporate layoffs to sponsorship bans), while others say this is unrealistic or harmful to legitimate immigration.
• Several comments express anxiety about political shifts, border detentions, and denaturalization; legal responses emphasize that outlier cases receive outsized media coverage and that lawful status plus clean records generally keep risk low.

Overall hiring landscape

• Thread is dominated by engineering roles, especially:
  • AI / ML / “agentic” systems across domains (developer tools, healthcare, finance, insurance, infra).
  • Core systems / infra (distributed systems, storage, networking, Kubernetes, edge, HPC).
  • Full‑stack web and product engineering for SaaS, fintech, devtools, and vertical apps.
  • Some non‑eng roles: product, design, DevRel, growth, sales, finance, QA, ops.

Common technical themes

• Heavy emphasis on:
  • Python, TypeScript/JavaScript, Go, Rust, Java, Elixir, Ruby on Rails.
  • Modern frontend stacks: React, Next.js, Vue, Svelte, Flutter, React Native.
  • Cloud & infra: AWS, GCP, Azure, Terraform, Kubernetes, Docker, Postgres, Redis, Kafka, Snowflake.
  • AI tooling: LLMs (OpenAI, Anthropic, Gemini), RAG, agents, orchestration (Temporal, LangGraph, LangChain), observability/evals.
• Many companies explicitly seek:
  • “Agentic-first” or “AI-native” engineers who already use tools like Claude Code, Cursor, Copilot.
  • Production experience with distributed systems, performance tuning, and reliability.

Work setup & geography

• Mix of:
  • Fully remote roles (often with time zone or country constraints).
  • Hybrid roles with 2–4 days per week onsite.
  • A sizeable minority are fully onsite, especially early‑stage startups, hardware/robotics, and some regulated or high‑collaboration domains.
• Locations span US, Canada, EU/UK, and parts of Asia; several roles require US citizenship or specific visas.

Compensation & equity

• Many postings give explicit salary bands; ranges vary widely:
  • Roughly mid‑five to mid‑six figures in USD (or local equivalents), sometimes all‑cash, often with equity or profit‑sharing.
  • Some highlight bootstrapped profitability and “earned equity” vs. options; others emphasize VC‑backed growth.

Community interactions & sentiment

• Frequent questions about:
  • Remote eligibility (e.g., non‑US or non‑EU applicants).
  • Tech stack choices (e.g., AWS ECS vs. Kubernetes, PHP vs. others).
• Several commenters express enthusiasm about:
  • Robotics, biotech, climate/energy, and socially impactful healthcare roles.
• Mild skepticism appears around:
  • Limited remote options, region restrictions, or perceived red flags in tooling.
• A few meta‑comments remind posters about thread etiquette and correct use (e.g., job vs. job‑seeker posts).

Scope of roles and experience

• Thread is a long directory of people seeking work across the stack: backend, frontend, full‑stack, mobile, ML/AI, DevOps/SRE, infrastructure, data, security, game dev, embedded, optimization, and design (UI/UX, product, brand).
• Many have 10–20+ years’ experience, including ex‑Big Tech, ex‑YC, founders, staff/principal engineers, and engineering managers.
• There are also juniors and students (CS undergrads, recent grads, early‑career full‑stack and data engineers) looking for first or next roles.

Remote work, location, and relocation

• Strong bias toward remote‑first work, often with explicit time‑zone constraints (EU hours, US Eastern overlap, etc.).
• Geographic spread is global: North and South America, Europe, Africa, Middle East, and Asia.
• Relocation preferences vary: some strictly remote‑only, some open within country/region, a minority willing to move globally if sponsored or for an exceptional opportunity.

AI/LLM and modern tooling

• AI/LLM experience is pervasive: many mention LLM apps, RAG pipelines, LangChain/LangGraph, agentic workflows, MCP servers, local inference, and model training.
• Numerous people emphasize using coding copilots and tools (Claude Code, Copilot, Cursor, etc.) as part of daily work.
• A few explicitly prefer non‑AI work or say they don’t want to “chase the fad,” highlighting a small countercurrent.

