S15 E54: Kruw on Running a Wasabi Coordinator

When the company behind Wasabi Wallet shut down its coinjoin coordinator in 2024, a lot of people wrote the obituary for Bitcoin privacy. Kruw did something less dramatic and much more useful. He spun up a coordinator on his own node, charged zero fees, and within a few months it became the biggest Wasabi coordinator around. By the time we sat down for this livestream, he told me it had pushed "28,000" bitcoins through its rounds, at roughly 500 BTC per day including remixes.

That alone would justify an episode. But Kruw is also a tireless researcher, a part-time historian of Bitcoin privacy drama, and the kind of guy who shows up to an interview with a folder of receipts and a cat napping beside him the whole time. We planned a live experiment, he traced coinjoins on screen, and the conversation wandered for over three hours into covenants, Lightning, Tor exit relays and the block size debate. It is one of those episodes where the format is that there is no format, and I mean that as a compliment.

A Torch Nobody Wanted to Carry

Wasabi had become a recurring theme of Season 15. I opened the season with Max Hillebrand and Lando Rothbardian on the "Buy Anything" button, then covered the shutdown, then the project's future. What fascinates me is that the wallet survived without the company. Kruw pointed out that decentralized coordinators had already been shipped quietly in 2023, built by the BTCPay Server developer Kukks with Nostr as the discovery layer. "There was no big announcement or a pre-announcement of a pre-announcement," he said, in a not-so-subtle jab at Samourai's long-promised decentralized Whirlpool. Yet nobody cared, because "people hate or hated Wasabi so badly that this didn't get any traction."

The 2024 crackdown changed that. Kruw listed the casualties as if reading a war report: Samourai's developers arrested, Sparrow pulling the Whirlpool client, Bitcoin Keeper deleting every tweet that ever mentioned it, zkSNACKs closing its coordinator, Trezor dropping coinjoin support and Phoenix leaving the US. I reminded him of his own post saying the winner of the Samourai versus Wasabi war was the regulators. He agreed, and admitted Bitcoin "was a little bit weaker than people thought it was."

My own read is that Wasabi came out tougher. The Samourai camp was louder about dying for privacy, but when someone actually went to jail, most of its fans scattered toward Monero. The Wasabi community stayed and picked up the pieces, and Kruw is the most visible example.

Coinjoins Are the Worst Way to Use a Blockchain (and That Is Fine)

We started with a bet. I gave Kruw a fresh address and he promised to pay me directly from inside a coinjoin, a little-known WabiSabi trick that lets you do the premix, the mix and the payment in a single transaction. Samourai, by comparison, needed three steps. Then we would let the audience try to trace it.

While the round filled up, I pushed him on the obvious objections. Monero hides amounts; Bitcoin's ledger shows everything. Kruw explained that this is exactly why coinjoins produce equal-sized outputs: "if you have a 100K sat output, you can't tell it apart from someone else's 100K sat output." He was honest about the weak spot too, admitting the WabiSabi protocol is weaker "for the very largest whale on the transaction."

On speed, he had a line I loved. A coinjoin, he said, is "just like the worst version of transacting on a blockchain." Ten-minute blocks already rule out paying for groceries, so adding half an hour of mixing on top does not break anyone's use case. "There was no one in between 10 minutes and 40 minutes who is now no longer going to use Bitcoin." On fees, he argued you can even save a little, because the transaction overhead is shared across the round.

I brought up the dust problem. Every time fees drop, people are told to consolidate their UTXOs, which I called "the worst privacy advice ever," because one lazy consolidation can undo years of discipline on a permanent public ledger. Kruw's answer was Wasabi 2.0's payment builder, which pieces together standard denominations so you can pay "within a couple percentage points" of the amount and leave no change behind. He compared it to the branch and bound coin selection in Bitcoin Core: "a privacy tool and a blockchain efficiency tool."

The Coordinator That Charges Nothing

Wasabi's user experience got a lot of love in this episode. I raved about the one-click transaction boost, the rocket button that spares you from choosing between CPFP and RBF. Kruw praised it as abstraction done right, but he also confessed that the original 2.0 design went too far by hiding UTXOs entirely. "I think this was a mistake," he said, adding that nopara73 "will still defend it to this day."

