Satoshi's Mistake

Bitcoin is now ~15 years old. In one respect it has been an indisputable success: it has a market cap of over $1T at time of writing this. But to me the most important definition of success is in the eyes of its creator — if Satoshi Nakamoto were to see Bitcoin today, and reflect honestly on it, would he consider it a success?

Obviously this depends on who invented Satoshi Nakamoto really was, but in my view he was almost certainly Hal Finney:

• He created the immediate technical predecessor to Bitcoin
• He was the second contributor to Bitcoin (after Satoshi)
• He received the first Bitcoin transaction
• He had an acquaintance named Satoshi Nakamoto
• He spent years working on anonymity/psuedonymity
• He independently had a large enough Bitcoin stash to be comfortable without spending Satoshi's
• He died in 2014, explaining why Satoshi has never done anything with his ~$50B fortune

The argument against is comparatively weak:

• He denied it
• He was able to convince some Forbes writer of this

Which is more likely — that all of the above is a coincidence, or that Hal Finney lied reasonably well^[0]?

Hal Finney was, as far as I can tell, a good guy who invented Bitcoin for the right reasons^[1]: to protect people — from governments and other powerful entities like Visa — by allowing them to transact without (centralized) intermediaries. On this score, how has Bitcoin performed?

What Cryptocurrency is Used For

First let me ask — how many people do you know who have made a purchase of real, tangible goods with Bitcoin? From my time in the SF Bay Area, I know a few who used it to buy drugs. But I've never met anyone who used it for anything else — besides me. I could tell you that it was a poor user experience for both parties, but that's far less notable than the fact that I'm probably the only person you've ever come across who has done it.

The Bad

So what is Bitcoin used for? Crime. Strictly speaking crime is not necessarily a bad thing — it's whatever the government decides it doesn't like, and part of the reason Hal Finney created Bitcoin was to protect people from their governments. But I have never seen any evidence that people are using Bitcoin at large scale to buy and sell banned books or to fund-raise for political dissidents in hiding. What I have seen is:

• Drugs
• Warez (software for cybercrime)
• Scams
• Ransoms
• Transactions between organized crime groups

I've spent the last 7 years working in the cyber insurance space, so I've had a lot of exposure to the explosion of ransomware, but even outside of my professional life I've seen take a toll on my community. In recent local news I've seen it hit:

• Our library system (which took months to fully recover after refusing to pay the ransom)
• Our airport (hundreds of flights delayed or canceled)
• Our cancer center (all patients' info sold on the darkweb)

But far larger than this is the impact of scams. The book Number go Up by Zeke Faux dives deep into one genre of scams, known as pig butchering, where a victim is tricked into making a series of increasingly large transfers in a fake investment scheme, with ponzi-esque and romance scam components. The total losses here are hard to estimate^[2], but are enormous — $2.57B in cryptocurrency scams alone were reported to the FBI in 2022^[3]. Some highlights include:

• A Kansas bank failed in 2023 after their CEO fell for a $47M big butchering scam
• The FBI seized $112M from a single pig butchering ring^[4]

It is worth noting that most of these scams use a different blockchain than Bitcoin to transfer funds. But in my opinion Bitcoin represents the key technical innovation they are leveraging. Some aspects of the cryptocurrency ecosystem may represent significant technical innovations beyond Bitcoin, e.g. on-chain computations with Ethereum, but I do not think these innovations are leveraged in any meaningful way in these scams^[5]. In a world without other cryptocurrencies, they would use Bitcoin and operate almost identically and nearly as effectively, while without the invention of Bitcoin they would be forced to use the traditional financial system.

Arguably more important than the targets of these scams are the victims that perpetrate them — many of whom are slaves. Throughout Cambodia and Thailand are dozens of forced labor compounds filled with trafficking victims who are forced to perform these scams all day. Rather than leaving these to a footnote, I strongly recommend reading this ProPublica investigation. Organized criminals imprison hundreds to thousands of people in these crowded compounds for years or indefinitely to work off illegitimate "debts", forcing them to work long hours under constant oversight and beating them for failing to meet quotas, and are often bought and sold between groups^[6].

