A retention policy is a promise you can't audit from the outside. Trezor's partners agreed to 90 days. The records that came loose this week reach back to 2019, and another 67,000 customers are in the pile, which tells me the retention number was decorative from the start. Seven years of names sitting in some vendor's support tool because deleting is work and hoarding is free.
Nothing leaked that touches a key. That line gets repeated and it's true and it misses the whole thing. What leaked is the map: who bought the box, where it shipped, roughly when they got serious about holding their own coins. I lived through the Ledger list going out. I watched the extortion emails start, watched a guy I've followed since 2017 move apartments over it. The threat model of self-custody was never the math. It's that ordering a hardware wallet writes your address into a database next to an implicit note saying this one probably has money, held by a company whose incentive to forget you is zero. 🔐
The other half of that story moved the same days. 20.45 BTC out of the Coldcard theft, about $1.6M, pushed through 34 swaps across the 2nd and 3rd, mostly into Ethereum, with the main destination later sitting roughly 5 ETH lighter. What I keep circling is the parking. The coins sat. Someone waited out the attention span of the chain-watchers, then moved with gas to spare and a destination that could actually absorb size. Panic-movers get caught. Patient ones get a spreadsheet. The professionalization of theft has been the most consistent trend in this space since 2017, and it never makes the year-in-review posts.
Meanwhile, the boring rail stuff, which is where the actual money is. Kraken listing SoFiUSD, a dollar token from a chartered bank, plus settlement membership so cash moves between them as tokens instead of crawling through an ACH batch. Read the structure and it's beautiful in a cold way. You hand over a dollar, SoFi holds it in reserves and short Treasuries, earns the going rate, and GENIUS makes it illegal to pass you a cent of that. SoFi pays interest on a savings account because it has to compete for that deposit. The token version of the identical dollar is funding it gets for nothing. I expect a line of chartered banks queueing up with their own dollar tokens on every major exchange, and not one of them will pay you to hold it. The most profitable deposit ever invented is the one that legally cannot pay you. 🧾
Which is what makes the Fed staff paper land harder than it should have. Their point is that a stablecoin can break without losing a dollar of backing, because if the only reason to hold the thing is usefulness, and usefulness lives on whatever chain your counterparties settle on, then congestion is the run. Fees climb, moving your dollars starts costing real money, you redeem at par or bridge somewhere cheaper, and the issuer sees both as the same event: tickets stacking up inside an hour against reserves that never lost a cent. Peg risk left the balance sheet and moved into the plumbing. I actually think that's an upgrade. A bank's solvency stays invisible until the morning it isn't. Congestion prints on a public explorer while it's happening. The issuers who survive the first real jam will be the ones already native on several chains, not the ones scrambling to bridge mid-panic.
Then the macro slapped everything sideways. 162,000 payrolls in August against 53,000 expected, unemployment holding at 4.1%. That's not a soft landing, that's an economy refusing to cooperate with the cut narrative half my timeline spent the summer pricing in. The same week, Japan's yield touches 4%. I sat with that one for a while. The corporate bitcoin treasury trade was never really about conviction, it was about a funding cost close to nothing, and a lot of that came out of Japan. Existing fixed debt keeps its terms, fine, nothing explodes. The next BitBond has to clear a hurdle that didn't exist eighteen months ago. When financing stops being free, the marginal buyer doesn't sell, it just stops showing up, and that absence takes months to become visible in the tape. 📉
The legitimacy drift is the thread I find hardest to hold steady. Hargreaves Lansdown, the same platform that spent years warning its customers off this stuff, now opening the door to crypto ETNs after the FCA lifted the retail ban in October. The US sheriffs' association moving from opposing the Clarity Act to neutral, months after calling it cover for crypto crime. Underneath both, the FBI story: three warrants in 2025, more than $560,000 seized off a Hamas fundraising pipeline, and then agents running the captured systems to pull in more donations. That's the real inflection. Law enforcement figured out it doesn't want the pipe shut, it wants the pipe running with its hand on the valve. Transparency stopped being the industry's liability and became the state's favorite feature. I don't know how I feel about that, honestly. It kills the dumbest criticism of this technology and confirms the deepest one in the same motion.
Kalshi is the counterweight, getting hit with $500,000 a day in Michigan until a final order, sports contracts forced offline, an appellate split hanging over it. Federal warmth, state teeth. That's the shape of the next two years for anything that touches money and prediction at once, and I'd guess it decides more outcomes than any bill does.
What feels different from six months ago is the location of the risk. It's not "will the chain hold" anymore. The chains held all week. It's whose float you're funding for free, whose congestion you're exposed to, whose customer list you're sitting in, whose state attorney general woke up in a mood. All the old fears were about code failing. All of this week's were about institutions doing exactly what institutions do, at scale, with your data and your dollars and their charter. 🌊
The chain kept every promise it made this week. The only promises that broke were the ones written in English, by companies, about what they'd delete and when.