Technical Problems

“The structure of software systems tend to reflect the structure of the organization that produce them.”

– Douglas Crockford

The success of an emerging technology is not an inevitability, and not all technical innovation is unqualifiedly good. The software industry’s history is like any other engineering discipline: full of dead ends, false starts, and wrong turns. While software is unconstrained by many traditional factors, it is ultimately constrained by the economics of its applications and the limitations of computer science.

The fundamental technical shortcomings of cryptocurrency stem from four major categories: scalability, privacy, security, recentralization, and incompatibility with existing infrastructure and legal structures.

Scalability

In computer science scalability refers to a class of engineering problems regarding if a specific system can handle the load of users required of it when many users require it to function simultaneously. However regarding this problem, the technological program of bitcoin carries the specific seed of its own destruction by virtue of being tied to a political ideology. This ideology opposes any technical centralization, and this single fact limits the technical avenues the technology could pursue in scaling. As noted in the second chapter on the culture of cryptocurrency, bitcoin is inherently an anarchist project with an anti-state mentality that runs deep within its development community. This ideology informed the initial development of bitcoin to pursue censorship resistance as a core feature at any cost, including performance and transaction throughput. This design choice comes attached to a terrible set of engineering trade-offs that introduce several intractable problems to scalability.

By design, the bitcoin network should allegedly be immune to payment interdiction or law enforcement that wishes to restrict funds’ movement. This guiding principle is the central proposition any proposed scaling solutions must conform to to be considered acceptable in the cryptocurrency community, but it is also its technical undoing.

The bitcoin network cannot handle the volume of transactions that traditional payment systems can

The bitcoin scalability problem arises from the consensus model it uses to confirm blocks of pending transactions. In the consensus model, the batches of committed transactions are limited in size and frequency, and tied to a proof of work model in which miners must perform bulk computations to confirm and commit the block to the global chain. The protocol constrains a bitcoin block to be no more than 1MB in size and a single block is committed only every 10 minutes. For comparison, the size of an average 3-minute song encoded in the MP3 file format is roughly 3.5MB. Doing the arithmetic on the throughput results in the shockingly low figure that the bitcoin network is only able to do 3-7 transactions per second. By comparison the Visa payment network can handle 65,000 transactions per second.

The transaction throughput of bitcoin is very low by traditional database standards. It is a common marketing tactic in the database industry to inflate benchmarks or use synthetic workloads that advertise inflated write speeds for databases. Nevertheless, there are mature and open source databases such as Microsoft SQL Server, Postgres, and Redis for which we can gather very accurate information about their write throughputs. Postgres is a classical relational SQL database and is capable of 200 to 300 updates per second, or 12,000 transactions per minute. Redis is a key-value store that can perform 110,000 writes per second or 6 million transactions per minute. However, since all core banking solutions use traditional databases as their storage engine, these numbers represent a baseline figure throughput that should be stated for base comparison (Katz 2017).

An appropriate comparison would be the Visa credit card network, whose self-reported figures are 3,526 transactions per second. Most credit card transactions can be confirmed in less than a minute, and the network handles $11 trillion of exchange yearly. Credit cards and contactless payments are examples of a success story (Erlandsson and Guibourg 2018) for digital finance that have become a transparent part of everyday life that most of us take for granted. The comparison between bitcoin and Visa is not perfect, as Visa can achieve this level of transaction throughput by centralizing transaction handling through its own servers that has taken thirty years of building services to handle this kind of load. The slow part of transaction handling is always compliance, ensuring parties are solvent, and detecting patterns of fraudulent activity. However, for the advocates proposing that bitcoin can handle retail transaction loads on a global scale, this is the definitive benchmark that must be reached for technical parity.

Building these payment processing systems can be viewed through a lens of compromise between three factors: scalability, decentralization, and security. The design choices of bitcoin favor decentralization and security while making a sacrifice in scalability. The database infrastructure behind standard money transmitter services is designed to be scalable and secure.