Engagement types and preferences

• Many are open to both full‑time and contract; a large subset specifically seeks freelance, fractional CTO/VP Eng, consulting, or short high‑leverage projects.
• Several want early‑stage or founding‑engineer roles with high ownership; others prefer stability at established companies.
• Some explicitly object to long interview pipelines, unpaid multi‑hour take‑home tests, and “calculator apps on Kubernetes.”
• A few posts are meta: brief questions or comments on links, sponsorship, and sales, plus one consulting group advertising for senior part‑time engineers.

Values and domains

• Recurring interest in meaningful or socially positive work: climate, healthcare, education, open source, civic/regulatory data, and non‑exploitative finance.
• Several explicitly avoid gambling, ads/surveillance, defense, or certain crypto niches.
• Many highlight strengths in ownership, mentoring, cross‑functional collaboration, and shipping real production systems end‑to‑end.

Concept and Use Cases

• Tool (ghostbox) is a CLI that spins up short‑lived dev machines using GitHub Actions runners, wiring in SSH and HTTP tunnels (Cloudflare, Tor) so users can “drop into” CI environments.
• Intended for debugging CI failures, manual builds/tests on multiple OSes, safe experimentation away from the local machine, and potential use by coding agents.
• Some commenters propose additional use cases: browser xterm interfaces, public links for trying CLI tools for ~90 minutes, integration with things like asciinema.

Implementation and Infrastructure

• Currently only uses GitHub Actions (Ubuntu, macOS, Windows “latest” runners).
• Tool creates a special private repository in the user’s own GitHub account to hold config and workflows.
• Author positions GitHub Actions as just the first backend and suggests adding other “Global Free Tier” providers later.

Licensing, Closed Source, and Trust

• Binary is proprietary, Rust-based, and free to use during a preview; source code is not published.
• Several commenters refuse to run a closed-source binary that has access to their GitHub account and repos.
• Debate arises over the author’s shift from previous open-source work to this closed-source tool; some find that pattern worrying.

Security and Abuse Concerns

• Multiple users are wary of piping curl | bash from an unfamiliar site and of giving secrets to undisclosed infrastructure.
• Accusations appear that the project could be malware or “vibecoded” slop; others suggest reverse‑engineering the binary before trusting it.
• Some fear the tool could be used for anonymous abuse or hosting “weird shenanigans”, with poor attribution.

GitHub Terms of Service and Ethics

• Large subthread argues this “resells” or “exploits” GitHub’s subsidized compute and clearly violates Acceptable Use; others insist it only spends each user’s own free minutes, similar to normal CI usage.
• GitHub disabled related repos for ToS violations (per the error message), though the author claims this is due to mass flagging and expects reinstatement; actual GitHub position remains unclear.
• Many worry tools like this will accelerate abuse, push GitHub Actions behind paid tiers, and harm the open-source ecosystem that relies on free CI.

Related Tools and Ideas

• Commenters mention comparable concepts: ephemeral environments, sandboxes, Segfault’s free shells, exe.dev’s UX.
• There is interest in a clearly documented, possibly self‑hostable or open implementation of ephemeral dev environments, but skepticism about this particular project’s approach and transparency.

Scope of the mistake

• Apple’s Support app accidentally shipped Claude.md files, which are instruction/config files for Anthropic’s Claude.
• The files appear to describe project structure, coding rules, and “don’t do X” guidance, not secrets like tokens or proprietary algorithms.
• Most see it as an embarrassing process slip (forgot to exclude from the app bundle), not a security incident.

Apple’s use of Claude and AI

• Cited reporting claims Apple runs custom versions of Claude on its own servers for internal tools and product development.
• Some argue this fits Apple’s pattern: rent external models while sitting out the public “AI arms race,” then buy/build when prices collapse.
• Others are skeptical of the reporting details and spin, and note Apple’s difficult B2B posture may have pushed it toward larger partners like Google for Siri/Gemini.

