YourBlockDoc💊 📈 @yourblockdoc

The entire crypto community right now

unusual_whales @unusual_whales

a week ago

JUST IN: Jim Cramer has said he's reportedly selling all his Bitcoin, per CoinDesk

559 253 5K 2.2M 233

Bill Maher says his vote is up for grabs due to the DSA takeover of the Democratic Party… America is currently experiencing a bigger political realignment than the mid-1900s. The party of Bill Clinton is dead. Dems are now openly running on Socialism.

“Money won’t matter in 2036,” claims @elonmusk on the latest episode of The Insider. He tells @zannymb, The Economist’s editor-in-chief, how people will live when AI and robots can do every job and why he expects deflation, not inflation: bit.ly/4b9UPYK

God bless America.

Still one of the greatest skits ever.

The more time you spend outside and off social media, the more you realize that the doom and gloom is all fake. The average person is enjoying their family, working, and doing their favorite hobby. The economy is booming. People are throwing parties. Boomers are retiring. Church activities galore. Movie theaters are packed again. Spend more time OFF of social media and your happiness will go through the roof.

voided @voided

a month ago

The moment you realize what life looks like for people who don't know anything about AI, Politics or the Economy.

240 2K 20K 1.3M 3K

The Agent Loop Tax Nobody Wants to Talk About Andrew Ng’s framing of loop engineering is right: the new frontier is not just prompting an agent once, but designing loops that let it build, test, receive feedback, and improve over time. There are at least three loops that matter: 1. The agentic coding loop: the AI writes, tests, debugs, and iterates. 2. The developer feedback loop: the human steers product judgment, taste, context, and direction. 3. The external feedback loop: users, testers, production data, and market response reshape the vision. That is the right product pattern. But there is a hidden production cost inside it: The loop itself can become the most expensive component of the system. Not the final answer. Not the external API call. Not even the model’s reasoning. The loop. Every agent step may carry a large repeated context payload: • tool definitions • schemas • routing instructions • prior tool results • failed calls • retry instructions • planner state • memory state • intermediate outputs • provenance notes • next-step reasoning That repeated context is not free. It is paid for in tokens, latency, variance, and failure surface. Anthropic’s tool-search framing makes the problem visible. In a typical multi-server setup, tool definitions alone can consume around 55K tokens before the model does useful work. Tool selection quality can also degrade once too many tools are exposed, roughly in the 30–50 tool range. Loading only the relevant 3–5 tools can reduce tool-definition overhead by 85%+. That is the architecture signal. If the model has to inspect a giant tool catalog just to decide what to do next, the model is not only reasoning. It is doing backend routing inside the most expensive part of the stack. Now stretch that across a long-running loop. If an agent runs one step per minute for 24 hours, that is 1,440 model steps per day. If each step carries a repeated 55K-token tool-context burden, that becomes 79.2M input tokens per day just for tool definitions. Not reasoning. Not outputs. Not external tool results. Not retries. Not memory expansion. Just the catalog tax. At a $5 per 1M input-token tier, that is roughly $396 per day per always-running agent. If the same workflow reduces tool-context overhead by 85%, the repeated burden drops to about $59 per day. That is about $337 per day saved per always-running agent. At 100 always-running agents, that is roughly $33.7K per day saved. Again, before counting output tokens, retry loops, tool-result bloat, failed calls, memory expansion, or long-context pricing effects. This is why a lot of agent architecture is backwards. For open-ended research, exploration, and one-off workflows, model-native tool calling can be useful. Flexibility matters. But for bounded product workflows, the model should not be responsible for everything: • inspecting every available tool • choosing the tool • generating arguments • repairing bad arguments • retrying failed calls • managing credentials • preserving provenance • deciding fallback behavior • carrying giant schemas in context • synthesizing after every execution step That is too much nondeterministic work inside the model loop. A cleaner production architecture looks like this: Client intent → semantic router → deterministic backend fetcher → Open API / MCP execution → compact evidence envelope → model synthesis In this pattern, the backend owns the deterministic work: • routing • authentication • allowlists • validation • retries • timeouts • provenance • caching • fallback states • testable execution paths The model receives the smallest useful evidence envelope and does what it is best at: reasoning, synthesis, explanation, and decision support. That boundary matters. Agents as demos push everything into the model loop because it feels magical. Agents as production systems move deterministic work out of the model loop because it is cheaper, safer, faster, and easier to test. Go is especially strong for this middle layer: • typed contracts • fast HTTP • cheap concurrency • predictable memory • strict timeouts • easy service boundaries • deterministic tests • backend-owned credentials The model does not need magical access to every possible tool. It needs the right evidence, at the right time, in the smallest reliable form. That is the real lesson. The future of agents is not simply “give the model more tools.” The future is better systems boundaries. Model = synthesis. Backend = routing, validation, execution, retries, credentials, provenance, and control. Stop making the model do backend work.

