Who are the money launderers in the Tornado Cash stack?

Over the last few years I've written a bunch of posts about Tornado Cash, an Ethereum-based mixing service, because I find it to be a fascinating tool. With the recent indictment of two people involved in the Tornado Cash "stack" for money laundering, here's another post to add to the list.

Let's get this clear from the outset. Somewhere in the Tornado Cash stack, someone is committing the crime of money laundering. That's been the case since at least mid-2020 or so, the moment that crooks started to send their criminally-derived ether proceeds to Tornado Cash for cleansing.

I'm going to repeat that. One of the parties (or groups of parties) woven together via the Tornado Cash apparatus has been knowingly acting as a financial counterparty to criminals, helping to "conduct" transactions that obfuscate dirty ether.

The question always was: who in the stack is guilty of money laundering? Is it the developers who are  laundering money? Miners? TORN token holders? Relayers? Licit users who engage with the smart contracts? And if so, are all licit users guilty, or just some users? Are the operators of the popular user interface the guilty parties?

A recent indictment from the U.S. Department of Justice claims to have figured out who the money launderers are.
 
Before getting to the indictment, let's tally up all the actors involved in the Tornado Cash stack. To begin with there are the users and developers. The central element of the Tornado Cash stack is a set of smart contracts, or pools, where users – both crooks and non-crooks – can send their easily-traced ether to be mixed, getting it back anonymized and untraceable. These core smart contracts were originally coded by three developers in 2019. In mid-2020, the developers removed the core contracts' upgradability, in effect "throwing away the keys" and ending their influence over them.

The next key set of actors are the relayers. Doing stuff on the Ethereum blockchain requires paying a fee to validators. The visibility of these fee payments effectively unwinds Tornado Cash's anonymity and reveals who Tornado Cash's users are. A group of third-party individuals, the relayers, are recruited to handle fees on behalf of users, thus restoring privacy.

The Tornado Cash stack also includes a popular user interface that acts as an overlay over the smart contracts, making them easier to interact with. Control over the user interface is delegated to individuals who own TORN tokens. TORN allows its owners to vote on how the front-end functions, in addition to earning profits from it. TORN holders have no influence over the core smart contracts.

Of these many actors, the DoJ has singled out Roman Storm and Roman Semenov, along with "others known and unknown," as the putative money launderers. (The government also accuses the two of failing to register as a money transmitter, but I'll set that aside.)

Storm and Semenov were the original developers of the core smart contracts, but that doesn't seem to be the nub of the DoJ's money laundering case. Rather, it is the accused's ongoing control over the user interface, exercised through their ownership of a large block of TORN tokens, that seems to have implicated them. Despite knowing that the Tornado Cash stack had become popular with criminals, the owners/operators of the user interface did nothing to screen bad actors from accessing said interface. On the contrary, they made efforts to both improve the interface and increase the profits they made from it.

The government's illustrates this by explaining the involvement of Storm and Semenov in managing the list of relayers that appear on the user interface, as well as in crafting the system for rewarding and levying fees on these relayers. The indictment cites a vote made by TORN holders in early 2022 that led to an update of the user interface's mechanism for listing relayers. The change allowed anyone to appear on the list, as long as they could stake a certain quantity of TORN tokens. The DoJ alleges that this decision improved anonymity by lengthening the user interface's list of relayers.  

The indictment further alleges that Storm and Semenov, through their ownership of TORN, profited financially from the user interface's new method of listing relayers. To get on the user interface list, a relayer had to buy TORN, which pushed up TORN's price. In addition, whenever a relayer that appeared on the user interface's list was selected, a portion of that relayer's staked TORN was "slashed," or reduced, forcing relayers to top up with additional TORN purchases in order to continue to qualify for the list. This added more upward pressure on TORN's price to the benefit of holders like Storm and Semenov.
 
