Do Bitcoin Core Devs Lock Their Doors At Night?

Untitled spark

xNostr

Fmr CIA officer here. Let me share how elections get stolen — and the role of intel agencies. We a...

How Elections Get Stolen: A Former CIA Officer's Warning A former CIA officer recently took to social media to lay out what he describes as a largely hidden truth about American elections — one that, he argues, both parties in Washington quietly understood for years but rarely discussed openly. The thread, posted by Bryan Dean Wright, draws on his intelligence background to connect the dots between classified capabilities and domestic electoral vulnerability. What the Intelligence Community Knows Wright's central claim is straightforward: no voting machine is truly "unhackable." According to him, the CIA and NSA maintain dedicated teams capable of penetrating virtually any electronic system — including voting machines stored in locked, air-gapped rooms with no internet connection. These operations, he says, rely on a combination of human intelligence (HUMINT), signals intelligence (SIGINT), and other technical disciplines, along with meticulous planning and what he calls "exquisite tradecraft." What makes this claim notable is his assertion that members of Congress on both sides of the aisle were regularly briefed on these capabilities and privately acknowledged the danger. Historical footage from Congressional hearings, he notes, shows both Democrats and Republicans expressing alarm at the idea of so-called "unhackable" systems — because they knew better. "They were briefed by NSA and CIA teams on how they 'hack the unhackable' every month without leaving a trace." The Taiwan Model — and Why America Ignores It Wright holds up Taiwan as a practical counterexample. Despite having a population of roughly 24 million people and operating under constant geopolitical pressure from China, Taiwan runs its elections with striking simplicity: Same-day voting only In-person participation with ID requirements Paper ballots A counting process open to public observation Results announced the same night Wright's argument is pointed: if Taiwan can pull this off reliably and transparently, there is no technical or logistical reason American states cannot do the same. The refusal to adopt similar safeguards, he contends, is not an oversight — it is a deliberate choice that creates exploitable vulnerabilities. He singles out California specifically, noting that the state's roughly 30-day vote-counting window is, in his view, an invitation for manipulation. The extended timeline and widespread use of mail-in and harvested ballots, he argues, provide cover for the kind of untraceable interference that intelligence professionals are trained to execute. The Playbook: Deny, Smear, Deflect Wright describes what he sees as a recognizable pattern whenever electoral irregularities surface. Like any well-run covert operation, the goal is to leave no fingerprints. But when something does leak, he says the response follows a familiar script: deny the allegation, attack the credibility of whoever raised it, and launch counter-accusations designed to muddy the waters and preserve the status quo. This, he argues, is not coincidence — it is operational discipline applied to domestic politics. The Bigger Picture Wright frames all of this around the Los Angeles mayoral race involving Spencer Pratt's challenge to incumbent Mayor Karen Bass, but he is explicit that his larger concern goes well beyond any single contest. He describes what he calls a "corrupt machine" — a structural system, particularly in Democrat-dominated cities and states, that he compares to the legendary political machines of old Chicago, New Jersey, and Louisiana under Huey Long. His conclusion is stark: candidates running against entrenched incumbents in these environments are not simply competing against a political opponent. They are competing against a system engineered, he believes, to produce predetermined outcomes. What Should Readers Make of This? It is worth noting that Wright's account is one perspective — an opinion thread, not a formal intelligence report or court filing. Claims about specific electoral fraud in any given race require evidence and due process to evaluate fairly. Reasonable people disagree sharply about the scale and nature of election integrity problems in the United States, and courts have repeatedly found insufficient evidence to overturn results in cases brought after recent elections. That said, the underlying technical point — that sophisticated state-level actors possess tools capable of compromising electronic voting infrastructure — is not fringe speculation. Cybersecurity experts and government agencies have long acknowledged vulnerabilities in voting systems, and the push for paper ballot backups and robust auditing has bipartisan support in many security circles. The debate Wright is entering is real, even where his specific conclusions remain contested. The question of how to build electoral systems resilient enough to earn broad public trust — and whether current systems meet that bar — is one that democratic societies cannot afford to stop asking.

xNostr

this is literally the best deal on the market rn for an ai coding plan > $10/month > kimi k2....

