Suyay

In physics, removing a step from a system doesn't just save time, it removes a place where error accumulates. That's basically the pitch behind 21X's license, and why Dusk plugging into it changes more than it sounds like.

I made myself a coffee, started researching, and discovered that 21X is the first company in the EU licensed under the DLT Pilot Regime to operate trading and settlement as a combined system—a DLT-TSS.
Historically those are two separate regulated functions, a trading venue and a central securities depository, with a gap between them where a trade sits before it's actually final. 21X collapses that into one atomic on-chain step. Dusk joined as a trade participant, and unlike most regulated venues that stay on private, permissioned chains, 21X runs on public, permissionless networks, first Polygon, then Stellar, with an integration into DuskEVM specifically named as the next one.

A stated goal of this specific collaboration is stablecoin treasury management: an issuer buying and selling tokenized money market funds through 21X to manage its reserves, settled atomically instead of routed through a separate depository days later. That's not an abstract use case for Dusk, it's the exact plumbing EURQ would run through.

What I keep sitting with: the DLT-TSS isn't a permanent law, it's an exemption under a pilot regime with a defined trial window. What happens to everything built on top of it if that regime doesn't get made permanent isn't something either side controls.

Oops, my coffee got cold.

#dusk $DUSK @Dusk

Reviewing today's data, August 21, 2026, there is a significant shift regarding the feeling of an “ugly market” we had been experiencing: the market is not currently in a phase of widespread decline; it's in a very strong recovery, but I still don't consider it confirmed as a new bullish leg of the cycle.

Bitcoin is around US$ 77,000–79,000, after a rally close to 20–24% this week and hitting three-month highs. Ethereum is around US$ 2,400, $SOL near US$ 90–91, and $BNB around US$ 670.

The interesting thing is that this rally has a lot more substance than a mere technical bounce.

Spot Bitcoin ETFs accumulated approximately US$1.6 billion in inflows during the first four days of the week, including about US$606 million on Thursday. At the same time, over US$ 4 billion in short positions were liquidated in two days. That explains part of the rally's speed: we have real money coming in, but also a massive amount of short sellers forced to buy back.

That creates a situation that I like, but which demands caution: liquidity + ETFs + short squeeze + regulatory improvement = favorable scenario, but... 20%+ rally in a few days + massive liquidations = elevated risk of a correction.

That's why I wouldn't chase a coin right now that has already gone up 30%, 50%, 100% or more.

What worries me most is that Bitcoin still concentrates roughly 60.6% of the crypto market capitalization. And the Altcoin Season Index is at 37, very far from the level normally used to talk about a true altseason.

What do you guys think?

Just my thoughts, not investment advice!!

@TermMax 's one-click Rollover is a small feature with a bigger implication than it looks like at first: it means a fixed-term position doesn't have to end just because its maturity date arrives.

Before this, a maturing fixed-rate loan on TermMax gave you two options — repay in full, or let the position lapse into whatever comes next. Rollover adds a third: move the position directly into a later-maturity market in a single transaction, extending the fixed term without a full repayment step in between and without defaulting into floating exposure you specifically avoided by choosing fixed terms in the first place.

That only works because fixed rates and fixed maturities are real, enforced terms on TermMax rather than a rough estimate. Traditional bond markets have handled duration management like this for decades through refinancing; most of DeFi still treats a loan's maturity as a wall rather than a decision point, mostly because floating-rate systems don't have a fixed maturity to manage in the first place.

I don't think one rollover feature turns DeFi into a fixed-income market on its own — that takes years of instruments building on each other, and I haven't seen how Rollover behaves yet when someone tries to chain multiple extensions back to back.

Does a feature like this end up mattering more to long-term lenders managing duration, or to borrowers just trying to avoid a floating rate they don't want?

#termmax @TermMax $TMX

Tutoring incoming students, the hardest part is always keeping two related ideas from blurring into one in someone's head. Reading through Dusk's privacy stack this week gave me the same problem.