In the government's view, the totality of these actions constitute money laundering, specifically a violation of  18 USC § 1956. The DoJ believes that the two defendants "conducted" transactions, a key element of money laundering, via their ongoing control over the user interface, along with other TORN owners. The indictment also shows that a large portion of Tornado Cash transactions were in fact criminal proceeds, including those made by the Lazarus Group. (I mean, we all knew that already.) Lastly, they show that the accused were aware that the funds coursing through the Tornado Cash stack were dirty, a mental state of knowing being a key plank in charging someone for money laundering.

It seems to me like the DoJ has a solid case, although we can debate whether operating the Tornado Cash user interface and its relayer list is tantamount to "conducting" transactions. The legal definition of conducts is a broad one, including "participating in initiating, or concluding a transaction." While the user interface, and thus those who operated it, never directly initiate transfers of ether to the underlying Tornado smart contracts, it doesn't seem a stretch to describe them as participating in the initiation of those transfers. We'll have to see what the judge says.

Counterintuitively, the indictment seems like a win, if only a lukewarm one, for fans of decentralized finance, or DeFi.

Proponents of DeFi have long worried that developers of autonomous smart contracts might be held liable in court for crimes. In this case, however, the same actors who happen to be the developers of Tornado's core smart contracts also built a complex and centralized business structure around those same contracts, and it is this tertiary apparatus that is serving as the basis for a money laundering charge, not the original coding of the core smart contracts.

It's a useful thought experiment to imagine how things might have played out if Storm and Semenov had acted differently. Let's imagine that the two coders hadn't created a profitable apparatus around the original smart contracts. Once the core smart contracts were up and running, they ceased to associate in any way with the Tornado Cash stack. Secondly, imagine there was no user interface. To deposit or withdraw funds, users had to interact directly with the smart contracts. Lastly, let's assume that TORN tokens had never been issued, so there was nothing to govern (or govern with), and thus no basis for the government to use "operating control" as a lever for a money laundering prosecution.

Given a very slimmed-down Tornado Cash stack, who does the DoJ now accuse of money laundering? Because they have to accuse someone. Crooks depositing dirty ether are still ending up with laundered ether, so there is by definition a "someone" in the stack who is providing laundering services to them.

In our hypothetical story, Storm and Semenov are not the money launderers, and the thrust of the DoJ's indictment confirms this. The two developers created software with presumably noble intentions: to provide regular folks with privacy from the panopticon that is Ethereum. Then they walked away, leaving the tool indelibly etched on the blockchain. It was only then that people started to interact with the tool, some of them to carry out illegal activity. It's this latter group who constitutes the guilty party.

Relayers are excellent candidates for a money laundering charge, a point I made last year. Because they process withdrawals on behalf of users, it would likely be a cinch to pin them for "conducting" transactions. Showing that relayers do this despite knowing that criminals may be their counterparties shouldn't be difficult for prosecutors to establish. And indeed, the DoJ's actual indictment is going in the right direction when it says that Storm and Semenov, along with "others involved in the Tornado Cash service, including the relayers," were engaged in the business of transferring funds, and goes on to accuse these "others," presumably relayers, of engaging in money laundering.

The second logical target for a money laundering charge is the licit users of Tornado Cash, in particular the large and savvy ones who used the tool regularly. A person who is aware that criminals are depositing dirty money into Tornado Cash smart contracts, yet decides to deposit their own funds into those same smart contracts, knowing that their effort will help these criminals conclude transactions that disguise the source of their funds, ticks all the boxes for a money laundering charge.

A licit user of Tornado Cash accused of money laundering might try to wiggle out of the charge by saying: "Sure, I knew crooks were using Tornado, and I know my efforts helped them. But I was only using it for legal reasons. I wanted to get privacy for myself." But that's not a very good defence against a money laundering charge, for the same reason that someone who tries to make a profit from obfuscating criminal funds can't evade a money laundering charge by saying they were only motivated by profit, and profits are legal. The desire to improve one's position, whether that be to get privacy or profits, isn't an excuse to launder money for crooks.

