Rezo🛡₿RRR

BIP-361 as Bitcoin's quantum moment of truth. Jameson Lopp and Casa are proposing a phased freeze of quantum-vulnerable wallets before theft becomes possible. Mechanics: - Block sends to old addresses = 3 years post-activation - Invalidate old signatures = 5 years post-activation - HD wallet recovery proofs = quantum-safe migration path Not a hack, just infrastructure hardening with a migration window... and Satoshi's 1.1M BTC sit on exactly these types of addresses: - If BIP-361 activates, Satoshi doesn't migrate - That's ~5% of supply frozen forever - Meaning the upgrade itself becomes the deflationary mechanism I've written a bit and thought a lot about the quantum threat, and I've said that the Bitcoin dev community responds fast when it has to. Here's proof. No panic, no regulation, just a concrete technical solution with a concrete timeline. The "code is law" crowd will lose this round and not coos they're wrong, but cos the existential argument beats the principled one. Infrastructure evolves when it has to, not when it wants to... and right now it has to.

Cypherpunk Jameson Lopp and other Bitcoin developers propose BIP-361 to freeze quantum vulnerable wallets. This could lock dormant BTC like Satoshi Nakamoto’s 1.1M coins, now worth $74B, before quantum computers can steal them.

"Tesla’s new AI5 chip is about 3 times more power efficient than NVIDIA Blackwell at under 10% of the cost. I am super hardcore on chips right now. I have chips on the brain. I dream about chips. A single AI5 has ~5 times the useful compute of a dual SoC AI4." 一 Elon Musk

Nobody ships anything that matters on the first try. The only mistake is stopping at the first.

Nobody ships anything that matters on the first try. The only mistake is stopping at the first.

Nobody ships anything that matters on the first try. The only mistake is stopping at the first.

Perplexity scaled from $100M to $500M revenue on 34% headcount growth. Plans another 2x in 2026 while keeping team flat. Meanwhile: - OpenAI spent $41B last year - Anthropic spent $17.5B We're fixated on the frontier labs while tooling-layer companies quietly do the actual math. Revenue scaling 5x with minimal hiring is operating leverage, and it's available to every AI company running the numbers. The real question: who captures the margin on that efficiency? - If it's builders, then "built for builders" is the answer. - If it's the model layer, then efficiency without moat is just a race to the bottom with extra steps.

The upgraded Google app for desktop is now available for Windows users globally in English. 💻✨ Use a simple keyboard shortcut (Alt + Space) to instantly find what you need—information from the web, your computer files, installed apps, and Google Drive files—all from the Search box. With AI Mode in Search and Lens built right in, you can ask whatever’s on your mind or search what’s on your screen to get helpful AI-powered responses with links to the web.

Nobody ships anything that matters on the first try. The only mistake is stopping at the first.

Two things people are complaining about with Anthropic: 1) Claude got dumber, and 2) Claude keeps crashing. Telemetry from an AMD Senior AI Director across 6,852 sessions show the "dumber" part is real: - thinking tokens dropped 73% - API retries surged 80x - contradictions tripled Anthropic confirmed they lowered the default effort from high to medium... /effort max restores it and the model itself is unchanged. But here's what ties both complaints together. OpenAI's leaked COO memo put it plainly: "Anthropic made a strategic misstep to not acquire enough compute." The downtimes weren't random bad luck... the quality cuts weren't a product decision but infrastructure reality hitting a product.

Two things people are complaining about with Anthropic: 1) Claude got dumber, and 2) Claude keeps crashing. Telemetry from an AMD Senior AI Director across 6,852 sessions show the "dumber" part is real: - thinking tokens dropped 73% - API retries surged 80x - contradictions tripled Anthropic confirmed they lowered the default effort from high to medium... /effort max restores it and the model itself is unchanged. But here's what ties both complaints together. OpenAI's leaked COO memo put it plainly: "Anthropic made a strategic misstep to not acquire enough compute." The downtimes weren't random bad luck... the quality cuts weren't a product decision but infrastructure reality hitting a product.

Two things people are complaining about with Anthropic: 1) Claude got dumber, and 2) Claude keeps crashing. Telemetry from an AMD Senior AI Director across 6,852 sessions show the "dumber" part is real: - thinking tokens dropped 73% - API retries surged 80x - contradictions tripled Anthropic confirmed they lowered the default effort from high to medium... /effort max restores it and the model itself is unchanged. But here's what ties both complaints together. OpenAI's leaked COO memo put it plainly: "Anthropic made a strategic misstep to not acquire enough compute." The downtimes weren't random bad luck... the quality cuts weren't a product decision but infrastructure reality hitting a product.

basically: anthropic sneakily turned down how hard claude thinks before editing code, changed the default from "high" to "medium" effort, and hid the reasoning from session logs. all without telling users. an amd director had 7k sessions of telemetry to prove the degradation was real and measurable (not just vibes). anthropic admitted to the changes. there's a workaround (use "/effort max"). the uncomfortable part is most users had no data to notice it happened at all.

Nobody ships anything that matters on the first try. The only mistake is stopping at the first.