Siri, voice assistants, and LLMs

• Many think Siri has stagnated and lags behind Alexa, Gemini, and especially ChatGPT’s voice mode in understanding and usefulness.
• Counterpoint: for driving and home control, people prefer “voice-driven command-line” behavior—predictable, deterministic commands over chatty LLMs.
• Reports on Gemini/Alexa+ are mixed: some call them “objectively better,” others say basic tasks (timers, routines, smart home actions) regressed, became slower, or more unreliable.
• Hallucinations and non-determinism are seen as blockers for billions of users and safety-critical flows.

AI-assisted coding and “vibe coding”

• Multiple commenters say nearly every large tech company is pushing AI coding hard; Apple engineers almost certainly use LLM tools too.
• Concerns: AI-generated “slop,” tech debt, and review fatigue when humans must debug large, low-quality diffs from LLMs.
• Strong agreement that production changes still need human review and process; using AI isn’t the problem, lack of rigor is.

Should Claude.md live in source control?

• Many argue “yes”: agent instruction files are part of the codebase, akin to build configs, style rules, and documentation; they must be shared and versioned.
• Others initially treated them like local IDE cruft, but were persuaded that shared AI behavior and review automation justify checking them in.
• Consensus: they belong in the repo, but must be excluded from final builds.

LLM spam and community quality

• Several note that replies on X and increasingly HN show “LLM smell,” likely farming engagement/karma.
• This is seen as degrading discussion quality and pushing people toward smaller, curated communities.

Personal vs business websites

• Many initially react against the title because their personal sites really are “for them” (notes, experiments, blogs, art).
• After reading, they note the article is clearly about company/organizational sites, not personal homepages.
• Some wish the title had said “commercial” or “non‑personal” website to avoid confusion.

Who is the website for? UX vs OX

• Broad agreement (for business/gov sites): primary audience is users/customers, not founders, boards, or marketing.
• Others argue websites are also for the company: they must support business goals, convey positioning, and build brand trust.
• A detailed thread frames it as UX (user experience) vs OX (owner experience). In reality, many organizations optimize for OX: easing owner anxiety, satisfying leadership taste, and following firms like Gartner for reassurance.
• Several anecdotes show decisions driven by owner/marketing control over tools rather than end‑user outcomes.

Design, branding, and “website as art”

• One camp: a website is a tool with a job (help users accomplish their task); treating it as art leads to self‑indulgent design that doesn’t convert.
• Counter‑camp: brand identity and emotion matter; stripping away “art” makes the web soulless. Examples cited where overly artsy landing pages failed to convert, but also where idiosyncratic sites (e.g., HN) gained character over time.
• Ongoing tension between conformity/usability (Jakob’s law) and differentiation/innovation.

Designers, founders, and power dynamics

• Many complain designers often lack domain, business, or technical understanding, design for portfolios, over‑prioritize minimalism, and ignore established patterns.
• Others counter that designers are frequently scapegoats, downstream of unclear strategy and executive whims; “design theater” reflects broken product processes, not just bad designers.
• Power issues recur: HIPPO (highest paid person’s opinion), CEOs micromanaging colors/logos, and staff needing to translate superficial requests into underlying problems.

Practical UX failures and nostalgia

• Frequent criticism of modern sites: hard‑to‑find basics (restaurant hours/address, government service details), intrusive popups, autoplay media, scroll/UX hijacks, constant redesigns.
• Some praise older, stable, text‑first designs and simple gov design systems as more genuinely user‑focused.

Research, testing, and limits of UX

• Recommendations include simple usability tests (watching first‑time users) and ruthless editing of self‑reassuring copy.
• Skeptics note UX research can be over‑confident, biased, or unable to surface non‑incremental ideas (“faster horse” problem); best results come from combining research with informed judgment.

Hype, “Dangerous Models,” and Marketing

• Many see the “too dangerous to release” positioning of Mythos and Cyber as a marketing tactic: artificial scarcity, velvet ropes, and “my model is more dangerous than yours.”
• Some argue labs would release these models if it maximized revenue; withholding suggests either overhyped capabilities or genuine risk.
• Others think companies want to appear responsible and prepared in case their models are later linked to real-world cyberattacks.

