Decentralized Finance and Its Discontents


staci warden, the former executive director of the Milken Institute’s Capital Market Development practice, is CEO of the Algorand Foundation.

illustrations by lincoln agnew

Published January 24, 2023


Do Kwon, co-founder of Terraform Labs, thought he might improve upon the financial model of “stablecoin” cryptocurrencies pegged to the dollar. Instead of holding dollar-denominated assets in reserve to defend the pegged value, he reasoned, why not create an algorithmic stablecoin — he’d call it Terra (which trades as UST) – worth one dollar of yet another cryptocurrency he would create, called LUNA.

This way, when the market value of UST became greater than $1, users could trade $1 worth of LUNA for UST for a profit, which would then drive down the value of UST. The reverse would happen if UST traded below $1. They’d balance each other on their own so he wouldn’t have to hold all that expensive U.S. dollar collateral to back the value of UST. Add a lending protocol on the Terra blockchain paying 20 percent interest rates in order to Anchor (the name of the protocol, really) demand for UST, and he was done. What could possibly go wrong?

I know, I know: I lost you somewhere between UST and Anchor — as, apparently, did a lot of investors in Kwon’s magic money machine. This much is clear, though: it all came crashing down in spectacular fashion last May, when the symbiotic relationship between UST and LUNA unraveled in a kind of process that Wall Street has traditionally, though not entirely helpfully, labeled a “death spiral.” The spiral started with a run on UST, which triggered an ever-increasing issuance of LUNA to maintain the peg, which then caused LUNA to decrease in value. Rinse and repeat.

The whole thing plummeted to nothing, wiping out $60 billion in value and earning the meltdown the title (against some tough competition earlier this century) of the largest singleasset failure in the history of finance. It goes without saying that countless small investors were wiped out in the process. Kwon is now wanted by Interpol in 195 countries.

One of the not-so-small investors to take a swan dive was the hugely influential hedge fund and proprietary trading firm, Three Arrows Capital. Three Arrows had made hundreds of millions of dollars in highly leveraged positions in LUNA (among other cryptocurrencies) with customer money that was lent, in a shockingly high percentage of the time, without collateral of any kind. Three Arrows went from a claimed $18 billion in net asset value to bankruptcy in a matter of weeks this past July — and something like $3 billion is still owed to creditors today.

The fallout reverberated across crypto markets, and institutions that had large, unsecured positions with Three Arrows began to fall like dominos. For example, the lending platform Celsius Network, went from $18 billion in assets under management and another $8 billion in client loans on its balance sheet to $167 million cash-on-hand by the time of its liquidation. An exchange called Voyager Digital lost $650 million “deposited” at Three Arrows at the time of its bankruptcy, and Voyager’s more than three million customers are now struggling to get their money back.

This implosion, in the teeth of the geopolitical and macroeconomic policy headwinds that have affected all financial assets, ushered in $2 trillion in cryptocurrency losses in nine months, in what has become the entire ecosystem’s terrible, horrible, no good, very bad year.

The Promise of Decentralization

“If we lose the battle to preserve public decentralized blockchains (aka ‘crypto’), with mathematical certainty in 10 or 20 or 30 years a long, dark night will fall over humanity,” tweeted @punk6529, a pseudonymous but highly influential NFT art collector, in October.

The remarks were a reaction to both the potential power of digital currencies issued by central banks and the recent U.S. Treasury sanctions on Tornado Cash, a “currency mixer” platform that facilitates anonymous payments. For libertarians like @punk6529, the value of a decentralized financial system built on a peer-to-peer framework rather than financial intermediaries is that the former enables a range of financial activities that are both permissionless and censorship-resistant. Anyone can access a decentralized cryptocurrency protocol and nobody, at least in theory, controls it.

Note, though, that crypto is more than a libertarian cause. Decentralized finance holds similar promise for progressives. Distrustful of the value of banks for the households and entrepreneurs of Main Street, and tired of fattening the market caps of large, centralized corporations like Facebook, many on the democratic left see the power of the blockchain as a way to capture more value for the creators of that value — as well as a model for bringing large swathes of the economically disenfranchised into the global financial system.

The fundamental operating principle behind both perspectives is that if multiple parties can exchange valuable things without having to trust or even know one another, then the cost of their financial interactions will be significantly reduced. That, in turn, will make it possible for many more people to participate in many more economic activities, and therefore have many more opportunities to create wealth. And despite several setbacks over the last decade, this belief continues to drive much of the interest in (and mainstream adoption of) decentralized finance by both retail and institutional investors.

Societies and economies do tend to naturally gravitate toward centralization. And last summer’s wealth destruction has indeed tested the zeal for decentralization among the crypto faithful. But crypto has always emerged from its winters of discontent stronger and more robust than before, and there is no compelling reason to think that this past summer will be an exception.