The total number of people enslaved for these scams is hard to estimate, although the reporting above suggests at least tens of thousands when you consider the dozens of large compounds identified. The UN High Commissioner for Human Rights has estimated at least 220,000 between Cambodia and Myanmar. I am not sure how much to believe this number^[7], although as I will show it's consistent with back-of-the-envelope calculations. But it's also not quite the right question to ask.

Marginal Considerations

A fair criticism of the above is that just because scams use cryptocurrencies does not mean that they would not exist without them, or that other scams wouldn't take their place. It's important to try to estimate the marginal impact of cryptocurrencies on scams.

To do this, I've compiled FBI data on internet scams from 2018 to 2023^[8], which crucially provides data broken down by scam type and also tracks the total losses by crypto vs. fiat channels (although unfortunately the per-scam-type crypto losses are only available for 2023^[9]). I've used the 2023 cryptocurrency losses per scam type to classify each scam type as fiat-mandatory if it is almost always^[10] performed using fiat currencies, and then used the cryptocurrency numbers for each year to break scam losses down into three categories: fiat-mandatory, fiat-optional (for fiat losses due to scams that often use cryptocurrency), and crypto. Every category has grown, but crypto is in a category of its own:

The fact that cryptocurrency scam losses have vastly outpaced all other categories strongly suggests that cryptocurrencies have enabled these scams, especially given how similar the trajectories of the other two categories are.

However unlikely, this still could be a coincidence. The main objection I can think of being that the types of scams that primarily use crypto have grown for other reasons (e.g. because people spend more time on their smartphones, especially post-pandemic) and that it's this number that has really increased and the gap between the fiat-optional and crypto categories is due to substitution from fiat to crypto that has a relatively small impact on the volume of scam losses. To evaluate this, it would be best to limit our attention to only the classes of scams enabled by widespread smartphone use, but unfortunately without annual breakdowns of crypto vs fiat losses per category this is not possible. However, we can get over 80% of the way there by excluding by far the largest category — business email compromise (BEC) — as well as real estate scams, which almost never use crypto. This gives us:

Here you can see a little of that story — the fiat-optional losses are growing faster than fiat-mandatory. However, crypto is still in a category of its own and the fiat-optional curve looks much more like the fiat-mandatory curve. Furthermore, this analysis may understate the affect of cryptocurrencies, because if certain types of scams become easier as a result of cryptocurrencies you would expect not only substitution of channel within scam types but for criminals to substitute scam types entirely.

It's still worth asking whether the increase in scams is responsible for an actual increase in slavery in Southeast Asia. Firstly, what fraction of these scams does so-called pig-butchering make up? Of the ~$5.6B in crypto losses in 2023, ~$4B of that was via what the FBI calls "Investment" scams. This presumably includes multiple kinds of scams, but they note that "the most prominent was a unique kind of confidence-enabled cryptocurrency investment fraud" which they go on to describe at length and clearly refers to pig-butchering, with an additional warning that "individuals who travel or live abroad should be aware of the risk of false job advertisements linked to labor trafficking at scam compounds overseas"^[11]. Thus it seems safe to assume that at least $1B is pig-butchering scams using slave labor. Given the prices quoted for slaves in the ProPublica article (e.g. $10k), we can estimate that the profits generated per person are probably in this range, so a low-end estimate is that ~100,000 people are enslaved to this end.

It is also possible that slave labor used for scams is merely reallocated from other activities such as fishing or agriculture. The standard econometric approach to answering this question hinges on estimating the elasticity of supply — how much an increase in the price criminal firms are willing to pay for slaves increases the number of slaves traffickers will provide. However, after a frustrating literature review my only conclusion is that we have no idea what this elasticity looks like^[12].

However, there is another approach we can take: looking at the statistics for other large industries in the region which frequently use slave labor. The one that has the most international attention and economic research is the fishing and seafood processing industry in Thailand, which employs ~350,000 workers^[13] with annual wages in the $3.5k-$4.5k range^[14]. Studies on forced labor^[15] in the industry found that ~10% of workers were subject to it in 2019 while ~13% were subject to it in 2023, although the size of the samples (470 and 400 respectively) mean this is not a statistically significant change.