To sum up, the task of any prosecutor trying to bring money laundering charges against the Tornado Cash stack is to find the actual third-parties who misuse the platform for laundering. In a slimmed-down Tornado, that means chasing down relayers and savvy licit users. In the DoJ's actual indictment, it's also trying to show that owners/operators of user interfaces qualify, and while it's not a bad theory, we'll have to wait for the court date to see if it gets confirmed.

Who should pay for scams? Victims or their banks?

Scam call centre on CCTV, via BBC.

Here's a question for you. Should banks be required to reimburse customers who have been scammed?

I was recently reading a CBC article about a 63-year old Toronto man who responded to a phone call from a scammer claiming to be a Bank of Montreal employee, warning him that fraudsters had accessed his bank account. He was soon cajoled into paying out $16,000 to the scammer. Not only did the Bank of Montreal not reimburse the victim the full amount. It continued to charge him interest on the stolen funds.

Which isn't surprising. As the law currently stands, Canadian banks don't have to reimburse their customers who fall prey to authorized push payment (APP) fraud, a range of scams that includes calls from impersonated bank employees, RCMP scams, and fake Revenue Canada refunds.

So why not flip the whole system on its head? Why not require the Bank of Montreal to fully reimburse victims of these sorts of scams? The idea isn't without precedent. In 2024, UK will require that most victims of APP fraud be reimbursed within five business days by their bank.

There are some good arguments in favor of this policy. 

As it currently stands, individuals and their families, friends, and support systems are the main lines of defence for detecting scams. But there are big gaps in these lines of defence. Everyone has vulnerabilities, which scammers skillfully exploit to induce panic. Once in a panic, the victim's ability to think clearly is short-circuited, opening them up to being exploited by the scammer. A victim's second line of defence is to seek a second opinion from a sibling or spouse, but these third-parties may not always be available to help out the scam target.

Banks, by contrast, don't panic. Like scammers, they are cold rational machines. In addition, bank computers never turn off, which means they are available 24/7 to detect fraud. They also have a vast amount of knowledge about their clients' financial lives. Combine this data with technology like AI, and banks are in prime position to intervene in the crucial panic stage of the scam process, thus scuttling the scam.

Banks already do plenty of fraud detection. But imagine how much more vigilant they will be if their profits are at stake because they must reimburse scams.

None of this would be free, though.

Making Canadian banks liable for scams will inevitably mean higher fees and more banking frictions for everyone else. After all, reimbursing victims adds a new cost item to bank operating expenses. To recoup these costs, banks will hike fees on a range of banking products. Bankers will also want to reduce costs by catching scams in progress, which means extra checks when any irregular payment occurs, thus slowing down everyone's economic lives.

While no one likes extra bank fees and delays, think of these burdens as an alternative to the implicit costs that families, friends, communities, and civil society are already absorbing due to APP fraud. For instance, to prevent his elderly parents from being scammed, Jack tries to vet all of his parents' bank transactions. If banks are obliged to reimburse victims, Jack no longer needs to burden himself by monitoring his parents transactions; the bank will now take on that responsibility. The cost of this bank-provided anti-scam insurance comes in the form of Jack, and everyone else, paying higher fees and dealing with the odd delayed transaction.

This isn't a net loss, but a swap of one burden for another. Which is the better option for Jack and his parents? Is it more cost effective for him to monitor his parents transactions, or to pay his bank to do the job?

This gets into the problem of moral hazard. If banks insure customers against scams, then folks like Jack and his parents will become less vigilant, which means the public will be more susceptible to scam calls. However, as long as the additional vigilance brought to bear by banks more than compensates for the lost vigilance of individuals and communities, and does so at lower cost, the policy probably makes sense.

Beware, though. The policy could backfire it it unintentionally unbanks the vulnerable.

