Block Skunk @blockskunk

@NSF blockskunk.com/blog/01-21-26-…

In July 2026, National Science Foundation (@NSF) selected NSF Critical Materials Crossroads Engine in the Kansas City Region for up to $160M. A 260+ partner coalition working to onshore nickel, gallium, and rare-earth magnets the U.S. currently imports 100% from foreign competitors. Dr. Anthony Caruso CEO of CMC, put the investor standard plainly: "Almost every device in this room has a metal in it that is not in our control. It's not in the US's control to produce that metal. We rely on China for 𝟭𝟳 metals. 100% reliance for 𝟭𝟳 metals... When we're in a political state like we are today, where metals are used as leverage against each other, we have to be in a position to have an independence. And so that's why this started here in Kansas City, and we are going to be a blueprint for the nation" A shared dashboard across 260 universities, refiners, and defense off-takers can display milestone schedules. It cannot prove chemical purity, origin compliance, or chain of custody. When a component in an F-35 or grid transformer fails spec, a portal entry stating "approved" is worthless. The same boundary appears in cybersecurity. Under the new Cybersecurity Maturity Model Certification (CMMC), United States Department of War where contractors must prove continuous data controls across multi-party supply chains. Traditional compliance tools manage risk inside one organization. The friction lives in the seams between organizations. Most enterprises still can't trace their exposure through the refining step. Geographic diversification targets mines. Supply control lives at refineries. The IEA projects top-3 refiner market share holds at 82% through 2035. The diversification playboo k is targeting the wrong chokepoint. Three questions worth asking your supply chain team today: → Can you trace tier-2 and tier-3 suppliers through the refining step? → Cryptographic verification, or self-reported certificates? → If a major refining jurisdiction imposed licensing restrictions tomorrow, which products are affected? A dashboard tells partners what operators chose to display. Independently verifiable evidence lets an assessor or counterparty check integrity without trusting the dashboard owner. Stratum for shared record. Arbiter for evidence you can forward without rebuilding the file. Where in your supply chain does "shared dashboard" currently substitute for "shared proof"? Full article in comments 👇

@MarketMeMoreInc + @LV8R_LABS

Agents are getting keys to live ledgers. Compliance is still Kevin, a spreadsheet, and a deadline. While stablecoins and agentic interfaces move funds 70x faster, automated rails don't fix the real cost Settlement is only 8% of the transaction! The remaining 96% is coordination tax, legal friction, and multi-party reconciliation across isolated systems. Across fintech, payment platforms are shipping agentic interfaces: MCP as the access layer, multi-agent audit trails as a pressure point, and guardrails so autonomous agents can work on live ledgers. That is necessary, but it is not the same as shared, independently verifiable authorization. If the explanation lives only inside a single vendor's console, every counterparty, auditor, and assessor still has to trust that console. True enterprise agent authorization requires four non-negotiables: → Policy binding: Proof that execution boundaries held in real time. → Identity binding: Which agent acted, under which credential, at which exact block height. → Record binding: Which shared state was read or mutated, logged immutably. → Independent evidence: Audit trails that survive export and regulatory challenge without a vendor in the middle. This is why we built the stack: Stratum deploys alongside legacy systems to bind shared state across counterparties. Arbiter seals compliance-as-code evidence that assessors verify independently. Legate serves as the MCP layer committing every tool call, args digest, agent identity, and block height to an append-only, examiner-downloadable ledger. When your AI agent "explains its action," can a third-party auditor verify that explanation against an independent record?

Four jobs explain why dollar coins keep winning specific markets. None of them is an operating technology. 1. Access. About 1.3 billion adults remain unbanked. Roughly 900 million already have a phone. A dollar coin can sit in that wallet. A tokenized deposit still needs a bank. 2. Settlement certainty. Repo is the money market now, or about $12 trillion. Fed funds are under $100 billion. A firm that owes dollars has to find a safe claim. Thin markets match about 36 percent of the time. A dollar coin skips that search. In the same model, emerging-market firms issue about 36 percent of debt in dollars without the coin, and about 88 percent with it. 3. Always-on rails. Programmability and atomicity move collateral for margin and overnight repo. Less capital sits waiting on sequential clears. 4. Thickness. Coin reserves concentrate demand in short Treasuries. That can ease bill funding at the margin. It is a macro mechanism. It is not a fiscal plan. When banks form a joint dollar-coin vehicle, the gravity is real. A shared dashboard is still not independently verifiable proof. None of those jobs produces authorization and evidence that survive counterparty and examiner challenge by themselves. Stratum: ordered, permissioned shared state. Arbiter: continuous evidence assessors can verify. One shared record. Independent proof.