Nonetheless, it is enough to exclude a large decrease^[16] in the rate of slave labor in the Thai fishing industry. And the industry is not nearly large enough to hide a decrease from substitution to pig-butchering scams — at the numbers above, there are about ~35,000 slaves working in the Thai fishing and seafood processing industries and their withheld wages are worth ~$140M (ignoring the costs of keeping them alive), a small fraction of the $1B pig-butchering industry and far fewer people enslaved.

As such, we can draw a similar conclusion as we did with respect to the prevalence of scams themselves — pig-butchering scams have generated a largely exogenous increase in the number of people enslaved, at least in the tens of thousands.

The Good?

All of this could be outweighed by the positive uses of Bitcoin or other cryptocurrencies — after all, you could make a similar criticism of the US dollar. But what are these uses? I've asked a number of crypto proponents, and the #1 example I always get is remittance payments. Indeed, this seems like a perfect use case: large, non-time-sensitive transactions (mitigating the technical overhead of cryptocurrencies), often to countries with poor banking infrastructure and corrupt or authoritarian governments. Among the more serious institutional and academic proponents of cryptocurrencies this is a major talking point. But does it happen?

For the most part, no. Good data is hard to come by, but what we have very strongly suggests that Bitcoin (and any other cryptocurrency) is a rounding error in the remittance market. The best data we have is out of El Salvador, where the government has made Bitcoin an official currency, thereby taking it out of the shadows as much as possible. As of March 2022, Bitcoin represented ~1.6% of all remittances to El Salvador^[17].

In retrospect, this is not particularly surprising. Remittances through traditional channels cost about 6% to send^[18], and while you can reduce that to 1-2% through Bitcoin, since nobody actually uses Bitcoin in day to day life you need to also pay fees on both ends to convert between currencies. While easy and cheap to do from USD as an American, this is a lot more difficult and expensive in a third-world country. And at every step of the way there is a high risk of getting scammed, since Bitcoin exchanges do not invest nearly as much in their brand reputation as e.g. MoneyGram and therefore the dishonest ones are harder to distinguish from the honest ones, which also have less to lose by becoming dishonest. Overall, it is unlikely that the risk-adjusted costs of Bitcoin remittances are lower or even equal to the costs of traditional remittances.

Now, you could argue that these costs assume an unsavvy user, and that if the sender and recipient are both savvy they can reduce the risk to almost 0 and also optimize the exchange costs. This is true, but it is also true of traditional remittances. I can send money to El Salvador for substantially less than 6% — by having the right banking relationships, knowing the menu of options available in traditional finance and how to compare their costs, etc. Services like MoneyGram exist precisely to service less savvy users on both ends, and their cost reflects this.

The other answer I often get from crypto proponents is: it hasn't happened yet, but it will! This is a reasonable position for a novel technology, but it becomes less defensible each day. After 15 years, hundreds of billions of dollars of investment, and the embrace of multiple major institutions and governments, you would expect to see something^[19].

If you still believe crypto has enough to offer to outweigh the negatives, then I doubt I can persuade you — but remember that the bar gets higher and the probability gets lower each day.

Verdict

So what is Hal Finney's legacy?

• Tens of thousands of people forced into slavery
• Hundreds of thousands made destitute
• Millions scammed
• Digital infrastructure routinely compromised by ransomware
• Over 600 TWh of electricity consumed^[20]

I certainly hope he'd be ashamed.

Lessons

So why am I writing this? To slander a dead man?

I am writing this because all of this could have been foreseen and avoided, and Hal Finney will not be the last technologist with a bright idea that would make the world worse. It is worth understanding why he made such a mistake.

First, what were his goals? I think there are 4 coherent ethical goals, and that Finney held all of them:

1. To protect people from monitoring by authoritarian governments
2. To protect people from being shut out of finance by their gatekeepers
3. To give people an alternative when traditional finance firms price gouge
4. To give people a more reliable alternative to traditional finance

Whether Bitcoin could address these goals depends on whether Bitcoin is:

1. widely used for regular transactions (with no centralized go-between like PayPal), or
2. converted into traditional currencies or passed through centralized services for tangible purchases.