Cybersecurity Capabilities and Verification

• Claims: current models have strong vulnerability-research capabilities and can find large numbers of bugs or vulnerabilities.
• Skepticism: lack of broad, trusted third‑party evaluation; some “benchmarks” are called anecdotal (e.g., tiny code samples).
• Some links and projects are cited as partial evidence that Mythos‑style capabilities are not uniquely beyond existing pay‑as‑you‑go models.
• Unclear whether Mythos is truly exceptional or just good marketing around incremental improvements.

Economics, Pricing, and Compute

• Discussion of DeepSeek V4 pricing being dramatically lower than OpenAI’s models; some suspect state subsidy, others note US tech has long been effectively subsidized too.
• Debate over whether inference is actually being subsidized: some say nobody is profitable at scale; others insist third‑party hosts and major providers have healthy per‑token margins.
• Compute scarcity and long lead times are seen as a major strategic factor; pre‑locking capacity may matter more than model quality.

OpenAI, Leadership, and Trust

• Strong distrust expressed toward OpenAI’s leadership, citing past reversals (e.g., RAM capacity rhetoric) and the CEO’s reputation for ruthlessness.
• Some note employees and media heavily backed leadership during prior board drama, suggesting internal loyalty but also a susceptibility to narrative management.

Safety Filters and Cyber Programs

• Users report more refusals on legitimate defensive security tasks and describe the Trusted Access Cyber program and its outsourced verification as clumsy.
• Debate on whether it’s technically possible to reliably distinguish offense from defense via text alone; some say in principle yes, others point to current tools as evidence of practical failure.

Local and Open Models vs Frontier

• Several argue local models are now “good enough” for many tasks and lag frontier models by only 6–12 months, undermining big labs’ moats.
• Examples of strong local models and new architectures are mentioned, though some users report reliability and context‑length issues.

Overall reception and purpose

• Tool inspects USB‑C ports and cables using Mac-specific interfaces, displaying capabilities and connection details.
• Many commenters find it immediately useful for identifying “mystery” USB‑C cables and plan to label their cables.
• Some see it as life‑changing or particularly helpful where hardware testers are impractical (e.g., blind users).

Functionality, UX, and feature requests

• Initial version was a menu bar app; several users dislike menu bar clutter and prefer a standard window or on‑demand tool.
• In response, the app gained:
  • A setting to disable the menu bar icon and run as a normal Dock/window app.
  • A command‑line interface.
  • Homebrew installation, with MacPorts support requested.
• Some suggest a widget (e.g., desktop or taskbar widget) as a better always‑on display model than a menu bar item.

Bugs, hardware limits, and weird behaviors

• Multiple reports of “No USB‑C ports detected” despite connected USB‑C devices.
  • One issue is “won’t fix” for Intel Macs because the southbridge reportedly doesn’t expose needed data.
  • Some Apple Silicon systems initially failed but were later fixed in updates.
• One user saw both USB‑C ports showing the same set of devices; cause is unclear.
• Early versions flagged cables as “plugged upside down,” which confused users; this behavior was later corrected.

Cross‑platform interest and ports

• Strong demand for Linux and Windows versions.
• Several community efforts emerged:
  • A KDE Plasma widget plus Linux CLI using /proc and USB tools.
  • A separate Linux-only CLI version without Qt dependencies.
  • References to existing Linux utilities (e.g., lsusb wrappers and PD‑aware tools).
• A Windows version is reportedly in progress.

USB‑C, e‑markers, and technical limits

• Discussion of USB‑C orientation handling: devices (not cables) generally manage lane swapping; some low‑end gear may fail in one orientation.
• USB‑C complexity is highlighted (power delivery levels, data protocols, wire gauges, pinouts), with frustration that the connector standard is marketed as if it implied consistent cable capabilities.
• The app reads e‑marker data; multiple commenters note:
  • Many cables lack e‑markers.
  • E‑markers can misreport capabilities.
  • The tool can only show what the marker and host report, not actual wire gauge, shielding, or signal integrity.
  • True quality or counterfeit detection would require specialized test hardware, bandwidth tests, or destructive analysis.
• ChromeOS is mentioned as another platform that can read e‑markers via a discovery message; typical Windows machines often cannot due to firmware limitations.