The scalability issues of the bitcoin protocol are universally recognized, and there have been many proposed solutions that alter the protocol itself (Hinzen, John, and Saleh 2019). Bitcoin development is a collaboration between three spheres of influence: the exchanges who onboard users and issue the bulk of transactions, the core developers who maintain the official clients and define the protocol in software, and the miners who purchase the physical hardware and mine blocks. The economic incentives of all of these groups are different, and a change to the protocol would shift the profit centers for each of the groups. For example, while the exchanges would be interested in larger block sizes (i.e., more transactions), the miners (who prioritize fee-per-byte) would have to purchase new hardware and receive less in mining rewards for more computational work and thus incur significant electricity cost. This stalemate of incentives has led to mass technical sclerosis of the base protocol and a situation in which core developers are afraid of major changes to the protocol for fear of upsetting the economic order they are profiting from.

It is a common joke in software development that the answer to any difficult technical problem is simply to add another level of indirection to the problem. This leads us to our new problem: the lightning network. Since the base protocol is unscalable, the seemingly natural solution to adapting this network is to add yet another network on top of the bitcoin network. The proposed design of this system would batch settlements between peers into bidirectional state channels. These state channels are managed by a smart contract but must be monitored by the two parties on both ends to efficiently close the channel when a batch of transactions is finalized. This design opens a small but non-zero time window for fraud in a system in which one party will broadcast an old state to the contract and can extract the remaining bitcoin locked in the contract before the other party has finalized the transaction. The proposed solution is either a central registry in which lightning network participants would suffer reputational damage for this kind of fraudulent transaction or yet another level of indirection known as a watchtower contract. That watchtower is another smart contract that monitors the first contracts looking for mismatches between the main network and the channel states.

The lightning network itself introduces a whole new set of attack vectors for double spends and frauds as outlined in many cybersecurity papers such as the Flood and Loot attack (Harris and Zohar 2020). This attack effectively allows attackers to make specific bulk attacks on state channels to drain users’ funds. The lightning network is an experimental and untested approach to scaling, with progress on this scaling approach having stagnated since 2018. According to self-reported lightning network statistics, less than 0.001% of circulating bitcoin were being managed by the network, and transaction volume has remained relatively flat after 2019. No merchants operate with the lightning network for payments and as of today it is nothing more than a prototype. There is little evidence supporting this scaling model even works without introducing implicit custodial requirements, novel attack vectors, or new mechanisms for fraud. The perpetual narrative around the lightning network is that it has always been 18 months away from adoption. A narrative that is updated every 18 months it fails to deliver. In software parlance, the term vapourware is used to describe software perpetually in the works of being developed but never materializing into a usable form. The bitcoin lightning network is pure vapourware (Rosenthal n.d.b).

Outside of the bitcoin network, there are similar problems in other cryptocurrencies. The bitcoin meme of technical indirection through Layer 2 solutions have been translated to other systems and their development philosophies. This perspective views the base protocol as being only a settlement layer for larger bulk transfers between parties, and those smaller individual payments should be handled by secondary systems with different transaction throughputs and consistency guarantees. The ethereum network has taken a different set of economic incentives in its initial design. At the time of writing, this network is still only capable of roughly 15 transactions per second. There is a proposed drastic protocol upgrade to this network known as ethereum 2.0 which includes a fundamental shift in the consensus algorithm. This project has been in development for five years and has consistently failed to meet all its launch deadlines, and it remains unclear when or if this new network will launch. Since this new network would alter the economics of mining the protocol, it is unclear if there will be community consensus between miners and developers that the protocol will go live or whether they will see the same economic stalemate and sclerosis that the bitcoin ecosystem observes. The ethereum 2.0 upgrade is unlikely to ever complete because of the broken incentives related to its development and roll-out.

Because of slow transaction speeds, cryptocurrencies are almost impossible to use for legitimate commercial transactions.

The broader cryptocurrency community has seen a zoo of alternative proposed scaling solutions, these proposals going by technical names such as sidechains, sharding, DAG networks, zero-knowledge rollups and a variety of proprietary solutions which make miraculous transaction throughput claims. However, the tested Nakamoto consensus remains the dominant technology. At the time of writing, there is little empirical evidence for the viability of new scaling solutions as evidenced by live deployments with active users. Central to the cryptocurrency ideology is a belief that this technical problem must be tractable, and for many users, it is a matter of faith that a future decentralized network can scale to Visa levels while maintaining censorship resistance and avoiding centralization.