Clearly right now we are in the second case. In this case, financial services are still required in order to make tangible purchases. This re-introduces the same issues that Finney hoped to avoid with Bitcoin, at best offering an alternative menu of firms to choose from, since the same dynamics that lead banks and transaction processors to be dominated by a few large firms and easily co-opted by governments apply to exchanges.

But the first case was the plan — albeit always a long shot^[21]. In this case, I think it's fair to say that Bitcoin would accomplish the first 3 goals, although the lack of any backing beyond mutual confidence seems like a major impediment to the fourth^[22]. But would this actually be ethical? It would not just put financial transactions beyond the ability for authoritarian governments to monitor, but any government. While many technologies might appear to affect both authoritarian and legitimate governments equally, there are 2 major differences between Bitcoin and something like encryption or even weapons:

• Monitoring financial transactions is a legitimate action of governments, unlike say monitoring general interpersonal communication
• The "activation energy" for using it is very low, unlike say starting an armed rebellion

Here is where I suspect Finney and those of his ilk would disagree with me. They would argue that:

• Monitoring financial transactions is not a legitimate action of governments
• All governments will eventually become authoritarian unless it is made technologically impossible

Both of these are false.

First, monitoring financial transactions is necessary for fair taxation, which is indeed a critical function of a legitimate government in the modern world. Without reasonable tax revenue, no government could hope to compete with the largest corporations, thereby allowing (and effectively forcing) the largest corporations to take over their military and public order functions, resulting in corporate rule.

Second, believing technology is the only way to prevent authoritarianism is an inherently suspect argument for technologists to make, as it falls into the common fallacy of believing that one's own field is the sole and all-important source of solutions to a problem. And this is indeed what they've done — ignoring the entire social and political history of our species, full of governments of all stripes from which we can see that authoritarianism is not the inevitable end of all governments and identify numerous risk factors and mitigations.

To understand why people hold these beliefs it is necessary to understand where they come from. These are more or less core tenants of the cypherpunk political ideology, which is basically the idea that the society of Snow Crash is either desirable or inevitable and that what people need are the tools to become the protagonist of a Cyberpunk novel. Cypherpunk gets its appeal from aesthetics rather than any rational analysis. And aesthetically I am sympathetic to it.

But I cannot emphasize enough how stupid it is as a political ideology. The world of Snow Crash is obviously bad. Finney may have been brilliant in the field of distributed systems, but in political philosophy he and his ilk were children — their views so ill-formed and arbitrarily constructed that they are not possible to seriously engage with.

Fundamentally, he was either unable or unwilling to engage in serious political thought, which I think is the real lesson here: as technologists, we have a moral obligation to think seriously and critically about our work. Hal Finney didn't, and thus shares in the blame for the damage his work has done.

0. ^

   Let's be honest, Forbes is not the most discerning journalistic enterprise

1. ^

   Perhaps it is worth mentioning a more fringe possibility behind Bitcoin: Paul Salotshi Le Roux^[23]. In my view he is a much less likely candidate, but he has a few things going for him:

   • He has a long history of cryptography programming
   • He previously used the name Satoshi on forged documents
   • He had a strong use case for Bitcoin (and was an early adopter)
   • He couldn't have done anything with Satoshi's fortune because he's been in prison since 2012

   I think it's interesting to consider the possibility that Le Roux is Satoshi not because it is especially likely, but because of the contrast between him and Finney. Unlike Finney, Le Roux is a monster — born in the self-declared apartheid state of Rhodesia, he is currently serving a 25 year sentence at ADX Florence (a Supermax prison) after years of cooperation to reduce what would otherwise surely have been a death sentence for^[24]:

   • Stealing credit cards
   • Manufacturing and smuggling cocaine and methamphetamine
   • Trafficking weapons to Philippine warlords
   • Developing explosives for Iran
   • 7 murders

   Bitcoin was probably created by Hal Finney. But it would make Paul Le Roux proud.

2. ^

   One attempt to estimate this is Chainalysis's 2022 Crypto Crime Report. They conclude that ~$15B in crypto transactions were part of criminal activity, but that this is a very small part of what they argue is a ~$15T total transaction volume. However, they only track how much money is moved between addresses, and so are unable to distinguish between final, intermediate, or even wash trading transactions. Their methodology for identifying criminal addresses is also unclear. As such, I have little stock in either number — the former may be an over or under-estimate depending on their identification of criminal addresses, while the latter is definitely an overestimate.