Andrew Ng @AndrewYNg

a month ago

“Loop engineering” is a hot buzzphrase after mentions of it by Boris Cherny (Claude Code’s creator) and Peter Steinberger (OpenClaw's creator) went viral on social media. Loops are now a key part of how we get AI agents to iterate at length to build software. In this letter, I’d

405 2K 9K 671K 10K

@Pentosh1 @bitsurfer99 @Pentosh1 you're the goat and f the haters. Thanks for all you do and sharing all your wisdom 🧠

It will never pass because it will take 2/3ds of Congress and 3/4ths of all 50 state legislatures to change the U.S. Constitution, which only allows for income taxation, not assets. It will be easily challenged in court as it violates the Constitution on multiple fronts - the direct tax clause, the 5th, 14th, 16th Amendments and more. It would unconstitutionally tax unrealized paper gains—like the increased value of a house even if it’s not sold—and even illiquid assets. It would take an expensive, intrusive army of bureaucrats to figure out asset values and those fights, too will tie up courts for years to come. Democrats instead repeatedly don’t focus on first stopping waste, fraud and abuse in government - including the $2.4T blown on wrong payments since 2004, the hundreds of billions of dollars wasted and stolen in Medicare, Medicaid, pandemic relief and more. Fix your house first before you overtax the house itself. Government should demonstrate it can effectively manage existing tax revenues before seeking additional revenue. 9 nations in Europe abandoned it-including France, Germany, Denmark, Finland and Sweden, because it cost more than the tax would ever collect. After capital flight and the top brackets fled, nations in Europe then started over taxing the middle class. California’s own proposed wealth tax violates California’s own state constitution, including limits on taxation, uniformity requirements, due-process protections against confiscatory taxes, and strong protections for property rights. @GavinNewsom @ap @x @ABC @CBSNews @nbc @wsj @nytimes @FoxBusiness @FoxNews @eveningedit #news @elonmusk #breakingnews

Gavin Newsom @GavinNewsom

2 months ago

It's time for a national billionaires tax and a new social contract. 10% of Americans own two-thirds of the wealth. Wages have stagnated. The cost of living has skyrocketed.  The system is fundamentally broken. The federal tax code, a corporate code, and an inheritance code