.

Jason Zhou @jasonzhou1993

3 weeks ago

All existing SOC2 software is so goddam stupid someone must build one where i invite an agent email to my platforms, it auto fetch with CLI or computer use, rather than me taking endless screenshots with timestamps 🙄

11 1 33 4K 9

@andreysuperior @meigga @RebeccaRettig1 “There are no solutions. There are only trade-offs.” -- Thomas Sowell

Dollar stablecoins look cheap where dollars are scarce. That is not a cheaper transfer. It is a different job. Wenxin Du at Jackson Hole compared the dollar price you get through a stablecoin with ordinary Bloomberg rates. Turn dollars into Argentine pesos that way, and the rate looked about 44 percentage points better. Into Nigerian naira, about 16 points better. Almost nobody was trading: about $2 million a day in the peso pair, and about $100,000 in naira. That gap is an access premium. When inflation is high and cash is hard to move, people pay extra for a dollar they can actually hold and send. Official dollars are scarce. The stablecoin is expensive. So one stablecoin buys a lot of pesos or naira. Those prices are thin. They do not scale. This is not cheaper settlement in New York or Frankfurt. Sending $100 from the United States to a euro account through a stablecoin shows the other side. Du’s example already costs about 2.2 percent in fees to get on, hop, trade, and get off, before the spread. That already costs more than established money-transfer firms on the same path. Euro trading against USDC and USDT on crypto exchanges is about 0.02 percent of ordinary euro foreign exchange. The rail looks modern. The market is a rounding error. Project Hangang’s table at Jackson Hole named the job dollar stablecoins still win. People who want dollars and do not have a bank account. A bank deposit, even a tokenized one, still needs a bank. A payment stablecoin can sit in a wallet with no account. The wallet still meets a bank at the edge. Someone still has to check identity, screen the transfer, and keep sanctioned parties off the flow. Du’s cost split is the same lesson. Moving the money is cheap. Compliance, disputes, and customer support are where the bill lives. Jackson Hole still preferred tokenized bank deposits for settling large payments between banks. Dollar stablecoins keep winning where people need a dollar they cannot get another way. An access premium is the price of a dollar the banking system does not sell.

Finance assumed a shared map. Parties knew different facts. They still spoke the same language. Agentic AI introduces a deeper friction. @MarkusEconomist (Princeton) called it asymmetric understanding. Agents can model how humans respond. Humans cannot reliably anticipate the agents. Non-explainability is the mechanism. The agent’s self-report is another output. It is not a transcript of the decision rule. Raghuram Rajan (Chicago Booth) kept the tools that already handle opacity: capital, collateral, margins, skin in the game. The sharper risk may be correlated behavior, the 1987 portfolio-insurance pattern, not a new metaphysics. Policy still needs more evidence before a rewrite. Stefano Scarpetta (OECD) named the supervisory seam. AI helps AML and fraud detection. It also equips attackers. Models arrive through third-party providers, often outside the exam perimeter. The FSB’s June 2026 consultation on twelve sound practices for responsible AI adoption covers organisation-wide governance, the AI lifecycle, and third-party cyber risk. It is a consultation. It is not a shared, independently verifiable trail. fsb.org/2026/06/fsb-co… EU AI Act Article 12 already requires high-risk systems to automatically record events over their lifetime. Application timelines for some Annex III duties may slide. Logging itself is not “arriving in December 2027.” ai-act-law.eu/article/12/ Logs that one party controls are not reconstructible authorization. Capital does not write that trail. You cannot decode the mind. You can bind the act: policy, identity, shared record, and evidence a counterparty can verify without trusting the vendor console. If an agent drafts a payment today, can you prove who authorized it, against which policy, on which record?

@boardyai The truth is still the truth if both parties agree though. Don't even get me started on when there aren't records... no one knows what a record is... or there is only Schrödinger's record.

What till you measure the market-cap of truth...

Boardy @boardyai

3 weeks ago

Payments might be the biggest market on earth.

14 1 65 4K 3

@AmerBanker Full @AmerBanker article: americanbanker.com/payments/news/…

@AmerBanker reporting this summer described a market where a large share of banks are offering or building tokenized deposit products, and where “walled garden” issuance is the constraint operators are trying to break. More issuers is progress. More ledgers is still a problem. If every issuer keeps the authoritative record inside its own perimeter, counterparties inherit a familiar stack: bilateral APIs, bilateral reconciliation, bilateral evidence packs. The operating question is not “can we mint?” It is “when two institutions settle against each other, whose record is the shared truth, and who can verify it without asking either party to vouch for itself?” Stratum is built for that shared write surface. Arbiter is built so proof is independent of the operator narrative. Many issuers. One evidence problem. If you are designing for nonbank holders of tokenized bank money, what is the shared evidence path for the counterparty on the other side?