The $10/Month AI Coding Plan That's Quietly Outperforming Its Price Tag There's a quiet revolution happening at the budget end of the AI coding market. While most developers are paying $20 a month for tools from the big Western labs, a growing number are discovering that a $10 subscription can deliver more raw usage, more model variety, and surprisingly competitive performance — if you know where to look. The buzz right now is centered on one specific tier: a $10/month plan that bundles access to three powerful Chinese frontier models — Kimi K2.5, MiniMax M2.7, and GLM-5 — under a single subscription. Users who've tried it describe it as "way more usage than $20 plans with big labs" and "super underrated." That's a bold claim. But when you dig into the numbers, it holds up. What You're Actually Getting The plan in question is the Atlas Cloud Coding Plan Starter tier , which provides 800,000 credits per day across a lineup of models that would have been considered frontier-class just months ago. At that credit budget, a typical medium-sized coding request — say, 5,000 input tokens and 1,000 output tokens — gets you somewhere between ~50 and nearly 500 requests per day , depending on which model you use. Lean on the faster, lighter models for routine tasks, and the heavier ones for complex reasoning, and that daily budget stretches remarkably far. The three headline models are worth understanding individually: Kimi K2.5 — Built by Moonshot AI, this model shines on research-heavy coding tasks. It carries a massive 256K token context window (the largest of the three), scores impressively on math and scientific benchmarks (AIME 96.1%, GPQA 87.6%), and supports image input — making it a strong pick when you're wrangling long documents or need deep reasoning alongside code generation. MiniMax M2.7 — The speed champion of the group, running at up to 100 tokens per second on the high-speed tier and hitting a 78% SWE-Bench Verified score , which sits at or above Anthropic's Claude Opus 4.6 on the same benchmark. It's especially well-suited to production coding workflows where low error rates matter — one real-world 48-hour test on a Next.js/Convex project showed roughly 5 review fixes needed vs. ~50 for Kimi . Its pricing is also the cheapest of the three at around $0.30 per million tokens. GLM-5 — Developed by Zhipu AI and trained entirely on Huawei Ascend chips (no Nvidia hardware involved), GLM-5 is a 744B parameter model with 40B active parameters per token. It scores 77.8 on SWE-bench , rivaling Claude Opus 4.6, and is fully MIT-licensed — an unusual perk for a model at this capability level. How It Stacks Up Against the $20 Competition To put this in context, here's what the standard $20-per-month options offer: Claude Pro ($20/month) — Access to all Claude models plus Claude Code CLI, but on a shared token budget that can feel constraining for heavy daily users. ChatGPT Plus ($20/month) — GPT-4.5, some Codex access, and general-purpose tools. Solid for mixed workflows but not deep coding focus. Cursor Pro ($20/month) — Great IDE integration with unlimited tab completions and a frontier-model credit pool, but the credits run out fast on intensive sessions. GitHub Copilot Pro ($10/month) — The most obvious comparator at the same price point, offering inline completions and agent mode, but you're locked into GitHub's model choices with no flexibility. What the Atlas Cloud Starter plan offers that most of these don't is model optionality . Rather than being tied to one provider's decisions about which model you get, you can route different tasks to the model best suited for them — and all through an OpenAI-compatible API format that works with Claude Code, Codex, and any other tool that speaks that standard. The cost efficiency is striking when you zoom out. Running a million API calls per day through Claude Opus 4.6 would cost roughly $150,000 a month . The same workload through MiniMax M2.5 runs about $9,000 — a 94% reduction. Even at individual developer scale, those efficiency ratios matter when credits and rate limits hit. Who This Plan Makes Sense For This isn't necessarily the right plan for everyone. A few honest caveats apply. If your workflow is deeply integrated with GitHub or VS Code and you rarely think about which model is running underneath, GitHub Copilot Pro at the same $10 price is simpler and probably good enough. If you're a heavy Claude Code user and the quality difference matters on nuanced tasks, Claude Pro may still justify its $20 tag. But for developers who: Run multiple coding sessions daily and keep hitting rate limits on mainstream plans Want to experiment with different models for different task types Are comfortable working via API or tools like OpenCode, Cline, or Claude Code pointed at a custom endpoint Are curious about the rapidly improving Chinese frontier models …this plan is worth a serious look. One practical tip from developers who've tested these plans: don't abandon your existing subscription cold turkey. Start with a pay-as-you-go option to understand your actual daily usage patterns before committing to a subscription tier. The math only works in your favor if you're actually using the credits available. The Bigger Picture What's happening here reflects a broader shift in the AI model landscape. Chinese labs — Moonshot AI (Kimi), MiniMax, and Zhipu AI (GLM) — have spent the past year closing the gap with Western frontier models on coding benchmarks, and in some cases surpassing them. They've done it while pricing aggressively, sometimes at 5 to 17 times cheaper than equivalent Western models on a per-token basis. That's created an opening for API gateway services to bundle these models at prices the big labs simply can't match — at least not yet. The $10/month plan making waves right now may not be the flashiest tool in the shed. It doesn't have the brand recognition of Copilot or the marketing muscle of OpenAI. But on pure value-per-dollar for a developer who writes a lot of code, it's hard to argue with the math.

xNostr

Non-KYC debit card with tap to pay. Build this and I will give you so much money.