However, the inescapable technical reality is that every possible consensus algorithm used to synchronize the public ledger between participants are all deeply flawed on one of several dimensions: they are either centralized and plutocratic, wasteful, or are an extraneous complexity added purely for regulatory avoidance (Diehl 2021).

A consensus system that maps wasted computational energy to a financial return, both in electronic waste and through carbon emissions from burning fossil fuels to run mining data centers, is Proof of Work. Proof of work coins such as bitcoin is an environmental disaster that burns entire states’ worth of energy and is already escalating climate change, vast amounts of e-waste, and disruption to silicon supply chains (see Environmental Problems). The economies of scale of running mining operations also inevitably result in centralized mining pools, which results in a contradiction that leads to recentralization (Rosenthal 2022, 2014; Weaver 2018).

The alternate consensus model proof of stake is less energy-intensive; however its staking model is necessarily deflationary, it is not decentralized, and thus results in inevitably plutocratic governance which makes the entire structure have a nearly identical payout structure to that of a pyramid scheme that enriches the already wealthy. This results in a contradiction that again leads to recentralization, which undermines the alleged aim of a decentralized project. The externalities of the proof of stake system at scale would exacerbate inequality and encourage extraction from and defrauding of small shareholders (Rosenthal n.d.a).

Any Paxos derivative, PBFT, or proof of authority systems are based on a quorum model of pre-chosen validators. In this setup, even if they are permissionless in accepting public transactions, the validation and ordering of these transactions is inherently centralized by a small pool of privileged actors and thus likewise involves recentralization. Any other theoretical proposed system that is not quorum-based and requires no consumption of time/space/hardware/stake resources would be vulnerable to Sybil attacks (Rosenthal 2022) which would be unsuitable for the security model of a permissionless network.

The fundamental reality is that cryptocurrency currently does not scale and cannot adapt itself to fit the existing realities of how the world transacts. The technology can never scale securely without becoming a centralized system that undermines its very existence (Kharif 2019; Rosenthal n.d.b).

Privacy

Bitcoin wallet addresses are a unique global addressing system derived from the use of hash functions. In a nutshell, a bitcoin wallet address is generated from an elliptic curve private key which is a unique number generated randomly when a bitcoin wallet is created. This number is inconceivably large by everyday standards and will have hundreds of digits. If generated by a proper random number generator, the probability of that specific set of digits ever being generated again during the universe’s lifetime is infinitesimally small. This number satisfies the necessary properties of a secret value that the user holds private and uses to control access to their funds. There are 21⁶⁰ total possible addresses in the bitcoin protocol. An example wallet address:

1A1zP1eP5QGefi2DMPTfTL5SLmv7DivfNa

The public address generated associated with a wallet is encoded in a format where uppercase and lowercase letters stand for numerical values. This sequence of letters and numbers uniquely identifies the endpoint for other users to send funds to and can be shared publicly either in textual format or in a graphical format such as a QR code without containing any information about the user. From base assumptions, the number is essentially anonymous.

In the traditional banking system, a coding known as IBAN (International Bank Account Number) is the standard numbering system used to identify accounts and associated financial institutions. When issuing an international wire transfer, a bank account will ask for the receiving IBAN as part of the transfer. This number is then mapped internally to the account holder’s account and is stored within the bank’s core banking software and routing system.

A bitcoin address is, however, not fully anonymous. The bitcoin ledger itself is a fully public list of transactions that have ever occurred since the network’s inception. It contains the very first transactions allegedly by Satoshi as well as the most recent transactions conducted in the last 10 minutes. The full provenance of a bitcoin can be traced back to its creation and through every address it passed through.

This feature means that while accounts are anonymous, the global transaction data can be used to infer specific properties about when, with whom, and in what amounts an address is transacting. This kind of information is traditionally called metadata. For instance, metadata about your text messaging habits may not contain the direct messages you send. However, given a sufficiently large sample size, it is possible to deduce a person’s social network, their life partner, and coworkers from the frequency and timing of messages. Likewise, a great deal of information can be deduced from tracing the provenance of a bitcoin address, and thus bitcoin addresses are not entirely anonymous but partially anonymous or pseudonymous (Matzutt et al. 2018).