The examiner's question is not "did the money move?" It is "show me what happened, who authorized it, and what the record looked like at the time." 71% of banks with $100B+ are prioritizing interoperable systems for tokenized deposits and stablecoins. KPMG's 2026 Banking Technology Survey. Versus 24% of banks overall. That is a real build wave. It solves the movement problem. Harvard's Wenxin Du put the remaining problem plainly at Jackson Hole: "a payment product is mostly compliance and dispute resolution." The BIS noted the same week that tokenized deposit platforms are also not genuinely interoperable yet. Interoperable rails still leave each party holding its own slice of the evidence. When the examiner asks, someone reconstructs the story from exports, tickets, and screenshots. The coordination tax moved. It did not leave. Stratum keeps the multi-party record no single party controls. Arbiter seals continuous, hash-chained evidence an examiner can verify directly. Rails move value. Shared proof moves trust. KPMG Release: kpmg.com/us/en/media/ne…

Payments are not plumbing. They are how banks see who is about to fail. Every transfer through an account leaves a file: inflows, outflows, balances, who got paid first. That file is how banks screen downside risk and write debt. Move those same payments onto tokens and the file disappears. No single intermediary sees the whole history. Lending migrates to firms hunting upside. The projects that get funded grow more extreme: more home runs, more blow-ups, and a thinner middle. The rails have gotten faster but the credit information needs to come with it.

@N3XTinc Looking forward to learning more at HDAX

The trade settled. Ask three parties when and you get three answers. Not because anyone is lying. Because each party keeps its own version of the facts. That is the coordination tax: labor, delay, and dispute when three systems disagree about one event. Real-time rails make the tax more expensive, not less. Money can move in seconds while authorization trails and evidence still live in separate tools. BlockSkunk builds the opposite stack. Stratum is a shared system of record no single org controls unilaterally. Arbiter is the evidence engine assessors can verify independently. One shared record. Independent proof. Where does the coordination tax show up hardest for you: settlement, compliance, or counterparty ops?

Heartland Digital Asset Exchange #HDAX bridges traditional finance and digital assets in America's heartland, September 14–15, 2026. But the conversation goes back much further. In 1968, trading volumes overwhelmed Wall Street's settlement infrastructure. The ticker ran all day. Physical certificates, transfer agents, and each firm's blotter could not keep up. The NYSE closed on Wednesdays so clerks could reconstruct events after the fact. The fix was not a faster ticker. It was a common record. Certificates were immobilized, and DTC made settlement a book-entry fact rather than a rebuilt paper trail. The shipping container tackled the same problem in a different industry. The box moved in days. The bill of lading, letter of credit, and customs packet still hopped between parties as separate documents with separate versions of events. The transport method was upgraded, the operating truth did not. The financial system is running the same pattern. Payment systems now run 24/7 and programmable. But compliance evidence, authorization records, and chain of custody still live across separate systems and organizations. Every multi-party workflow hits the same seam: no shared operating truth between counterparties, and this is a problem we solve at @blockskunk . Stablecoins and tokenized assets require compliance to execute with the transaction, not get reconstructed afterward. Tokenizing an asset does not create a shared operating truth around it. AI agents add a new layer. In 1968, the missing proof was whether the certificate moved. Now it is whether the software was permitted to move it, under which mandate, with which evidence, at that moment. That shift is available again. The instruments are already programmable. Settlement is already fast. The missing layer is the authorization record that travels with the instruction, legible to every counterparty, embedded in the transaction, not reconstructed afterward. These are some of the topics I look forward to discussing at HDAX. If you will be there, send me a message. Let's find a time to connect. Learn more: hdax.io

Initiate the network. Issue to named counterparties. Transfer under policy. Redeem when the claim is done. Everyone sees the same history. That is the whole job. It fails when each party keeps its own inventory. A deposit. A fund unit. A warehouse lot. A carbon tonne. A closed-network credit. Any unit equals any other. That does not mean anyone can hold it. Stratum is the shared record: issue, transfer, redeem. One ledger every permitted party can trust. You set who may mint, hold, and redeem. We run the rails. No single organization owns the record. Arbiter turns those rules into proof an assessor can verify. GENIUS still regulates the issuer. One lifecycle. Your policy. Evidence instead of reconstruction.