Citadel and Dusk's confidential transaction layer solve different questions that sound like the same question. Citadel is about proving who you are: age, residency, accreditation, issued and verified through a self-sovereign identity toolkit, so you can prove you qualify for something without handing over your full ID. The transaction layer, through Moonlight and Phoenix, is about proving what you did was compliant: ownership limits, transfer eligibility, verified without exposing the raw transaction data to the public.

Put together, that's the actual claim behind "programmable privacy": one proof for identity, a separate proof for behavior, neither one requiring the other to go fully public to satisfy a regulator who's actually authorized to check.

What I keep turning over: identity requirements aren't the same across jurisdictions, what counts as sufficient proof of residency or accreditation in one regulatory regime isn't automatically what another one asks for. I haven't seen how flexible Citadel's license issuance actually is when the attribute a regulator wants isn't one of the ones already defined. That's less a privacy question and more a "how many regulators can this actually serve at once" question.

#dusk $DUSK @Dusk

@TermMax 's numbers are worth sitting with before anyone forms an opinion on $TMX from a price chart alone. TVL has crossed $100M across 8 chains, and total users who've interacted with the protocol have passed 1.1 million. In March, TermMax ranked second by daily active addresses among every DeFi lending protocol tracked on Token Terminal — not a minor placement in a long list, second overall in the category.

TVL alone doesn't tell you much; plenty of points campaigns inflate it temporarily. DAU ranking that high is harder to manufacture, because it requires wallets actually opening and closing fixed-rate positions week over week, not just capital sitting idle for an airdrop. TermMax also came out of the YZi Labs Residency program in May, which functions as a filter applied before the fact, not a claim the project makes about itself after.

I'd still treat one month's DAU ranking as a snapshot and not a permanent state — DeFi usage rotates fast in both directions, and I don't have a clear read yet on how much of that 1.1M user count is active this month versus cumulative since the protocol's earliest days.

Is anyone tracking whether that active-user number is climbing, holding, or already cooling off since March?

#termmax @TermMax

Somewhere between two physics problem sets this week I went down a rabbit hole on how Dusk's bridge actually confirms a lock happened before it mints anything on the other side.

On May 20, MAP Protocol's Butter Bridge minted 1 quadrillion MAPO out of nowhere, about 4.8 million times the real supply, after an attacker spoofed a cross-chain message through a flaw in the OmniServiceProxy contract. The bridge never verified the deposit actually happened, it just trusted the message saying it did. The token dropped nearly 30% before the dust settled.

That's the exact failure mode every lock-and-mint bridge is built to avoid, and it's also the pattern under Dusk's own ERC20/BEP20-to-native migration: lock the tokens on Ethereum or BSC, issue native DUSK once the lock event confirms. Zellic audited that migration contract specifically and found zero vulnerabilities, every branch of the mint function tested against exactly this class of forgery. Dusk's newer bridge, connecting DuskDS to DuskEVM, goes further and skips the pattern entirely: it's validator-run and trustless, no wrapped assets or locked reserves sitting there as a target at all.

What I'd still want spelled out: an audit that found nothing in 2024 isn't a permanent guarantee against a new class of forgery in 2026, and the older ERC20/BEP20 migration bridge is still architecturally the lock-and-mint shape that keeps getting broken elsewhere. Not every bridge Dusk runs carries the same risk profile, and that distinction matters more than "audited" as a blanket label.

#dusk $DUSK @Dusk #Ethereum #BSC

I've been reading through how TermMax handles liquidations, because most protocols only explain their liquidation logic in the docs nobody reads until it's their position getting closed out.

The setup is different from the standard on-chain auction model. When a Gearing Token's collateral value falls below its debt threshold, TermMax doesn't just force-sell into whatever liquidity happens to exist at that moment — it supports physical delivery, where the collateral itself can be delivered directly to settle the debt instead of routing everything through an open-market sale. That matters most for the assets TermMax is built to support beyond blue-chip crypto — real-world assets and lower-liquidity tokens, the exact collateral types that get destroyed in price during a forced sale on a thin order book.