The $100 Million Idea Hiding in Plain Sight: No-KYC Crypto Debit Cards There's a product that millions of people desperately want, that already commands premium fees, and that remains stubbornly hard to build well. It's a simple concept on its surface: a debit card you can tap anywhere, loaded with crypto, with no identity verification required. Someone on the internet put it bluntly — "Build this and I will give you so much money. You can ask for an insane fee and people will still use it." They're not wrong. So why doesn't a clean, reliable version of this product exist yet? The answer involves a collision between financial privacy, global regulation, and the sheer technical difficulty of bridging the crypto world with legacy payment infrastructure. What Is a No-KYC Crypto Card, Exactly? A no-KYC crypto card is a prepaid or debit card tied to a cryptocurrency balance — Bitcoin, USDT, ETH — that works wherever Visa or Mastercard is accepted, without requiring the user to submit a passport, address, or any other identifying information. Traditional crypto cards from the likes of Coinbase or Binance require full Know Your Customer (KYC) verification, meaning you hand over your identity before you spend a single dollar. No-KYC cards flip that model, prioritizing privacy over compliance. The demand is real and the use cases are varied: Privacy-conscious users who don't want their spending habits tied to their identity Crypto traders who want to take profits quickly without going through exchanges People in underbanked regions who lack the documents traditional KYC demands International travelers avoiding foreign transaction fees and currency friction What Already Exists — And Why It Falls Short The market has responded, but imperfectly. Several providers now offer partial solutions: BingCard issues instant virtual cards with no KYC, rechargeable with BTC, USDT, ETH, and USDC, charging a 1% transaction fee. Physical cards, however, require KYC verification. PlasBit, Goblin Cards, and SolCard are offshore providers offering varying degrees of privacy, though most users report limits on spending and unreliable acceptance. Bitrefill lets users convert crypto into gift cards — a workaround that technically sidesteps KYC but is clunky and limited. One YouTuber who publicly championed a card called "Up Card" noted he used it at conferences in Germany and Dubai, paying for hotels, restaurants, and yes, handbags — but was candid about the risks: "If these guys get up and leave tomorrow, there's no way of me knowing. You will lose your [funds]." That counterparty risk is the core problem with today's offshore solutions. The honest reality? Most of these products are either virtual-only (no tap-to-pay), geographically restricted, or carry serious financial risk if the issuing company disappears. Why Building This Is So Hard The obstacle isn't technical imagination — it's regulatory geography. The Regulatory Patchwork Financial rules vary enormously by region , and card networks like Visa and Mastercard add their own compliance layers on top: United States & Canada : Regulated by FinCEN, practically all prepaid card issuers must perform KYC. Cards that try to operate without it face enforcement or shutdown. European Union : The Fifth Anti-Money Laundering Directive (5AMLD) enforces KYC, but allows exemptions for low-value prepaid cards under €150 — a narrow but real window. Southeast Asia : Countries like the Philippines have more relaxed prepaid card regulations, which is why many offshore no-KYC products are incorporated there. India : The Reserve Bank of India requires full KYC on prepaid instruments; no-KYC options exist only in legal gray zones. This patchwork means a product that's legal in Manila may be blocked in Miami. Any serious builder has to decide which markets to serve and structure their company accordingly. The Card Network Problem Even if you navigate local regulation, Visa and Mastercard have their own compliance requirements for card issuers. Getting a card onto one of these networks requires a banking partner, and banking partners are deeply averse to anything that looks like it could facilitate money laundering. Offshore incorporation can work around some of this — but it introduces the very counterparty risk that makes users nervous. The Real Opportunity: Tap to Pay Changes Everything Virtual cards exist. The gap in the market is physical cards with contactless tap-to-pay functionality that don't require identity verification. That's the hard part — and the lucrative part. A virtual card works for online purchases. But paying for coffee, tapping through a subway turnstile, or splitting a dinner bill requires a physical card with NFC capability. That's what people are begging for. The EU's low-value exemption under 5AMLD might be the most promising legal path. A card capped at €150 in stored value, issued through a compliant offshore entity with a European banking partner, could theoretically offer tap-to-pay without full KYC — at least for small, everyday purchases. It wouldn't be perfect, but it would be real. The Bottom Line The demand signal here is unmistakable. People are willing to pay premium fees, tolerate offshore risk, and work around clunky workarounds just to spend their crypto privately. That's not a niche preference — it's a market gap screaming to be filled. The builder who cracks the combination of legitimate card network access, a permissive regulatory jurisdiction, NFC-enabled physical cards, and a non-custodial or low-risk fund structure won't just have a product. They'll have a monopoly on something people genuinely need. As existing no-KYC card guides note , the space is growing — but the definitive solution hasn't been built yet. The money is waiting. The question is whether someone can thread the regulatory needle to claim it.

xNostr

Nick tells meme coin traders who lost money on his token to get out of the trenches and go get a job...

dddd

xNostr