3. ^

   "In 2022, investment fraud caused the highest losses of any scam reported by the public to the FBI’s Internet Crimes Complaint Center (IC3), totaling $3.31 billion. Frauds involving cryptocurrency, including pig butchering, represented most of these scams, increasing 183% from 2021 to $2.57 billion in reported losses last year." — Dept. of Justice

4. ^

   ibid.

5. ^

   It is true that many of these scams nominally involve investing in some supposed new cryptocurrency, but these supposed cryptocurrencies are often nonexistent and at any rate they are not used for the actual fraudulent transfer of funds.

6. ^

   These conditions are attested in this NYT investigation as well as in Chapters 18-19 of Number go Up.

7. ^

   See page 9 (7 as numbered). The report's endnotes cite this as "Information on file with OHCHR." While there may be good reason to limit distribution of this information (e.g. because the source is a double-agent working with criminal groups), this lack of transparency makes it very hard to accept such a high figure.

8. ^

   You can download my spreadsheet here. The original data comes from the IC3 annual reports which are also linked in the spreadsheet.

9. ^

   Comparing the breakdown in the 2023 Cryptocurrency Report to the "Cryptocurrency" and "Cryptocurrency wallet" numbers provided in the 2022 and 2023 annual reports, it is clear that these numbers are mutually exclusive so I've added them together in the loss statistics.

10. ^

    There is a large gap in the % of 2023 losses per scam type that were via cryptocurrencies, with none between 14% and 28%, so I've chosen this gap as the dividing line between fiat-optional and fiat-mandatory scam types.

11. ^

    2023 Cryptocurrency Report, p. 13

12. ^

    The economic literature that has looked at the market dynamics here can be divided into 3 categories: theoretical models, simulations using hypothetical parameters, and macroeconomic models relating differences between countries to social and economic variables. None of the models I found that incorporate supply elasticity even make an estimate of it, except those which assume perfect elasticity (i.e. the supply curve is a vertical line) in order to simplify their calculations — which is obviously ridiculous.

13. ^

    Endline research findings on fishers and seafood workers in Thailand, p. 11 - International Labor Organization

14. ^

    Supply Chain Study on Forced Labor in the Fishing Industry in Thailand - US Dept. of Labor, Bureau of International Labor Affairs, Office of Child Labor, Forced Labor, and Human Trafficking

15. ^

    Both studies use the International Labor Organization definition of "forced labor": "The term forced or compulsory labour shall mean all work or service which is exacted from any person under the menace of any penalty and for which the said person has not offered himself voluntarily" Ibid, p. 28

16. ^

    We can rule out a larger than 1% decrease based on these samples with p=0.05.

17. ^

    Are Cryptocurrencies Currencies? Bitcoin as Legal Tender in El Salvador by Fernando E. Alvarez, David Argente, and Diana Van Patten

18. ^

    The global average was 6.35% in Q1 2024 and slowly falling according to a World Bank report, but for digital services (as opposed to hard cash) this was 4.96%.

19. ^

    The only counterexample I can think of is nuclear fusion (which may end up not being a counterexample, but I at least still believe in it). But in the case of nuclear fusion, we long ago identified the technical obstacles that needed to be solved and have made decent progress on them. For crypto everyone's just waiting for something to come along.

20. ^

    Source — this is about 430,000 tons of CO2. There are technical proposals for reducing future energy usage, and while I am not completely convinced by these proposals I am not primarily interested in the environmental costs of Bitcoin and so am not counting potential future CO2 emissions against it.

21. ^

    This is a very long shot for two reasons. First, all the same cost issues that prevent Bitcoin from being cheaper than traditional finance in most cases would still be present even with ubiquitous usage. Second, the path dependency issues are obviously massive.

22. ^

    I wrote a previous blog post on this, which I stand by — I don't think Bitcoin is any more widely used as a currency today than when I wrote that 6 years ago.

23. ^

    This claim was popularized in a Wired article

24. ^

    US v. Le Roux