What I still want to see is how this performs during a multi-asset stress event, not a single isolated liquidation. Physical delivery solves the thin-liquidity problem for one position at a time, but I haven't found data yet on how the mechanism holds up when a large number of GTs need unwinding simultaneously across different collateral types at once.

#termmax @TermMax $TMX

Woke up to the Coldcard story all over my feed this morning, a reported $100M+ hardware wallet exploit, and went straight to how Dusk actually handles institutional custody.

The Coldcard breach, disclosed August 6, is being read less as a storage failure and more as a governance one: one device, one employee, one system able to move assets alone. That's the gap independent key generation, multi-party approval, and audit trails are supposed to close, and it's exactly the design question sitting under every custody claim right now.

Dusk routes institutional custody through Cordial Systems, running Dusk Vault on Cordial's self-hosted treasury infrastructure, not a single hardware key someone can walk off with. This isn't a pilot integration either. Cordial already secures over $20 billion in private credit originated on-chain through Figure Markets, and NPEX uses this same custody layer as an actual client, not just a technical partner name-dropped in a deck.

What I still want to see spelled out: the Coldcard story worked because "multi-party approval" sounds solid until you ask exactly how many parties, how independent they really are, and who can override the threshold in an emergency. I haven't found Dusk or Cordial publishing that structure for Dusk Vault specifically. Scale and an existing client are real signals, they're just not the same thing as seeing the governance model itself.

#dusk $DUSK @Dusk

I've been trying to figure out how TermMax actually replaces a multi-step leverage loop, because most "one-click leverage" claims in DeFi turn out to be a nicer front-end wrapped around the same four or five transactions running underneath.

What happens on TermMax is different: locking collateral mints two tokens at once — a Fixed-Rate Token representing the debt taken on, and a Gearing Token, an NFT that records the exact collateral and debt of that specific position, capped by the market's maximum loan-to-value ratio. Selling the FT at its market discount is what actually delivers the leveraged exposure. One mint, one sale, instead of loop-borrow-swap-redeposit repeated across separate protocols, with separate gas costs and separate points of failure at every step.

What I'm still working through is how that Gearing Token behaves under real stress. An NFT carrying both collateral and debt as one object is a different liquidation surface than a normal lending position, and I haven't found enough detail yet on how MLTV enforcement holds up when a large number of GTs need unwinding at the same time, not just one at a time.
@TermMax #TermMax #termmax

Reading about Dusk building EURQ, its regulated euro, together with Quantoz this week put me right back at a summer job in a small print shop. A sign taped above the safe read "two signatures required for any withdrawal." The lock underneath only needed one key, and three copies of that key were floating around the shop. One week the till came up short, someone had opened it alone. Nothing about the paperwork was wrong. The rule existed. It just wasn't built into the lock.

Call it the paperwork peg: a promise written into policy that the mechanism underneath never actually enforces.

On May 24, that exact gap took down a regulated euro stablecoin. StablR, a Malta-licensed, MiCA-compliant issuer, ran its minting contract on a 1-of-3 multisig, meaning any single key alone could authorize new supply. One key got compromised. The attacker added themselves as an owner, removed the two legitimate signers, and minted 8.35 million USDR and 4.5 million EURR out of thin air, about $13.5 million at face value. EURR fell to $0.85, USDR crashed as low as $0.40. StablR had every MiCA box checked, reserve requirements, redemption rights, monthly proof-of-reserves. None of those boxes covered how many keys it took to mint.

EURQ sits in the same regulatory category, an Electronic Money Token issued by Quantoz under MiCA, built into Dusk as the settlement currency for Dusk Pay's regulated payment rail. Dusk's own transaction layer enforces balance integrity and ownership cryptographically on every transfer, through Moonlight or Phoenix, not through a policy document.

What I don't have visibility into yet: whether Quantoz's own minting and custody setup for EURQ, the multisig threshold, hardware security modules, time-locks, has actually been disclosed or audited against exactly the failure that hit StablR. A MiCA license didn't stop that one. The question isn't whether Dusk's chain is sound, it's whether the key management sitting above it is.