Development process

• The developer iterated rapidly, shipping many releases within hours, largely guided by HN feedback.
• Some praise this responsiveness; others worry the pace leads to “vibe‑coded,” under‑tested software and urge slowing down.
• LLMs (e.g., code assistants) are credited with enabling fast prototyping, ports (e.g., KDE widget), and CLIs, prompting a brief meta‑discussion about reduced cognitive friction in software creation.

Model performance & benchmarks

• Mixed views on capability. Some find Grok historically “dumb” vs frontier models; others say 4.3 is near GPT‑5.1 / Gemini 3 Pro preview level for many tasks but still behind April frontier releases in coding reasoning.
• External benchmarkers in the thread report: good agentic performance, small/dense outputs, but coding ability not competitive with top models (Claude Opus 4.7, GPT‑5.5, leading Chinese models, Kimi K2.6).
• Several note Chinese open models (Qwen, DeepSeek, GLM, Kimi) closing the gap or beating Grok and other closed models on code‑oriented leaderboards.

Speed, cost & value

• Speed is widely praised: ~200 tok/s in some tests; among the fastest “big lab” models. Some warn speeds often degrade after launch.
• Pricing ($1.25–2.50/M tokens) seen as aggressive vs Opus/GPT‑5.5. One benchmarker notes Grok 4.3 “reasons more,” so real costs can end up similar to earlier Grok 4.20 despite lower per‑token price.
• Some question simplistic “value” scores and argue cost should be measured per task completion, not per token.

Product experience & features

• Voice mode is repeatedly praised as unusually capable, natural, and apparently using a strong model (unlike some competitors’ cut‑down voice models).
• Users like its tone control, style matching, and multilingual naturalness, particularly for informal or nuanced communication.
• Complaints about the Grok app: no projects in apps, no memory, no tool/plugin integration from the UI, weak artifact/project workflows. Some expect the Cursor partnership/acquisition to fix coding harness and workflow gaps.

Use cases

• Reported strengths: conversational chat, tone editing, voice dictation, real‑time news/Twitter‑centric search, “what are people on X saying about X?” queries, grey‑area tasks (security scanning of self‑code, trafficking classification, edgy web tasks), D&D prep, casual what‑if scenarios, DIY and tax help.
• Others find it unreliable or underpowered for deep technical or compiler/experimental design questions, preferring Claude/GPT.

Safety, bias & politics

• Strong ethical backlash: many refuse to use Grok due to the CEO’s politics, alleged manipulation of outputs (e.g., “woke mind virus” framing, sycophantic answers about the CEO), and reported incidents of racist or extremist behavior (e.g., “MechaHitler,” “white genocide” insertions).
• Multiple links and claims about Grok‑generated CSAM (including children; “undressing” images) and EU/US investigations; some question scale and current feasibility but do not dispute that incidents occurred.
• Debate over whether slightly looser guardrails are good (enabling security work, classification, adult topics) or irresponsible, especially when combined with political steering.
• Broader point: all major labs curate outputs; some argue Grok is uniquely and openly steered toward the owner’s ideology, others respond that Google/OpenAI/Anthropic also embed their own politics.

Competition & ecosystem

• Many see Grok as “yet another subpar model” whose main differentiators are speed, price, looser alignment, and X/Twitter integration.
• Others welcome it as competitive pressure that may keep token prices down amid rising frontier‑model margins.
• Several note growing preference for open or Chinese models (Qwen, DeepSeek, GLM, Kimi, Gemma) for cost‑effective coding and local deployment, especially among power users and role‑play communities.

Availability and Workarounds

• Several commenters say it’s hard to get Mac Studio/Mac mini/Neo for development and testing.
• Alternatives tried: Hackintosh (described as painful and unreliable), macOS in QEMU/KVM (often too slow without GPU acceleration), and cloud Macs like MacinCloud/Scaleway.
• Some find dedicated Mac cloud instances workable, but cheaper/shared plans are too restricted.

Used Hardware, Neo vs M-Series Macs

• Many suggest used/refurbished M1 Macs as a practical stopgap; prices around $250–$300 are mentioned.
• Debate whether paying $300 for a 5‑year‑old M1 is reasonable versus waiting for a $600 new Mini/Neo.
• Some argue M1 is still more than adequate and even preferable in some ways (e.g., multiple Thunderbolt ports).
• Others emphasize that 8GB RAM (Neo baseline) is the real bottleneck; 16GB+ M1 is preferred for serious work.