23K 5K 28K 12.5M 2K

What a day for $STRC. When it rains it pours. When panic is setting in, it helps to revisit the fundamentals of why you made the investment in the first place. The main thing to focus on is that STRC is designed as a "bend, not break" instrument. Or, better yet, a "Stretch, not snap" instrument - yes, cringe. First, let's get some facts out there: 1. Strategy's total annual cash outlays are about $1.7B for dividends and coupons. Most of that is for STRC dividends. 2. Strategy currently has $1.4B in cash, and raised $300M last week on the $MSTR ATM. They raised $100M each of the prior two weeks as well. The stock is still trading at a small premium to mNAV as of today. They have likely raised a few hundred million additional this week, maybe even $300M . 3. Strategy's converts have put dates mostly in mid 2028; however, a third of the converts are still trading at a premium to par, meaning they would be unlikely to be put. 4. Strategy has a number of mechanisms at their disposal to help bring the price of STRC back to par. Ok, now for some additional thoughts: 1. The $1.7B that Strategy "owes" each year amounts to $33M/wk. The MSTR common traded $11B last week. $IBIT traded $6B last week. $BTC traded about $3B on Coinbase alone, while globally, traded multiples of that. If they need to raise $33M/wk, they can, basically forever. Even if they were just selling BTC and turned off the MSTR ATM, they'd be selling 545 BTC/wk at current BTC prices. They could do a mixture of MSTR ATM and BTC sales to spread it out. To put 545 BTC into perspective, so far in 2026, Strategy has acquired 174,863 BTC. So, if they simply sold what they acquired this year on a weekly basis, they could fund dividends and coupons for 74 months beyond the 10 months they already have - total of 84 months, or 7 years. What if BTC drops to $30k you say? They can fund things for 47 months, including existing cash, so 4 years. To be VERY CLEAR, this is only selling the BTC they bought this year, it leaves the other 672,500 BTC acquired prior to 2026 in tact. If you throw that all into the mix, they can fund cash needs for 31 and 16 years at $60k and $30k BTC respectively. If BTC is at $30k in 16 years, I think it is dead. TL;DR - cash outlays aren't a problem. No "death spiral" is triggered from selling a few hundred BTC/wk. 2. The converts seem really scary, with put dates coming up in 2 years roughly. However, there are a few things to consider. First, with a volatile stock like MSTR, they can very likely refinance the converts with another convert that has a later put date, later maturity and lower strike price on the conversion price. This is due to the very high value of the imbedded option in the convert. There's some dilution there for the common holders, but not crazy, since it's somewhat offset by the elimination of the old converts. Based on today's prices and a 30% conversion premium, that would be about 10% in net additional shares issuable upon conversion of the new convert. If the common float doubles between now and that future conversion (very likely), that's only a 5% dilution. Not amazing, but also not horrible for the MSTR common holders. That 5-10% dilution would push back the conversion date by many years. Additionally, other companies have successfully swapped converts for preferreds in the past, e.g. Strive, so there's no reason that Strategy couldn't come to an arrangement there with a number of the convert investors. TL;DR - the converts are not a huge problem and wouldn't require the company to come up with $6.7B in cash right on the spot. 3. Strategy has no margin debt. So, BTC could actually fall to $1 tomorrow and they would not be "liquidated". There is no liquidation price in the short-term. Period. 4. So, what is the "death" scenario for Strategy? Turns out, there's no killshot, just a slow bleed. Sure, if BTC falls to $20k and stays there for 6-7 years, then Strategy has a problem. But only at that point do they have a problem. If you think this is a realistic scenario, you should leave bitcoin immediately and short every company in crypto. Very few crypto businesses would survive that. So, putting it all together, Strategy is specifically designed as an anti-fragile fortress for bitcoin accumulation. Sure, leverage creates some risk, but it is also the ONLY reason they are able to accumulate BTC. If they had zero leverage, then IBIT would be the better investment. Any big hedge funds that want to "liquidate" them would need to keep the price of BTC generationally low for over half a decade. The biggest advantage that Strategy has is time. They can wait and be deliberate. To flip a common phrase on its head, Strategy can stay solvent MUCH longer than the market can stay irrational. Now, regarding STRC specifically, this is Strategy's flagship product. If they stop paying the dividends on it, they lose S3 eligibility for a year. That's a HORRIBLE outcome, so they would only stop paying the dividend in the most dire of circumstances. In fact, very few preferred stock issues ever stop paying dividends, and most that do, do it because of bankruptcy. They very much want to see STRC trading back at par, because it is their primary cash-raising mechanism for accumulating more BTC. That does NOT mean they are going to panic, raise the dividend to 15%, and sell a bunch of BTC to buyback all the STRC. They have plenty of time, and as long as you don't sell, you have time too. You continue to collect the dividend that you signed up for, and likely an increased one as they increase the dividend over time. You still get $11.5 per year per share of STRC, regardless of the market price. STRC can certainly go lower, but as long as you're not margined against it, it shouldn't really matter. It's very clear that Strategy has both the capacity and the desire to continue paying the dividends and ultimately support getting the price back to par. Maybe it takes a few weeks or maybe it takes a few months. STRC is a low time preference instrument issued by a low time preference company. The important thing is, there's no "breaking point" anytime soon. BTC basically must completely fail before there's a "breaking point". There is certainly a "panic point" though for investors, and with STRC down almost 15% so far this week, nearly 3x worse than BTC, it's very clear we are at "panic point". Ultimately, STRC was designed to remove the possibility of "breaking points". The goal is to "bend, not break". "Stretch, not snap."