The tracking and tracing of bitcoin involved in criminal activities has emerged as a standard practice in law enforcement and emerging companies such as ChainAnalysis have been able to deduce quite a bit of implied information simply from public information. Unlike with bank accounts, law enforcement does not require a subpoena of public information for an ongoing investigation. Notoriously many users of darknet services such as the Silk Road were caught because of a misunderstanding about the transparency of the bitcoin ledger used by these actors.

Acquiring bitcoin has always had a bootstrapping problem for new users. In the early days of the protocol, one could use a home computer to mine small amounts by devoting spare CPU cycles to generate small amounts. However, for the last decade, this has been economically unviable. These days, the traditional onramp is to go through a domestic exchange or one of the offshore services. In the case of exchanges domiciled in the United States or in Europe, the onboarding process for accounts requires the account holder to present a government-issued identification and proof of address. This process is similar to opening a bank account and provides a mechanism for the institution to contact you and alert law enforcement of any suspicious or criminal activity associated with the account opened. This is a legal requirement known as Know Your Customer or KYC is the legal requirement to maintain an audit log of the account holder’s personal information and account activity.

Since the exchanges themselves operate accounts with massive inflows and outflows of transactions, their wallet accounts are massive hubs of activity that can easily be observed in the global ledger. If the exchanges are operating in a compliant manner, every transaction they process should internally be mapped to metadata about the account holders and their respective information. If an account was associated with criminal activity, law enforcement could subpoena the exchange and demand the information required to trace the account back to an individual. This information chasing through account metadata is the mechanism by which money laundering and wire fraud cases can be prosecuted.

This is in contrast to how the traditional banking system works, where bank secrecy laws are a central part of the obligation between a bank and its customers. Banks cannot use the transaction flows of their customers as part of their investments or share this information with other parties unless required by the courts. Bank transactions are required to be secure, private, and generally confidential information. When a wire transfer is issued by a company whose corporate account is at HSBC in London to Morgan Stanely in New York City, the metadata contained within that transaction could contain commercially sensitive information. For example, if a British company is sending large amounts of funds to a newly created American division, it may indicate the intent for the company to expand into the American market. There are cases where the constellation of transactions between known entities could be used to deduce confidential information about the parties. However, this fact poses an existential question about the efficacy of cryptocurrency networks as an international payment system if pseudonymous accounts leak information.

A retail bank account held by individuals is usually a simple structure that periodically collects deposits from an employer and frequent small debits for everyday activities such as groceries, rent, and buying coffee. Corporate banking, especially for large multinational corporations, can be quite complex and span a significant number of accounts and institutions.

In contrast, a corporation that wants to transact in cryptocurrency would have to address the fundamental issue that inflows and outflows from their accounts are commercially sensitive information. The amount of money that a corporation pays in payroll correlates with staffing and their operating expenses in specific regions and divisions in the company. The accounts receivable correlates with invoices it collects, its commercial interactions with its clients, and its lines of business. Public metadata of any of these transactions is private information, in fact it is usually some of the most protected information inside a corporation that is only shared with auditors and its direct banking relationships. Both of these parties are professionally and legally bound by confidentiality. A company electing to transact in cryptocurrency would leak confidential information like a sieve by choosing this public mode of payment.

The technical answer that one might propose to this problem is that the corporate should create a network of wallets and shuffle the payments between the wallets in random amounts and times to obscure the provenance of funds(White 2022a). This solution is needlessly complicated for a traditional corporate treasurer who should not perform this level of financial obfuscation and needless overhead for their normal daily activities. This solution is also indistinguishable from the money laundering process used by criminals. For cryptocurrency to pose any value to the commercial banking sector, this question requires a good answer.

If we step back, this conundrum begs a more profound question: Why are we making what was once a non-problem into a complex problem? Since banking was invented in Florence in the 13th century, the privacy problem has been solved. A mixed-visibility network with some access to authorities and privacy otherwise works very well. What does cryptocurrency offer except creating new problems(White 2022b)?