Switching coordinators is now just a matter of pasting an address and a fee limit into the settings, and his own coordinator lives at kruw.io. Kruw loves the idea of a market of coordinators, each catering to a different user. His rounds run for an hour, which makes them big. zkSNACKs ran ten-minute rounds, and "two out of three of them would fail." He would like Wasabi to eventually pick a coordinator for you based on how much you want to pay and how long you want to wait, a bit like makers competing on JoinMarket.

So why zero fees? After zkSNACKs left, the new coordinators launched with matching fees, but running one costs maybe "$150 a month at the max." Kruw went for liquidity first, not money, and the first-mover advantage did the rest. Then he revealed the part most people miss. Because WabiSabi registers inputs before outputs, every participant leaves a few leftover satoshis behind, and a coordinator can collect them. "These little sats add up," he told me. "It's not tons and tons of bitcoins, but it'll pay for your server." He thinks that incentive could drive decentralization if more people knew about it.

He also let slip that Trezor's coinjoin code is still open source and could be pointed at another coordinator. He plans to test it on one of his many Trezors and write a guide, since cold-wallet coinjoins were always one of Wasabi's most underrated tricks.

Easy Mode, Medium Mode, Hard Mode

Then came the spicy part. I asked Kruw to share his screen and show how Samourai coinjoins get demixed. He pulled up an old OXT Research graph of a Whirlpool transaction: a premix split into five outputs, a change output and a Samourai fee output. The user later merged his change with all five mixed outputs, several of them on reused addresses, which drew a neat little circle on the graph.

Kruw was clear that this is user error, and that Samourai's own researchers would say so. His complaint was the double standard. "Samourai for some vicious reason every time this would happen, they would go out on Twitter and claim this means the implementation of their competitor was broken." Address reuse defeats everything: "No privacy tool, no Monero swapping, Lightning swapping, Whirlpool, nothing will protect you from that." He sorted tracing into tiers: reused addresses are easy mode, merged change is medium mode, and recombining the mixed amounts is hard mode.

He also went after the famous "132 arrests" talking point used against Wasabi. People who used Wasabi may have been arrested, he said, but claiming they were arrested because of Wasabi "is a leap of logic." Even F/Dev got a mention for asking Burak to say "coinjoin" instead of WabiSabi when crediting the protocol behind Ark. Fittingly, WabiSabi itself credits CashFusion from Bitcoin Cash, something I always enjoy pointing out.

I noted that privacy seems to always split into two camps: GrapheneOS versus CalyxOS, Zcash versus Monero, Wasabi versus Samourai. JoinMarket is the peaceful exception. Kruw called it "OG," the only place on Bitcoin where you can earn yield and privacy at once. "Count me in, that's awesome." He just prefers Wasabi for its privacy and block space efficiency.

"We're All Money Transmitters"

The question I most wanted to ask was the uncomfortable one. Kruw sounds American, admits he lives in "the territory of the Empire," and coordinates coinjoins at the same scale that got the Samourai developers charged. How is this safe?

He traced his answer back to 2020. Before Covid he was a number-go-up guy. Then he lost "not just one job but two jobs," the first to lockdowns and the second over the vaccine. "You start to realize that if you don't fight now, there's nothing left to fight for." On the Samourai charges, he refuses to play the legal-definitions game. "I'm a money transmitter, you're a money transmitter, we're all money transmitters. It's great, that's the whole purpose."

When I pressed him again, he did not pretend. "It's not" safe, he said. If he shuts down his coordinator, does he then shut down his Lightning node, and then his full node? "If you forfeit every piece of ground from here until self-custody, then you're guaranteed to lose the battle before it starts." He does not trust the Constitution to save him either, pointing to how gun rights have been chipped away. "Cypherpunks write code, and we build Bitcoin to be strong, we don't build it to be regulated."

I told him honestly that I worry about him. My previous guest was Charlie Shrem, who went to jail for treating bitcoin as fungible money. Kruw answered with a grim statistic about the US correctional system, then held up a physical Vlad coin and a Samourai Dojo node in a red case that cost him 0.1 BTC, as proof he puts his money where his mouth is.

Receipts on Samourai

Kruw is a Samourai critic who has actually contributed to Samourai, opening bug reports and buying its hardware. His main grievance was that Tor and Dojo were both off by default, which meant new users handed over their IP address and their xpub. "Nothing will matter if you hand them your xpub."