Victims of scams are vulnerable. They may be elderly, lonely, have low income, are weighed down by debt, or are working multiple energy-sapping jobs. Requiring Canadian banks to reimburse scams will make it more costly for them to service these groups. In response, banks may close the accounts of those they deem most likely to be tricked by scams. And so one of the unfortunate side effects of trying to protect the vulnerable from scams may be to actually burden them with a worse problem, no bank account.

There may be a fix for to this. Legislators may need to add a companion rule prohibiting banks from discriminating against customers on the basis of "scammability." However, keep in mind that this new rule will go on to have its own round of unintended consequences, which one hopes doesn't necessite a third rule, and a fourth one, and a...

Let's not forget the scammers, by the way, who won't sit idly by. 

Scam call centres will incorporate the new policy as a way to make their attacks even more effective. Imagine a panicked customer who is on the verge of transferring funds to the scammer. She suddenly blurts out loud: "Wait, is this a scam?" The scammer, reading off his script, pounces. "This is not a scam, ma'am, and even if it was, you live in the Canada. Your bank will cover it." The victim's worries allayed, the money is transferred, whereas without a policy of reimbursement the alarm bells in the victim's head might have been sufficient to get her to call a level-headed friend or family member for advice, likely putting an end to the scam.

In response to these tactics, banks will have to roll out their own information campaigns. Thus begins a cat and mouse game, whereby scammers adapt to banks and banks adapt to scammers, who in turn adapt to banks. But this is a cat and mouse game that has always existed, albeit with a different cast of characters, that is, scammers being pitted against individuals and communities. By changing the status quo and pitting scammers against a group that is more well-equipped for the game, bankers, we may all come out ahead.

Central bank digital currencies and the fallacy of immaculate adoption

I recently noticed that the Bank of Jamaica, Jamaica's central bank, has implemented two new marketing strategies to drive adoption of its new central bank digital currency, Jam-Dex. Jam-Dex is one of only four operational central bank digital currencies (CBDCs) in the world, having been introduced to Jamaicans in July 2022.

Now, I have no idea if these efforts will get Jam-Dex to succeed. In 2022, the system processed just J$357 million (or $2 million U.S. dollars.), but that comprised just six months of operations, so that's probably not enough time to judge it. What particularly interests me is how Jamaica's strategy of offering incentives to businesses and consumers serves as advance warning to other central bankers that one of the key tenets of the CBDC intellectual enterprise, what I call the immaculate adoption doctrine, is wrong.

In their white papers on CBDC, central bankers generally assume that the product will be immaculately adopted. The thinking goes like this: "We don't have to worry about devising a marketing plan for our new digital currency, nor think about incentives to promote usage, or the possibility that the product fails. All we've got to do is design it, put it out there, and – presto! – the public will instantly flock to it."

But as I've been saying for a while now the immaculate adoption doctrine is wrong. CBDC will probably just be a middling payments product. Existing options like cash, insured deposits and fintech balances work just fine for most folks, CBDC adding no extra features to the mix. It's just not possible to take a middling payments product and launch it, effortlessly and immaculately, into wide adoption. A big and expensive marketing push from central banks will be required if CBDC is to ever be adopted, Jamaica's Jam-Dex being a good example. And even then there's no guarantee of success.

The immaculate adoption doctrine gets even worse, though. Supremely confident in the success of their product, many central bankers fret that it will be too popular, hurting the banking system by stealing their deposits. To prevent this, they are building flaws into the product, effectively turning a middling product into a crappy one. A crappy product is even less likely to succeed.

Dirk Niepelt and Cyril Monnet recently make this same point with respect to a euro CBDC. In order to protect the business models of European commercial banks, the ECB wants to "trim the digital euro's attractiveness," the authors say, by adding holding limits for consumers and merchants. However, given the fact that European private sector payment options are already quite convenient, Niepelt and Monnet worry that the imposition of these hurdles condemns the ECB's CBDC to death on arrival.