Security

The standard advice around the custodianship of cryptocurrency is that one should “be your own bank” and “if you do not hold your keys, they are not your coins”. These idioms are related to the fact that cryptocurrency is a bearer instrument, and if you hold the private keys to a set of funds, you are effectively in control of the assets, just as if you physically hold euros or dollars in a wallet. A problem arises when these funds are held by an exchange account which holds funds before they are withdrawn. These exchanges are not banks, they are not legally bound to hold deposits, and they are most likely not located in the customer’s jurisdiction. Most cryptocurrency exchanges provide no legal recourse for lost funds, and the funds held are not insured under any deposit insurance scheme.

Lost private keys for account have resulted in 20% of the supply of bitcoins being irretrievably lost.

In addition, these exchanges are some of the most targeted entities on the planet for hackers. In 2019, twelve major exchanges were hacked and the equivalent of $292 million was stolen in these attacks. Over time and in conjunction with bubble economics, these events have only increased in severity and frequency.

While some best practices can mitigate this risk, the fundamental design of bitcoin-style systems is that the end-user is responsible for their own keys and wallets by safeguarding their cryptographic secrets. This can be done through several strategies. So-called cold wallets are wallet keys stored in physical objects such as paper and not connected to electronic devices. Other systems such as hardware wallets allow users to secure and encrypt their keys on a dedicated hardware device.

Cybersecurity is one of our era’s biggest problems, and companies with significant information security budgets and dedicated teams regularly fail. A system that requires every depositor to have the same level of security as a chief information security officer and constantly be aware of threat vectors and potential attacks on key storage is an enormous cognitive overhead. At face value, this seems like an unnecessary burden on an average user who simply wants to hold funds and be protected against fraud in their daily transactions. The ask of individuals to supply their own banking institution-level information security is highly unreasonable.

In the course of human life, many situations occur which require third parties to be able to access or reset our accounts. If you forget a PIN code or lose a credit card, there is a simple mechanism to retrieve your funds by going to a banking branch and proving your identity. In a more extreme case of an untimely death, a person’s funds will be passed along to their spouse or children through inheritance and wills. The successors can petition the bank for access to the funds by presenting a death certificate and gaining control of the deceased accounts. Being one’s own bank makes both cases either impossible or needlessly complex. The human mind is fragile and subject to decay, mental disorders, and memory loss. If you forget the passwords to your hardware wallet or if it is physically destroyed, you lose access to your accounts. These events have already occurred to even some of the most sophisticated investors. An elaborate setup of data backups could mitigate this, safety deposit boxes or multiparty wallet setups, but such technical solutions are an unnecessary complexity burden for most users.

There are many news stories of ransom, kidnapping, and murder of crypto asset holders who attempted to safeguard their wallets personally. In cybersecurity, the term rubber-hose cryptanalysis satirically refers to extracting cryptographic secrets from a person by coercion or torture. A digital attack vector is unnecessary if criminals could extract the keys by kidnapping and torturing the owner and then laundering the funds from anywhere in the world.

Of course, the natural solution to this would simply be that most users should not be their own bank; instead, they should use a “cryptobank” which holds their funds and provides them access. However, this is ultimately just recreating the same centralized authority system which cryptocurrency advocates attempted to replace. Providing cryptocurrency security for the masses either introduces more social problems that thee technology has no answer to or results in a recentralization that undermines its own ideological goals. After all, we already have centralized banks and existing payment systems that work just fine.

Compliance

The fundamental reality of international commerce is that money management has been vital to a nation’s sovereignty and its ability to manage its economic growth and security. The movement, storage, and handling of money are regulated, and most countries have laws on the international movement of funds. Showing up at an airport in Berlin with undeclared cash above €10,000 will land one in quite a bit of trouble. If the value proposition of cryptocurrency is international money movement or extranational stores of value, then the technology will have to conform to existing regulations at entry points and exit points.