Then he scrolled through his notes. He showed old tweets where Samourai's team rejected block filters (BIP 157) with the claim that they were "a full node wallet" on phones, posts where Keonne Rodriguez warned against closed source wallets, and a later post welcoming a closed source Whirlpool client. He also showed Samourai saying it was not against Tor by default in principle, but worried about support costs, and then telling a contributor, ten days later, that it would ignore his merge request.

The sharpest point was Ricochet, the feature that hops your coins through several of your own addresses. Samourai charged 100,000 sats for it. "Does this sound like a service you should pay for? Not really." Worse, the fee output made those hops identifiable. The indictment, he noted, estimates about "$1.1 million for Ricochet transactions," which tells you the government could see them. Without the fee, those hops might have offered real deniability.

I admitted his note-keeping both impressed me and made me a little nervous. "That's my talent in this world," he said. "I'm a very vindictive person and I don't forget a lot of things." He says he applies the same standard to himself so that he argues "linearly and not in circles." For the record, he also refused to kick the Samourai developers while they are down: "they're victims, but not heroes." If you want to help them, he suggested, "go run your own coinjoin coordinator and just prove that you can't stop them."

Covenants, Ordinals and Old Pennies

The top of Kruw's Twitter profile at the time was a call to activate CTV, BIP 119. He admitted he wore a UASF hat during SegWit without really knowing what he was advocating, and supported Taproot mostly on the strength of his web of trust. CTV appeals to him because it is simple: "this is a pre-signed transaction, but it's enforced by consensus rules." His enthusiasm has cooled a bit only because the list of rival covenant proposals keeps growing. "Instead of true or false, you have A, B, C, D, E, and it's much harder to get a high score on that test."

I shared my view that CTV was the first serious proposal from outside the usual circle of developers, and that Adam Back did a masterclass in stalling it in the Cult of the Covenant Telegram group. Kruw credited Jeremy Rubin with doing "everything realistically possible" to prove its merits, from Sapio to his Advent calendar of blog posts, and called Speedy Trial a miscalculation, because a lack of pushback is not the same as support.

On ordinals, he was even more passionate. "The filter is the transaction fee," as I put it, and he agreed that the community's failure to stop inscriptions "is a success case for Bitcoin that proves how censorship resistant it is." Data has always lived on Bitcoin, from the Genesis block headline to the ASCII tribute to Len Sassaman.

He also dug up a 2017 Jimmy Song tweet saying coins signed by Charlie Lee would be worth more to him, much like collectible pennies. That gave me a chance to defend Litecoin's role in activating SegWit and my soft spot for MWEB, which I think is better than anything currently proposed for Bitcoin. When he asked me to rank privacy coins, I said Litecoin wins on liquidity, Monero on overall privacy and Zcash on cryptography.

How Blockchain Analysis Actually Works

Somewhere past the two-hour mark, Kruw gave what I called one of the best breakdowns of chain surveillance I have had on the show. He started with Roman Sterlingov, convicted as the operator of Bitcoin Fog partly on chain analysis evidence. Bitcoin Twitter called it pseudoscience. Kruw thinks that was the wrong defense: "start with the strongest argument: say this guy, whether he did it or not, let him go."

Then he went through his list. Address reuse comes first, which is why the whitepaper tells you to use a fresh address every time. Next is common input ownership, the heuristic that coinjoins and payjoins break. Then comes following the change, which usually matches the script type of the input. Round payment amounts give away the recipient, as do version numbers, locktimes and RBF flags.

Single-output transactions are where the guessing starts, because they could be a self-transfer or a spend. Add KYC data or a public Electrum server watching wallets sync, and "you can start to trace a huge amount of the Bitcoin network activity." Blocks are small, he reminded me, and "looking at 3,000 transactions or something is not that hard."

Why does he know all this? "I want to audit my own transactions," he said. "You don't want to find out later that you don't have privacy you thought you had."

Our live experiment, by the way, failed. Two hours in, the payment still had not arrived, and his logs said "overall balance must not be negative." It was a code problem, not a protocol one, but I made the point that if a power user cannot pull this off in two hours, it is hard to blame people for picking Monero or Zcash. "I can definitely see the appeal," he said.