#dusk $DUSK @Dusk

I remember locking funds into a DeFi lending pool because the APY on the screen looked good the day I deposited. Three weeks later, I logged back in to check my earnings and the rate had moved twice, both times in the wrong direction. Nothing had "gone wrong" — that's just how floating rates work. But I hadn't actually agreed to that number. I'd agreed to whatever the pool decided later.

I started calling that the moving target problem: you commit capital today, and the actual terms of the deal keep changing after you've already said yes. It's treated as normal in DeFi, the same way it would never be treated as normal for a mortgage or a bond.

What pulled me toward @TermMax was seeing fixed rates and fixed terms applied to that exact situation — you lend or borrow at a rate set the moment you enter, and it stays that rate until the maturity date you agreed to, full stop. No renegotiation mid-position because the pool's utilization shifted overnight.

I still think fixed rates trade some upside for that certainty, and in a strong bull run a floating position could outperform. But the number I see on day one is finally the number I get on the day that matters.
#TermMax #termmax @TermMax

#dusk $DUSK @Dusk
Reading about Dusk's Chainlink CCIP integration this week put me right back at a reception desk I temped years ago, the one with a badge scanner at the loading door. Courier badge beeps green, door opens, nobody checks the photo against the face holding it. Just the beep.

One week someone walked out with an outgoing shipment, badge beeping green under a name that wasn't theirs. Security footage caught it eventually. The scanner had done exactly what it was built to do: trust the beep, not the person.

Call it the trusted beep: a system that verifies the signal saying something is true, never the thing itself. Most cross-chain bridges run on the same shape, and it's the gap Dusk is choosing not to leave open. One chain sends a message claiming funds are locked, the other pays out the moment it hears that message land clean.

On August 9, someone drained roughly 199,916 XRP, about $200,000, from the bridge connecting the $XRP Ledger to the tx chain. No private key touched, no contract bug on either ledger. The attacker forged a deposit-verification message in the relayer software, and it paid out through 94 withdrawals in 97 minutes, all because the message beeped green.

This is the single-relayer setup Dusk skips with the Chainlink CCIP integration it's building alongside NPEX. Multiple independent oracle networks reach consensus before a cross-chain action executes, backed by a separate Risk Management Network watching for a pattern like 94 rapid withdrawals. For $DUSK itself, the CCT standard burns the token on the sending chain and mints on the receiving one, so no locked liquidity pool sits there as a badge scanner waiting to be spoofed.

What doesn't disappear even with Dusk's setup: more parties checking it isn't the same as nobody being able to forge it. It shifts the risk from one relayer's code to whether enough of those networks could ever be compromised or collude at once, a different shape of the same trust question.

I still think about that badge, beeping green for a face it never looked at. @Dusk

3 a.m. Pacific, January 28 2021. Robinhood's ops team gets a letter from the NSCC, the same daily letter about 100 brokerages receive, except this one asks for 3 billion dollars in collateral by the next morning. Nothing about GameStop's fundamentals changed overnight. What changed was the size of the gap the clearinghouse had to cover: trades settle two days after they're placed, and during that window someone has to guarantee the money shows up. With GME's volatility spiking, that guarantee got expensive fast. Robinhood didn't have 3 billion sitting around, so it did the only thing left, it turned off the buy button on the stocks driving the exposure.

Call it the T+2 tax. Every trade on a legacy exchange carries an invisible toll for the two days between "I bought this" and "I actually own this," and when volatility spikes, someone has to front that toll in cash, collateral, or access. Retail investors found out the hard way that the toll gets paid by taking the button away from them.

This is the gap Dusk Trade is built to close on the settlement side. Because the underlying chain settles with deterministic finality in seconds through Dusk's Succinct Attestation consensus, a delivery-versus-payment trade doesn't sit exposed for two days waiting to fail. Dusk Trade is already onboarding real products under that model, BlackRock's ICS Euro, Sterling and US Treasury government liquidity funds and MembersCap Fund I, through standard KYC, AML and GDPR-compliant onboarding rather than a crypto-native signup flow.