In fact, one can argue that the collapse of crypto in 2022 says little about the promise of decentralized finance. Rather, the weakness of the LUNA protocol and the centralized entities bludgeoned by its downfall do more to bring into focus the merits of decentralized finance than they do to highlight its weaknesses. As such, it is worth trying to understand the problems that decentralized finance is trying to solve in more detail, and thereby to gauge the viability of its underlying promise.

Gilding the Poisonous Lily

The implosion of Terra-LUNA and Three Arrows Capital this summer was the Bear Stearns of crypto. In mid-November, just as this article went to press, came the Lehman Brothers version. FTX, one of the most popular exchanges, valued at $32 billion in January, and Alameda Research, its sibling hedge fund and market-maker, blew up through a combination of trades gone sour, excessive leverage and (alleged) fraud in the form of lending out customer balances to support trading.

We will likely see billions of dollars in customer and investor losses before it’s all over. But whereas the fall of Terra-LUNA hit more sophisticated, “crypto native” investors, FTX was a retail meltdown. It was a well-known, respected, global exchange, with the likes of Tom Brady and Steph Curry as its spokespeople and a founder who testified regularly (about optimal regulatory regimes!) on Capitol Hill. The FTX implosion shocked everybody, even the most sophisticated crypto OGs. But FTX proves the rule, more than it is the exception. Centralization requires trust, and trustworthiness is hard to come by. For those who fear they will never find it, decentralization and code-based regimes are at least part of the answer.

The Problem With “Tradfi”

The market structure for trading stocks in the United States boggles the mind. For one thing, retail pricing is opaque, as noninstitutional investors are passed through multiple intermediaries — some of whom pay to see those retail orders to gain an advantage in executing their own trades. For another, exchanges charge intermediaries tens of millions of dollars annually for access to their data feeds. And they need to subscribe to multiple exchanges as the only way to provide price discovery to retail customers. Execution speeds are down to the millisecond, forcing market makers and “quant” trading firms to pay millions of dollars more to keep servers on exchange location. And, to make sure nobody’s electrons get an unfair advantage, each institution’s cable to the exchange’s data source is exactly (exactly) the same length. All that, but exchange trading shuts down completely every day at 4 p.m. and the entire trillion-dollar ecosystem takes the weekends off.

The banking and consumer credit markets of traditional finance (“TradFi” to crypto hipsters) aren’t much better. Despite the revolutionary technological and organizational changes that have taken place in the past century, the cost of financial intermediation in the U.S. has remained more or less constant at around 2 percent of transaction value since 1890, while bank consumers in the U.S. still pay an average of $14 per month in basic account maintenance fees and $5 every time they want to access their own money at an ATM. And despite efforts to make finance more accessible to more people, the system is still structurally discriminatory. For example, researchers at the University of California (Berkeley) found that Latino and African American borrowers pay almost eight basis points (0.08 percent) higher mortgage rates — a whopping $765 million more annually than white borrowers — and had some 1.3 million more mortgage applications rejected between 2009 and 2015.

By a range of measures, even something as simple as making a payment is still excessively costly and cumbersome for the vast majority of people. First, from a personal security perspective, it’s crazy that Americans still have to put their name, address, bank account number and signature on a piece of paper, and then hand it over to strangers in order to pay their bills. But if they use a credit card instead, the merchant will be charged 3 percent of the total purchase price for the privilege of using the payment “rails” of the two dominant card providers. (Meanwhile, those providers are reaping the rewards: Visa has been the largest financial services company by market capitalization for several years running now.) To make an international payment, a wire transfer will set an individual or small business back $25, and it will (or might not) arrive up to 10 days later.

Third-generation blockchains use negligible electricity, and their potential deployment at TradFi-like scale would reduce power consumption in the financial sector by at least an order of magnitude.

For the impoverished, these costs and other inefficiencies can make a big difference. International remittances, an income lifeline for millions of families in the developing world, cost 6 percent or more of the sums remitted in fees. And Business Insider estimates that in the U.S., the underbanked will spend $230 billion in 2023 on fees and interest charges alone. Don’t get me started on the predicament of the fifth of the global population that doesn’t have access to any kind of financial services.

And then there’s the environmental impact of the technology behind TradFi. While “proof-of-work” blockchains like Bitcoin do use an indefensible amount of energy, it’s nothing compared to the energy consumed in the operations of the traditional financial sector. A basket of the largest banks and asset managers collectively emit something like two billion tons of carbon dioxide annually and, if combined, would become the fifth largest carbon emitter in the world, behind Russia.

Of course, the Bitcoin blockchain processes nothing close to the volume of transactions processed by TradFi. But third-generation blockchains use negligible electricity, and their potential deployment at TradFi-like scale would reduce power consumption in the financial sector by at least an order of magnitude. For example, Algorand, the blockchain with which I am most familiar (I’m the CEO of the Algorand Foundation), currently processes 6,000 transactions per second and only emits the carbon equivalent of a small neighborhood. And, by the way, Algorand buys carbon offsets in order to back its carbon-negative pledge.