Lightning Is Confidential, Not Private

It took us two and a half hours to say the word Lightning, which might be a record. Kruw's verdict was precise. The sender gets good privacy, but the receiver leaks unless they use blinded paths or BOLT 12, and an LSP sitting on both sides of a payment sees everything. "You only gain privacy on Lightning from extending the length of your route, which is not what Lightning is designed to do." He is a huge fan of Phoenix because splicing turns every channel change into something like a payjoin, "the coolest thing ever."

He was less kind to ecash. Fedimint, Cashu and friends are custodians either way, and he worries they will pull users out of non-custodial privacy tools rather than out of banks. He reminded me of a Mutiny federation that lost its DNS and left users unable to redeem their tokens.

I defended Nym, which I described as an incentivized mixnet inspired by David Chaum's work and a possible upgrade on Tor, whose exit nodes may well be run by agencies and embassies. Kruw, true to form, revealed that he runs a Tor exit relay himself. "You're a brave man," I told him. "You're fighting the fight, or else you're a fed."

Turning Into a Big Blocker

The last stretch was the most surprising, at least for me. I confessed that I am "turning into a big blocker." With Starlink and cheap storage, raising the block size is the most conservative way to add capacity, and the rest is politics. Maybe 16 or 32 MB, I said, enough to help millions onboard to coinjoins and layer twos without custodians.

Kruw was sympathetic but called it a slippery slope. "What if I say hey, I want eight?" He reminded me that Zcash had a long stretch of spam when block space was cheap, because "someone will do it just because they're an [expletive]." His benchmark for decentralization is running a full node on your phone, and he admitted he is a bit of a hypocrite there, since Wasabi itself uses block filters. I mentioned Amaury Séchet's speed limit analogy, and he argued that Ethereum is "what Bitcoin Cash would look like if Bitcoin Cash got adoption." I told him about the failed HTC Exodus 1s full node phone sitting in my drawer.

As we wrapped up, with his laptop battery dying and his cat still napping, Kruw added one more stat I had not expected. Under zkSNACKs, Wasabi was about four times bigger than Samourai in fresh on-chain volume. "There's some hidden bitcoiners out there with a lot of money who really like Wasabi, who are not very vocal on social media." That might be the best summary of this whole conversation: the loudest people in Bitcoin privacy were not the ones doing the work. If you want to see who is, connect to his coordinator, follow @Kruwed, and read my older essay on the limitations of Bitcoin privacy with his tracing tiers in mind.

Listen to Kruw on Spotify, Apple Podcasts, YouTube, and More

Watch on YouTube

Time stamps:

00:00:39 – Introduction
00:01:33 – Recap of Wasabi Wallet Discussions
00:02:28 – Samourai Wallet's Decentralization Announcement
00:03:56 – Challenges Faced by Privacy Projects
00:04:40 – Shift in Bitcoin Culture
00:05:32 – Impact of Regulatory Pressures
00:06:40 – Wasabi's Resilience
00:07:53 – Exploring Bitcoin's Privacy Features
00:08:36 – Demonstrating CoinJoin Transactions
00:10:24 – Differences Between CoinJoin and Monero
00:11:23 – Challenges with Bitcoin Transaction Privacy
00:13:11 – Concerns Over High Transaction Fees
00:17:40 – Wasabi's New Features for Privacy
00:20:14 – User-Friendly Features in Wasabi 2.0
00:21:18 – Transaction Speedup Options
00:23:05 – Critique of UTXO Management in Wasabi 2.0
00:23:38 – Wasabi Wallet User Control
00:24:03 – Switching Coordinators Feature
00:24:50 – Decentralized Coordinator Community
00:26:18 – User Preferences in Coordinator Selection
00:30:56 – Coordinator Fees Discussion
00:31:04 – Drivechains and Mining Concerns
00:34:25 – Market Dynamics of Coordinator Fees
00:37:43 – Collecting Dust from Transactions
00:38:09 – Trezor's Coinjoin
00:41:14 – Wasabi Wallet's Development Funding
00:42:36 – Samourai Wallet's Blockchain Analysis Tool
00:46:49 – Address Reuse and Privacy Challenges
00:49:40 – Mistakes in Privacy Usage
00:50:21 – Discrediting Wasabi Wallet
00:51:42 – Cash Fusion
00:53:23 – Reputation and Arrests
00:54:33 – Bitcoin Privacy Camps
00:55:56 – Join Market's Unique Position
00:56:21 – Legal Risks in Bitcoin Coordination
01:00:01 – The Fight for Bitcoin's Future
01:01:42 – Personal Ethics and Legal Protections
01:03:14 – Concerns Over Martyrdom
01:04:12 – Physical Tokens and Authenticity
01:05:10 – Critique of Samourai Wallet
01:08:09 – Transparency Issues in Samourai Wallet
01:09:55 – Closed Source vs. Open Source Debate
01:11:26 – Default Privacy Features in Samourai Wallet
01:13:13 – Ricochet Transactions and Fees
01:14:41 – Government Awareness of Transactions
01:15:01 – Implications of Fee Collection
01:15:53 – Personal Standards and Hypocrisy
01:16:21 – New Pepe Avatar Presentation
01:18:21 – Nostr vs. Twitter
01:20:11 – Support for CTV & Covenants
01:22:38 – Governance and Community Dynamics
01:22:53 – Concerns about ETF Privacy
01:24:08 – Complexity of Consensus Changes
01:28:20 – Community Understanding of Proposals
01:31:01 – Censorship Resistance in Bitcoin
01:32:46 – Market Dynamics and Filters
01:34:08 – Historical Context of Data on Bitcoin
01:35:31 – Criticism of Jimmy Song
01:36:30 – Litecoin's Role in Segwit Activation
01:37:47 – Bitcoin's Cultural Stagnation
01:39:45 – Ranking Privacy Coins
01:40:26 – Liquidity vs. Privacy
01:41:39 – Technological Advancements in Privacy
01:42:54 – User Resistance to New Technology
01:43:09 – 10101 Shutdown
01:44:16 – Drivechains and Altcoins
01:45:58 – Monero Liquidity Concerns
01:46:29 – Future of Bitcoin vs. Altcoins
01:47:20 – Cultural Conflicts in Bitcoin Development
01:48:26 – Blockstream's Influence
01:51:49 – Whale Bots in Bitcoin
01:52:52 – Michael Saylor's Influence
01:55:07 – Bitcoin's Future and Development
01:59:08 – Censorship Resistance vs. Privacy
02:00:05 – Satoshi's Coins and Community Reaction
02:01:08 – Blockchain Analysis Case
02:04:15 – Challenges of Privacy Services
02:05:20 – Tracking Methods in Blockchain
02:06:39 – Understanding Transaction Outputs
02:09:03 – Heuristics for Blockchain Analysis
02:10:06 – Personal Experience with Blockchain Analysis
02:10:58 – Samourai Wallet's Controversies
02:13:37 – Victims vs. Heroes in Privacy Projects
02:16:46 – Bitcoin Soft Forks for Privacy
02:18:18 – Statechains and Privacy
02:20:08 – Interactive Transactions Challenges
02:22:25 – Monero and Chainalysis
02:23:01 – Network Level Privacy Issues
02:24:01 – Future of Privacy Tools
02:24:49 – Zano's Hybrid Consensus Model
02:25:13 – Tail Emission Interest
02:26:11 – Consensus Mechanisms Explained
02:27:05 – Future of Bitcoin on Layer 2
02:28:28 – Lightning Network, Finally
02:29:24 – Confidentiality vs. Privacy
02:29:39 – Privacy Concerns in Lightning
02:30:12 – Splicing and Lightning Efficiency
02:32:40 – Public vs. Unannounced Lightning Channels
02:34:12 – Custodial Systems and Privacy
02:35:31 – Investment in Privacy Technologies
02:36:57 – Nym Security Token Overview
02:40:03 – Challenges of Decentralization
02:41:22 – Tor and Bitcoin Privacy
02:42:10 – Running a Tor Relay
02:48:55 – Ordinals & Spam
02:51:20 – Block Size Debate
02:52:41 – Challenges of Block Size Increases
02:57:48 – Ethereum vs. Bitcoin Cash
02:58:57 – Dynamic Block Size Mechanisms
03:00:10 – Running Full Nodes on Mobile
03:03:49 – Wrap Up and Future Prospects