The immaculate adoption doctrine of CBDC needs to be replaced by the it'll-be-a-hard-and-dirty-slog doctrine of CBDC. First, if they are to flourish, CBDCs can't just be carbon copies of existing private payments options. They need to offer something unique. Figuring out what these features are will take years of trial and error. Second, central banks will have to resort to dirty marketing tricks, incentives, bribes, arm twisting – all usually the domain of the private sector – to kickstart their CBDCs. Lastly, central banks need to stop deluding themselves that they can simultaneously launch a decent CBDC while also preserving the banking status quo. Those two things aren't possible! Stop pussy-footing around and admit that the whole effort will involve breaking a few banks.

Given that CBDC will be a hard and dirty slog, and not an immaculate ascendance, central banks need to think deeply about whether they truly want to undergo the pain of issuing a CBDC, which means being sure that society really needs one. Otherwise, they shouldn't get into the game.

UK's core payments settlement system fails... again. Some thoughts

As they increasingly forsake cash, regular folks are making dozens of digital payments every month. What they don't realize is how this growing reliance on digital payments increasingly yokes their commercial lives to the fate of a single piece of infrastructure: their central bank's large-value settlement system. When that system experiences a glitch, everyone's financial life gets put on hold.

In the United Kingdom's case, it is the Bank of England's RTGS settlement system that lies at the core of the economy. RTGS's centrality is highlighted by the fact that all the arrows in the chart below converge on it: every payment in the UK, big or small (except for cash), ultimately gets finalized using RTGS.

Alas, RTGS failed this Monday for six hours. No reasons were given, although I can't help wonder if it is was due to a software glitch stemming from Bank of England staff having been recently upgraded RTGS to the ISO 20022 payments language, rather than something like a cyberattack.

RTGS's centrality illustrated. Source: Bank of England

This isn't RTGS's first long failure. Back in 2014, a poorly-managed software update caused RTGS to shut down for 9 hours, leading to a revealing independent review.

The failure of the nation's key piece of payments infrastructure, even for just a few hours, is not a good thing. During those hours of unavailability, costly delays are imposed on day-to-day commerce as well as financial markets. Even when a buggy system is up and running, the uncertainty of another potential long failure acts as a pervasive cost on commercial society. 

To reduce these costs, central bank large value payments systems are typically built with multiple layers of redundancy. In RTGS's case, the hardware is hosted at two different sites, so that if the primary site goes down, the other one can quickly kick in. Presumably whatever knocked RTGS down last Monday was fierce  enough to incapacitate both sites.

A third layer of redundancy comes in the form of the Bank of England's Market Infrastructure Resiliency Service, or MIRS. With RTGS's two sites incapacitated, the Bank can "fail over" to MIRS, payments recommencing. MIRS uses different software, programming, and hardware, as well as being  hosted in a geographical remote location with a separate group of staff. This is achieved by an outsourcing arrangement with SWIFT, the same folks who run the global SWIFT messaging system.

There's no indication that the Bank of England failed over to MIRS earlier this week, staff preferring to focus on fixing RTGS instead. Alas, this choice subjected the UK economy to a long settlement delay. Why no fail-over to MIRS? Why choose such a long period of settlement deprivation?

A reading of the inquiry into the 2014 failure gives some clues into what may have happened two days ago. When RTGS failed on Monday, October 20, 2014, the Bank of England likewise chose not to fail over to MIRS. Why? The inquiry pointed to the fact that it would haven taken 2-2.5 hours to get MIRS up and running. Given this length of time, it made sense to try to fix RTGS instead, an inherently-preferable system because of features like the ability to save on liquidity, which the back-up system MIRS lacked.

Management was also reticent to switch on MIRS because they weren't sure if, after having activated it on Monday, they could turn it off on Tuesday night and manually return to a now-repaired RTGS without making a mistake. Bank officials only felt comfortable doing this manual switch back to RTGS on a weekend, because it afforded them much more time than a weeknight.

And thus trepidation about switching on the back up system led to it never being activated in 2014, which forced 9 hours of settlement deprivation on the UK economy.