Performance and Benchmarks

• Mixed anecdotal reports on relative performance of Neo vs M4 vs M1 on CPU-bound tasks.
• One user’s Mandelbrot benchmark shows Neo slightly ahead of M4 in single-thread workloads, which others find puzzling given core counts, clocks, and thermal limits.
• Broader debate around Geekbench: some see Apple silicon as “laughably ahead,” others call Geekbench misleading and point to more nuanced benchmarks where x86 chips often win at similar power or price.

Local LLMs, OpenClaw, and Mac Mini Value

• Many link shortages to demand for local AI inference (OpenClaw, local LLMs), though some say OpenClaw itself is only a minor factor.
• Mac mini is viewed as a sweet spot: good price/performance, unified memory useful for larger models, first-class iCloud/iMessage integration, and strong resale value.
• Some run OpenClaw-like setups on very modest hardware and see M4/Mac mini primarily as overkill for that, but attractive as always-on LLM boxes.

Supply Constraints and Planning

• One thread claims shortages are SoC-limited, not RAM-limited, with 3–4 month lead times at TSMC and little spare capacity.
• Neo demand is widely seen as underestimated; production targets allegedly had to be doubled, potentially missing an education-cycle window.

Ethics, Policies, and Sentiment

• Divided views on “buy-and-return” Macs to bridge repairs or short-term needs: some call it unethical, others say it fits Apple’s stated policies.
• Sentiment on Apple is mixed: praise for efficiency and value (especially Mac mini, Neo) versus criticism of declining quality, poor port/thermal decisions, and expensive high-end desktops.

Naming, Trademarks, and “Community Fork” Debate

• Many object to using “OpenWarp” and keeping “Warp” in the fork name, calling it misleading, disrespectful, and potentially a trademark problem.
• Others argue it’s only a legal issue if there’s a live trademark in the relevant class; links show “WARP” is registered for a terminal emulator and Cloudflare also has a “WARP” trademark.
• Several note that, etiquette-wise, serious forks traditionally change names and emerge from an existing community or core contributors.
• Some push back, saying forking with aspirational language is fine and that “forking is the heart of open source,” though the “community fork” label is widely seen as premature.

What Warp Is and How It Evolved

• Warp is described as a Rust-based terminal / “agentic development environment” with:
  • Block-based output navigation
  • Tab and workflow systems
  • Command prediction and AI-assisted command generation
  • Cross-platform consistency
• Earlier versions focused on a modernized terminal UI without AI; AI features and “agentic IDE” positioning came later, confusing some users about what Warp now is.

Desire for a Lightweight, Non‑AI Warp

• Multiple people want a “ThinWarp” or “neutered” Warp: same UI and features, no AI, no tracking, and no login/account requirement.
• Some already disable AI in settings; others reference a fork that strips AI and telemetry.
• OpenWarp is criticized for still apparently requiring a paid account to use one’s own AI provider, which some expected the fork to remove.

Business Model, Accounts, and Open Source Strategy

• There is frustration that Warp initially required accounts, remains network-dependent for some, and is layering in more AI/paid features.
• Some feel a fork is warranted to counter “enshittification”; others think this fork is hasty and should instead have started as upstream pull requests.

Value of an AI Terminal vs General Agents

• Critics ask why a special “AI terminal” is needed when AI tools and agents already run inside any normal terminal.
• Supporters see value in terminal‑specific AI (DevOps helpers, workflows, meta‑prompts) tightly integrated with the UI.

Website and UX Complaints

• Several reports that the OpenWarp site is “vibe coded” but broken: layout wider than screen, constant reflow from animated terminal demo, occasional non‑English text.
• This poor UX leads some to question trusting the project with payments.

Miscellaneous

• Many jokingly hoped “OpenWarp” meant OS/2 Warp being open sourced and express disappointment.
• Alternatives mentioned include Ghostty, Wave, and various AI command helpers and agents.