There are MINI MAMDANIs popping up all around the country.   It is a dangerous thing — this is NOT a joke.   We are in a fight RIGHT NOW to save the Republic and EVERY AMERICAN needs to take this seriously.

@Pentosh1 @grok I love you pentoshi lol

I’m sick of it. The Bitcoin gloom and disappointment. Don’t you see what is underway? When you close your eyes, there’s one number you should see in your mind: $500T of fiat assets. That’s how much global asset value is sitting in Bonds (fixed income) & Money (M2 fiat currency). Why does that matter? Because that giant reservoir of ~½ the world’s asset value contains the potential energy necessary to power hyperbitcoinization. This is what Saylor sees. But do you see it yet? Consider Hoover Dam. The reservoir behind it contains 12 TWh of usable hydroelectric energy – it just looks like one big lake, calm and placid. But if you stick a pipe through that damn and put a turbine generator in the middle of it and let the water run through it… you can generate enough energy to power the city of Las Vegas for 5 years. That’s potential energy. Stored, untapped power. And by removing the barrier for the water to flow towards a lower energy state, you can harness the pent up power of the reservoir. This same mental model works for capital. A high Sharpe ratio is the financial analogue of a low-energy equilibrium state. Capital flows downhill, always seeking lower risk per unit of return. (Yes, I know everyone thinks about it as “highest return per unit of risk”, but this is the equivalent and helps understand the physical metaphor) Do you see it yet? Think about all the capital parked in fixed income instruments or money market funds. All of this capital is parked there because it has historically provided an acceptable trade-off of modest nominal returns for minimal risk. The entire premise of fixed income is “here’s a way to park cash in low-risk instruments that will generate a positive return slightly greater than inflation.” Adjacent to this asset category is “cash and cash equivalents” where the value proposition is somewhat smaller returns in exchange for even less risk. And over the decades, a steady stream of capital has found its way into these asset buckets that promise low risk and modest nominal returns via future fiat cashflows. These buckets have become a giant fiat reservoir, brimming with nearly $500T of capital. Do you see it yet? Along comes Strategy. @saylor realizes that much of this $500T of capital would be better off if it flowed into Bitcoin. But Saylor also recognizes that this reservoir of capital is inherently constrained. Boxed in by convention, investment mandates, risk management, volatility aversion, etc. It won’t flow to Bitcoin on its own. It can’t – it’s walled off, dammed up. Strategy engineers a solution. Creates a product to meet that capital where it’s at. The $500T fiat asset reservoir wants low risk, low volatility, fiat cash flows. Strategy designs preferred equity instruments that solve for these constraints, while Strategy uses the fiat capital proceeds to buy Bitcoin (which it believes will appreciate at 29% CAGR for the next 20 years). In exchange for capital today, STRC offers 11.5% annual returns with volatility asymptotically approaching 0. The Sharpe ratio is off the charts. It breaks everything in tradfi portfolio allocation. At first glance, it seems impossible. But it works because it’s not powered by risk-taking layered on top of fiat inflation; it’s powered by the ongoing monetization of a superior monetary asset whose endogenous properties ensure its appreciation when valued in fiat currency units over time. Saylor terms this kind of Bitcoin-powered fixed income offering “Digital Credit.” When a commodity flows from a high-energy state to a low-energy state, it releases energy. In the case of Hoover Dam, that energy can be used to power a hydroelectric turbine. In the case of Bitcoin treasury companies with Digital Credit offerings, that energy can be used to power shareholder returns for common equity holders. This can happen in every major capital market in the world. Do you see it yet? Strategy has stuck a pipe through the dam. A conduit through which capital can flow out of the Fiat Asset Reservoir and towards a low-energy equilibrium state. Digital Credit offerings (e.g., STRC, SATA, and others) create that value proposition. And what’s the Total Addressable Market (TAM)? All $500T of fiat assets in the reservoir. The recent SpaceX IPO Prospectus recently made a splash by claiming the company had a combined $28.5T TAM, proclaiming that this was the “largest TAM in human history.” But my essay from 2023 titled “Bitcoin’s Full Potential Valuation” already articulated how Bitcoin’s TAM is all value itself, above and beyond the usual lens of annual economic activity across industries. Saylor read it, adopted it for his presentations, and built on it with the Bitcoin24 valuation model. The SpaceX Prospectus is wrong. Bitcoin has the largest TAM in human history. And Digital Credit has the second largest TAM in human history – the $500T Fiat Asset Reservoir. Do you see it yet? Digital Credit offerings will redirect some % of the $500T Fiat Asset Reservoir into Bitcoin. This will happen because the value proposition of Digital Credit offerings is higher Sharpe than anything I am aware of in the entire $500T reservoir, inflation-adjusted. Think of it as the Second Law of Capital Dynamics: capital flows toward assets offering superior risk-adjusted returns. If Digital Credit ingests 1% over the coming decades, that’s $5T. It seems unreasonably pessimistic to think that only 1% of the $500T Fiat Asset Reservoir would be interested in vastly better returns with a similar (or better) risk profile. Let’s say Digital Credit appeals to a (still-conservative) 10% of the $500T fiat asset reservoir, that’s $50T. Bitcoin is currently a $1.5T asset. Do you see it yet? Digital Credit may direct a torrent of $50T of capital into Bitcoin over the coming decades. All of it bidding for a finite supply of Bitcoin. The scale of that inflow would likely drive Bitcoin’s valuation to $10m/BTC, or ~$200T total. Digital Credit is the plumbing of hyperbitcoinization. This is how it happens – you’re watching the early stages of Bitcoin’s monetization megatrend. The question is: do you see it? Or will it have to play out first?