Among its suggestions, the 2014 inquiry called for an upgrade to the MIRS back up option in order to make it a less anxiety-inducing option to turn to. The passage is worth reading in full:

Work should be undertaken to remove or reduce the barriers to invocation of MIRS so that
the Bank can "switch and fix" in parallel and in confidence. This should focus on testing the process to fail-back to RTGS intraweek (which is the primary barrier to invocation). If it is not possible to reduce this barrier, consideration should be given to enhancing the resilience and functionality within MIRS. In addition the Bank may wish to consider other back-up options for RTGS.

These were all good ideas. They would have reduced the hassle of resorting to the backup option by either improving the switching experience, or by upgrading MIRS's features so that being stuck on it for a few days posed less of a nuissance.

Which brings us back to 2023. If there is an inquiry into Monday's RTGS outage, investigators will need to explore why a multi-hour delay was once again imposed on UK citizens. Was it because, once again, the costs of using the back up system were deemed too high relative to the benefits? If so, were the costs deemed too high because none of the improvements suggested back in 2014 were adopted?

Failure to learn from the past would be unfortunate. These issues are especially salient because the Bank of England will introduce the next version of RTGS in 2024. Given that the updated RTGS will be built with more modern technology, it will (hopefully) fail less often than the older version. But it will still fail. What will the updated back up scheme look like? Will RTGS quickly switch over to tertiary site, or will the economy be forced to endure multi-hour settlement failure as a fix is pursued?

These are not just questions for the UK, but for every nation, since we all have large value payments systems on which commercial society is entirely dependent. It seems to me that if you have designed and built a back up system, that back up system should be, ya know, used. Those who operate them, usually central banks, should not be afraid to switch over. In the UK's case, that means that the decision to turn on MIRS (or whatever back up system the updated RTGS will use after 2024) should always be an easy decision for the Bank of England to make, not a gut-wrenching one.

Why bitcoiners should learn to accept bitcoin's ponzi nature as a good thing

When I describe bitcoin as a type of ponzi or pyramid, idealistic bitcoiners usually view this as an attack on the nobility of the bitcoin project. But they shouldn't. Bitcoin's ponzi nature is one of its greatest strength.

Imagine that you are trying to bootstrap an unorthodox and potentially illegal asset, one that you are marketing (or mis-marketing) to the world as a new monetary system. The standard way to get your asset to have a positive price is to have some funds in a vault somewhere and then link that new asset to whatever is held in the vault. Examples include failures like e-Gold and Liberty Reserve. Neal Stephenson's gold-backed currency in Cryptonomicon is another that followed this script.

It is this underlying intrinsic anchor that imbues the overlying novel asset that you are trying to kickstart with its original positive price.

The problem with this approach is that if the asset is illegal-ish, then to shut it down the authorities need only locate the underlying source of value and confiscate it. With the backing gone, it's game over.

Instead of relying on the tangible world of vaults and backing, bitcoin's creators got their asset to have a positive value by taking a non-tangible route through people's mindscapes. The original two dozen or so adopters began to play a self-referential mind game which went a bit like this: "If I think Joe will buy bitcoin at a higher price in the future because he thinks Susan will buy at an even higher price, for the reason that she thinks that I will buy even higher up, then I'll buy some bitcoin now."

This circular process, a ponzi game, worked. Very early on, bitcoin was imbued with a positive value. The mind game has since grown from a few dozen players to hundreds to thousands to millions, but the self-referential logic of a ponzi continues to be the basis for bitcoin's price.

It's easy to confiscate gold or shut down a bank account. These things are tangible and have actual physical locations with addresses. But it's tough to shut down a million imaginations participating in a collective illusion. First, there's a million of them. And second, imagination isn't something that authorities can fine or confiscate. People's dreams and delusions, their mental gymnastics and schemes, are all locked safely in their skulls.