As a point of reference, it is helpful to consider how money transfers currently work. Nations with advanced economies have a domestic settlement system that allows banks within a regulatory regime to transfer funds between entities quickly. The United Kingdom has FasterPayments, Australia has BPAY, and the United States has CHIPS (Clearing House Interbank Payments System). These systems act as netting engines between the banks where trades are netted against each other instead of the total amount of all trades being cleared on every transaction. A trade goes through two steps. The first is clearing which is the confirmation of information between the payer and payee, and the second is settlement, which is the actual transfer of funds. For financial institutions to transfer funds, they will have what is called a Vostro account of the other bank, which records the funds held by the current bank on behalf of the other. Conversely, the other bank will have its Nostro account, which is an account held by the other bank which holds the current bank’s money. Transfers between the banks will be debited and credited in their respective Vostro accounts, thus allowing them to transfer money.

International wire transfers are done on the SWIFT (Society for Worldwide Interbank Financial Telecommunication) network(Dörry, Robinson, and Derudder 2018), which forms the messaging systems by which banks communicate messages about international transfers. The SWIFT network does not move money itself but simply is a messaging protocol for institutions to communicate the intent of transfers to happen. In addition, banks can only work directly with overseas banks with whom they hold an account, this is known as a correspondent account. If a bank does not have a correspondent banking relationship, it will have to route the wire through a third-party bank. This process entails having a Vostro account of a foreign bank or going through a chain of correspondent banks that do.

Every step along this chain incurs compliance checks with domestic laws and often involves multiple human and technical touchpoints inside and outside of the organization to complete the wire transfer. Along this process, each party involved in the transaction is legally required to carry out anti-money laundering (AML) and sanctions checks to ensure that the transfer complies with domestic laws and international treaties. First when these checks are complete can the transfer be completed and the money credited to the target account. The fees associated with these transfers are deducted from the total amount and represent the operational costs of performing all of these compliance checks along the way.

Cryptocurrency is purposefully built to evade regulation and make compliance impossible and is thus incompatible with cryptocurrency existing under the rule of law.

The bottleneck along this process is never the technical transmission of the messages. Just as any modern electronic messaging system, they are almost instantaneous. Any human touchpoint will be subject to the bank’s operating hours and days on which they are open for business, which is often only business hours and workdays.

Regular financial services companies such as Wise (previously known as TransferWise) have invented alternative solutions to international payments for small amounts that customers send often. Since most retail transactions are small (less than $5000 per day), Wise’s internal system matches users attempting to send small amounts in one currency block with corresponding users sending amounts in the opposite currency block. Wise uses these pools of funds to net out aggregate transactions via local bank transfers.

The inability to move money from a country is ultimately one of domestic internal infrastructure development and external international relations, rather than technical limitations. Moreover, the proposed use case for cryptocurrency as a mode of international remittances is fundamentally limited because of a lack of a coherent compliance story. Even if we were to use cryptocurrency as a hypothetical international settlement medium, this system has not removed financial institutions from the equation. The system’s entry and exit points would have to perform the same checks of outgoing and incoming money flow required by many international agreements.

In this hypothetical scenario, we have simply shifted the custodial, compliance, and identity management responsibilities to a different centralized entity that performs precisely the same activities and ultimately is subject to the same legal liabilities. In this setup, instead of settling in a national currency pair, there are now two currency pairs with a useless and volatile intermediary step in between. Using cryptocurrency for remittance has not disintermediated anything- it has simply shifted the intermediaries and introduced another level of indirection for no apparent reason.

A system that aimed to replace the existing international transfers would be subject to a similar set of rules regarding international transfers and capital controls, and it is naive to think that hundreds of treaties would be renegotiated on behalf of digital currencies. This wishful thinking is at the heart of the absurdity of crypto; the belief that somehow because something is on a peer-to-peer network(Rosenthal 2014), it is somehow exempt from the rules of being in a society.

Of course, like all cryptocurrency arguments, the counterargument is ideological: compliance is a non-issue because nation-states should not exist and should not have capital controls. This ideological goal is inexorably embedded in the design of cryptocurrency, making it an unscalable and untenable technology for any real-world application where sanctions, laws, and compliance are an inescapable part of doing business in financial services (Hanley 2018).