@BitcoinMagazine @jelliebeeann confirmed @Drake is Satoshi 🔥 🦉 🔥

JUST IN: Legendary rapper Drake raps about Bitcoin in his new song: "I am a BTC crypto big-timer" 🚀

It will soon be common knowledge that China has been pushing harmful policies from inside of our own country… DEI, absurd climate regulations, wealth taxes, soft-on-crime reforms, illegal immigration and even the anti-standardized test movement.

Insider Wire @InsiderWire

3 months ago

#BREAKING: California mayor Eileen Wang charged as an illegal Chinese agent; set to plead guilty.

326 1K 16K 2.8M 2K

"21M"

⚡️The actual finding, stripped of all comfort, is brutal. Your brain is too powerful for the life you’re living. That’s what the anxiety is. A machine built to process enormous complexity being fed a diet of emails and social media and minor interpersonal drama. The mind responds the only way it can. It manufactures complexity where none exists. It creates problems to solve because you won’t give it real ones. It spirals because spiraling is what an understimulated high-performance engine does when it’s idling. The overthinking isn’t a symptom of intelligence. It’s a symptom of waste. Wasted capacity. Wasted potential. Wasted time. The mind is screaming at you that it was built for more than this and instead of listening you’re liking a tweet that tells you the screaming means you’re special. The people who actually channel high IQ into something real don’t overthink. They don’t have time. They’re consumed by the problem they chose. The anxiety burns off like fog when the sun comes up because every cycle is allocated. Every thread is committed. There’s no spare capacity left to spiral with.

World of Statistics @stats_feed

4 months ago

People with higher IQs are more prone to anxiety and overthinking, according to multiple psychological studies.

173 386 5K 339K 695

@skyyhh @jelliebeeann

@jelliebeeann It's more effective this way 😎