And so the authorities very quickly gave up and let the bitcoin ponzi game play out, which probably wouldn't have been the case if bitcoin's creators had gone the non-ponzi route and tried to establish an anchor to real world assets.
 
In sum, bitcoin is resilient because its source of value is ponzi logic playing out in millions of people's heads. Bitcoin idealists may bristle at the idea of being associated with a ponzi, but they should embrace it as a strength.

AML works, even if it intercepts less than 1% of criminal proceeds. Here's how

Critics of anti-money laundering (AML) often point to the fact that just a tiny fraction of 1% of criminal proceeds are seized every year, their point being that AML is not working. Either update our approach to AML, they say, or just throw in the towel and legalize money laundering. (Here's a good example of this genre of critique).

But the critics' focus on the money-that-gets-caught misses out on one of the key channels through which AML works; by making crime more expensive. It isn't just the caught cash, but the costly detours that criminals are forced into along the way, or what I call the AML tax, that are signs of AML performing adequately.

I was reminded of this when I recently read Cash and Carry: the high cost of currency smuggling
in the drug trade, a fascinating paper about drug trafficking and high denomination banknotes. According to the paper, drug traffickers operating between Holland and Colombia typically contract with third-party broker specialists to move their funds back to Colombia. In six different situations from 2003 to 2011, coordinators of these Dutch brokerage operations were arrested. Everything from sketchy financial notes to detailed Excel spreadsheets were seized, allowing the authors of the paper to crunch the costs of running these businesses. 

Put differently, we get to see a juicy insiders' view on how much criminal resources must be sacrificed to successfully evade AML measures.

The authors report that the typical cash shipment to Colombia measured €300,000. The massive 500 euro note was the most popular medium for packaging shipments. The brokers relied on coordinators residing in Holland to supervise individual couriers traveling home to their families to carry these €300,000 packages. Couriers carried these funds in specially built parts of their suitcase.

The cost of moving these €300,000 packages to Colombia adds up very quickly. First, the coordinator had to absorb a 3% fee to convert the trafficker's original low denomination euro banknotes to the very compact 500 euro note. To that, add a 2-3% fee to pay the individual courier and 0.5% on airline tickets, as well as additional administrative costs like modifying luggage and bringing couriers to and from the airport.

A table of costs for the six apprehended coordinators is inserted below:

Data from six criminal investigations by the Dutch police showing the costs of smuggling cash [source]

Not listed in the table above, the broker had to absorb the cost of the occasional intercepted courier as well as having to pay their coordinators to manage the whole process. Lastly, brokers would have added a profit margin of their own. Adding all these amounts up, the authors peg the all-in cost of smuggling cash from Holland to Colombia at 10–15%, with a recorded maximum of 17%.

(This doesn't includes the costs that traffickers had to absorb back in Colombia in order to recycle the euro banknotes back into the local banking system.)

If AML doesn't work, as its critics claim, then Colombian drug traffickers wouldn't have had to contract with cash-smuggling specialists to move their funds in the first place. They'd have just used a bank, their cost of getting their funds transferred from Holland to Colombia falling from 15% to the fee of a wire transfer; a minuscule 0.5% or so. And so drug traffickers' profits would be 15% higher. (Or, if the cost savings are entirely passed on to the public, the price of cocaine would by around 15% lower, which would mean more Dutch cocaine consumption.)

But Dutch AML isn't toothless, the evidence being that drug lords must take a costly detour through the criminal note brokerage industry. This effectively acts as a tax on traffickers, hurting them in their wallets and probably putting a few of the marginal ones out of business. This AML tax may also result in higher cocaine prices, which helps to nudge Dutch consumers into legal alternatives.

These are all successes of AML, even when none of the cash is actually being confiscated. Those who claim AML is a "99% design failure" because it only intercepts a tiny amount of dirty money are naively omitting these benefits from their cost-benefit calculation. Sure, AML suffers from a boatload of costs, but the positive side effects of the AML tax deserve to be accounted for.