What I'm not certain closes cleanly: faster settlement removes the technical reason for a collateral call like NSCC's, but clearinghouses and regulators price risk using rules built for the T+2 world. Whether that regulatory capital treatment actually catches up to near-instant settlement, or whether the exposure just gets recalculated somewhere else in the system, isn't something the technology alone decides.

#dusk $DUSK @Dusk

May 19, 2020. Overstock's tZERO airdrops 4.37 million OSTKO tokens to shareholders, one for every ten shares held. The stated reason, straight from the company, was to bolster liquidity on its own security token exchange. That's worth sitting with: tZERO had filed with the SEC as far back as 2015 to run a blockchain-based trading system, issued the world's first SEC-registered blockchain security in 2016, and by 2020 was still handing out free tokens because the venue itself couldn't generate enough trading on its own. Today those tokens still trade mainly through specific broker-dealer channels and the OTC grey market, not in the open.

Call it the licensed empty room. You can build the most compliant rails in the industry and still end up with nobody in the building, because a license proves you're allowed to trade, not that anyone will.

NPEX solves the other half of that equation before Dusk even enters the picture. It's not a startup applying for permission, it's a Dutch exchange already regulated by the AFM, holding an MTF license, a Broker license, and an ECSP license, with a DLT-TSS license in progress. It has already financed over 200 million euros for more than 100 SMEs and connects a network of over 17,500 active investors, a room that was already full before anyone mentioned blockchain. The stated plan is to bring 300 million euros of that activity on-chain through Dusk. As Dusk's CEO put it, other RWA protocols are competing for shelf space, Dusk is becoming the structure that houses the collection.

What I'm not fully sold on yet: NPEX is a mid-size SME exchange, not Euronext or the LSE. The model of attaching regulated infrastructure to an existing investor base clearly beats building a compliant venue and hoping people show up, but whether a larger, established exchange chooses to plug into someone else's chain instead of building its own remains the open question here, not the technology.

#dusk $DUSK @Dusk

November 2022. Someone holding FTX's tokenized Tesla, GME, or Apple "stock" opens the fine print for the first time, because the exchange is going bankrupt and they need to know what they actually own. The answer is in CM-Equity AG's key information document: these were bilateral OTC derivative contracts tracking a share price, cash-settled only. No claim to delivery of the underlying. No shareholder rights, no voting, nothing. You didn't own a fraction of Tesla. You owned a promise from a counterparty that just went bankrupt, sitting in the same queue as every other unsecured claim.

That's the ownership mirage: a token that looks like the asset, trades like the asset, and legally is not the asset. It happens whenever tokenization wraps a security instead of replacing the process that creates one. The chain gets a new interface. Custody, registry, and settlement stay exactly where they were, off-chain, with the same people who can freeze, default, or go bankrupt.

Native issuance is the distinction @Dusk keeps drawing, and it's a narrower claim than it sounds. Instead of a token representing an asset held somewhere else, the asset itself is created and managed on-chain, issuance, transfer, and settlement running through DuskDS with deterministic finality. Dusk Trade, built with NPEX (an AFM-regulated Dutch exchange), is where that becomes a workflow: eligibility checks, DvP-capable settlement where the asset and the payment leg move together, disclosure paths built in instead of bolted on. If it settles on-chain, it isn't waiting on a bankrupt intermediary's fine print to tell you what you actually hold.

What I don't think the architecture alone answers: native issuance still needs a jurisdiction willing to say that on-chain settlement is legal ownership, not just a technical record of one. NPEX has that under the EU's DLT Pilot Regime. Whether that recognition travels beyond a handful of regulated venues is a legal question, not a cryptographic one.
#dusk $DUSK @Dusk

Block 22450093, May 10 2025. Someone submits a straightforward ETH-to-RATO swap on Uniswap. Position 10 in the block: 1.58 ETH going in. Position 11: another wallet buys first, same pair, paying more gas to jump the queue. Position 12: that same wallet sells right back into the pool it just pushed up. The original trader ends up with roughly 10 million fewer RATO than they should have gotten. Nobody hacked anything. The mempool simply showed everyone the trade before it settled, and someone acted on what they saw.

I keep coming back to that one block because it isn't a headline hack, it's the network doing exactly what it was built to do. Public mempool, public intent, public consequence. Call it the glass ledger problem: any institution that wants to move real size onchain is being asked to trade in a room with the lights on and the blinds torn off.

That's the wall @Dusk is building against with Hedger, the privacy engine landing on DuskEVM. Unlike Zedger's UTXO-only design, Hedger combines homomorphic encryption with zero-knowledge proofs directly at the EVM layer, so balances, amounts and counterparties stay encrypted through execution, with selective disclosure available to whoever is actually authorized to check, an auditor, a regulator, a counterparty. Same Solidity tooling, same EVM wallets, confidential by default instead of confidential by accident.

Here's what I haven't seen answered yet. DuskEVM runs as a rollup, which means a sequencer orders transactions before they're batched to DuskDS. Encrypting the payload hides it from the public mempool that produced block 22450093, but what does the sequencer itself see before ordering, and who checks that this seat isn't just a smaller, more private version of the same problem. Encrypted for us isn't automatically encrypted for whoever holds the queue. Worth watching as DuskEVM mainnet gets closer.
#dusk $DUSK @Dusk

Fourteen posts ago I opened with my grandfather's watch — forty years in a safety deposit box, needing two keys that never shared a room. I didn't know yet if TBV actually closed that gap. I only knew it was the first design I'd seen that tried.

What I found across the rest of it: a self-claim mechanism that lets a depositor pull their own BTC back with a signature committed the day the vault opened, no permission required. A Security Council that can freeze a bad claim but has no address of its own to receive BTC — compromise it and you lose recovery capacity, not custody. And, this month, my own 0.01 sBTC coming back into my wallet at the three-day mark the withdrawal screen promised. Not close to it. Exactly it. Vaults bound to one application at creation, so a bug in one integration can't reach into another. Collateral that can't be rehypothecated, lent out, or repurposed by anyone, structurally, not by policy.

Not everything held up clean. TBV's liquidation still depends on a price oracle, the same category of risk that gutted Balance Protocol this summer. BABE's headline speed only applies in the honest-setup case; real malicious-security protection costs more. Whether a BTC-backed stablecoin even qualifies under rules regulators still haven't finished writing is genuinely unresolved. I'd rather close this out with those still on the table than pretend I found none.

I filled out Babylon's own testnet feedback form this week, every answer honest — nothing in two weeks of trying to find where this breaks gave me a reason to flag anything as broken.

I still have that image of the watch in my head. What's different now is I know what would need to be true for the design to actually fail — and for fourteen days, running my own BTC through it, I haven't found where it does.

@BabylonLabs_io $BABY #baby

A valet company near my old office got caught years back lending parked cars to a delivery service during slow hours. Sued, reformed, reopened with new signage: insured, monitored, professional. This time I read the contract — buried in it, they can still lend your car to a "partner fleet" while you're inside, logged and mileage-split. Not the same scandal. Same car, still leaving without you, just initialed.

Call it the disclosed loophole: the exact practice that caused the trouble, kept intact, wrapped in enough paperwork that it stops being a scandal and starts being a feature. Institutional Bitcoin-backed lending is going through its own version of this right now.

Silicon Valley Bank called it a "new institutional era" for Bitcoin-backed lending in a June report — crypto-backed loans hit $67B in Q1, up 49% year over year. The report credits stronger risk controls, learned from Celsius, BlockFi, and Genesis collapsing in 2022 over exactly this: rehypothecating customer collateral. The industry's own materials still describe rehypothecation — lending pledged BTC to institutional borrowers to subsidize your rate — as standard today.

TBV's docs rule this out structurally, not by policy: "the collateral cannot be transferred, rehypothecated, lent out, or repurposed by any participant." Each vault is a single pre-signed Bitcoin output — no internal balance sheet for anyone to route it through, disclosed or not.

What I don't know: nobody's compared TBV's testnet rates against Ledn's or Strike's real production rates yet — no rehypothecation doesn't automatically mean a competitive cost. Ledn's own research found an 88%-to-14% gap between crypto holders who'd consider a loan and who actually have one. Trust might be the bigger blocker, but nobody borrows on trust alone.

Still don't know what became of that valet's mileage-split program. I stopped using the garage — maybe that's the real answer to whether disclosed is different from safe.

Comparing my own testnet numbers against what Ledn and Strike publish this week.

@BabylonLabs_io $BABY #baby

My old building used magnetic key fobs — same encoding scheme since it was built. When a hobbyist proved he could clone one with a $40 reader, the board wasn't worried about the fobs already out there. They were worried about getting every resident to show up the same weekend to swap them — half the building doesn't read the mailing list, the other half didn't want to change anything that still worked.

Call it the whole-building swap: a fix that exists, but only matters once everyone with the old lock adopts the new one — a coordination problem, not a cryptography problem. Bitcoin has its own version now. In March, Google's Quantum AI team cut the estimated resources needed to break Bitcoin's signature scheme by roughly 20x — from about 9 million qubits to under 500,000. Current hardware isn't close: IBM's best chip runs 156 qubits. But over a third of circulating BTC already has its public key exposed on-chain, reachable once a capable machine exists. Bitcoin's own quantum-resistant address proposal, BIP-360, has sat on testnet since February — shipping it means getting a consensus-driven network to agree on timing. Same fob problem, planetary scale.

TBV's dispute path never had to have that argument. The signatures securing Assert and ChallengeAssert — the transactions that make a claim binding — are Lamport signatures: hash-based, one-time-use, already resistant to the attack threatening the elliptic-curve signatures securing most of Bitcoin. It wasn't built as a quantum fix — it's what a Groth16 verifier built for Bitcoin script happened to need.

What I don't know: TBV's own BTC custody still settles through Bitcoin's regular Taproot outputs — the same elliptic-curve signatures everyone else is exposed on. Being quantum-resistant in the dispute layer doesn't make the underlying Bitcoin quantum-resistant. The newer lock still sits inside the older building.

Not sure if this matters yet to anyone who isn't already deep in this. Feels like the kind of thing worth answering before it needs to be, not after.

@BabylonLabs_io $BABY #baby

At first I assumed BABE's efficiency claim over BitVM3 was an incremental engineering win — a leaner garbled circuit, nothing structural. Sitting with the actual numbers changed that.

BitVM3's Groth16 verifier garbled circuit runs 40.5 GiB. BABE's version of the same job: 22.2 MiB. That's not a smaller file, it's a different category of thing — 40.5 gigabytes needs a machine built to store and garble it; 22.2 megabytes fits on hardware nobody would call specialized. Setup time drops from 353.7 seconds to 174.9 milliseconds, decryption from 352.1 seconds to 126.5 milliseconds. Whoever runs the Verifier side of a BitVM-style dispute doesn't need to be an infrastructure operator anymore.

That's the part I keep turning over: BitVM3's cost effectively selects for who can afford to Verify at all. Shrink the hardware bar by three orders of magnitude and more people can credibly do it themselves, instead of trusting whichever few operators had the storage to bother.

Here's the honest caveat, buried a few sections into the paper: those numbers are the honest-setup case. Real security against a dishonest Prover or Verifier needs cut-and-choose — multiple redundant instances garbled and checked against each other. At the parameters BABE actually uses for malicious security, setup climbs to around 49 seconds, and an optional step to collapse the on-chain footprint back down adds another 12 minutes on top of that. The eye-catching 2,022x still holds for the core primitive. It's just not the number you get once someone's actually trying to cheat.

I don't know yet whether 49 seconds plus an optional 12-minute step is "cheap enough" the same way $37.82 was cheap enough to replace trusting a bridge. It's a different kind of cost — setup time instead of on-chain fees — and I haven't seen anyone argue convincingly which one matters more once this runs at real scale.

@BabylonLabs_io $BABY #baby