One Ledger to Rule Them All

Much could and should be done to reduce the cost and environmental damage wrought by TradFi. But the system is so interconnected that it is virtually impossible for any single intermediary to initiate a meaningful improvement on its own. And while the digital disrupters collectively known as fintech have brought a real upgrade in user experience, the privilege of using the payment systems of the two dominant card providers fintech apps still depend on TradFi.

Public blockchain technology, though, is fundamentally different from the ground up, and the way in which it is different is probably the single most important thing to understand about crypto. In traditional finance, transactions are sent — mostly batched together at the end of each day — from bank balance sheet to bank balance sheet through the correspondent banking system. Cryptocurrency transactions, by contrast, are all written to one shared ledger, and that ledger is called the blockchain. For this reason, even the idea of “sending” cryptocurrency somewhere is a misnomer.

On the Algorand blockchain, for example, you can transfer the equivalent of a million dollars to someone halfway around the world for less than a penny, and they will receive it safely and securely in the same instant you are debited — just 3.7 seconds after you sign the transfer. This low-cost, final-settlement efficiency comes in large part from the fact that nothing is really “sent” or “moved” around at all.

Of course, if all transactions are posted to the same ledger without a trusted third party to verify them, it’s important that these transactions be cryptographically secure (this is where the idea of a “crypto” currency comes from), and it’s very important that no single entity controls that ledger. It’s a lot of power for one institution in an era of declining trust in institutions. A centrally controlled ledger would also be an enticing target for bad actors, making it inherently vulnerable to attack.

What makes a decentralized public blockchain secure, by contrast, is that it lives across a large number of computers (referred to as “nodes”) simultaneously, so that even if multiple nodes are corrupted, the ledger will continue to operate, updating without interruption on all the other nodes. The genius of Satoshi Nakamoto, the elusive creator of the Bitcoin blockchain, was Bitcoin’s incentive mechanism whereby all of these diverse and non-cooperating nodes — some benign and others perhaps malign — could both agree about the order of transactions on the ledger and ensure that no subset of them could change things around. In a word, the ledger would be immutable.

Since then, third-generation blockchains have evolved better and, in particular, faster and more environmentally friendly mechanisms to ensure consensus among the nodes. But in all cases, the bedrock idea is that the integrity of the ledger is achieved without centralized control.

Decentralized Finance: Code Is Law

Decentralized finance, or “DeFi,” tries to replicate the full panoply of basic financial services — savings and investment products, borrowing and lending, derivatives, insurance products, etc. — on global, decentralized and permissionless ledger-based systems, available 24 hours a day, 365 days a year.

Imagine, goes the DeFi thought experiment, that anyone from anywhere could create a financial-service offering using open-source software — and that anyone from anywhere and at any time could make transactions using that service. The value proposition is akin to that of the internet itself. The system is universal and publicly available, mostly without censorship. Its vision, essentially, is that universal DeFi could actually become a new-and-improved third generation of the internet — and it is often called “Web3” — because it would allow for the transfer of value globally, not just information.

To put theory into action, the various important roles played by financial intermediaries would need to be replicated without recourse to those intermediaries. And, in particular, there would need to be structures that would enable various parties, all unknown to each other, to interact with confidence. These structures would need to be replaced by code and programmable for many specific uses. Crypto’s solution to this challenge is called the “smart contract,” which is the most fundamental tool of decentralized finance.

Smart contracts sound fancy, but essentially they are straightforward “if-then” conditional logic statements that govern the release of funds from one crypto wallet to another. The smart contract temporarily takes custody of the funds of the sender and then automatically releases them when specific verifiable conditions are met. (But until those conditions are met, the sender can remove the funds.) This idea is revolutionary because it enables conditional payments and eliminates the risk built into TradFi that one party to a transaction might fail to meet its obligations.

Any transaction requiring money to be held in escrow is a good candidate for a smart contract. But to give one concrete example, the Algorand Foundation is now working to deploy this technology for disaster relief payments. When a family’s home is flooded, family members need relief immediately, but relief agencies still need to be wary of duplicate or fraudulent claims. Using blockchain technology, the family could upload a picture of their flooded living room, together with proof of identity and residency, to the ledger through a simple interface. Once verified, a relief agency’s smart contract would then release funds immediately and without human intervention. Before doing so, though, it would also scan the blockchain’s transparent and complete payment record to make sure that the victim had not already been paid.

The smart contract was dreamed up by the polymath Nick Szabo back in 1994 and popularized by a 19-year-old named Vitalik Buterin in 2011 when he co-founded the Ethereum blockchain. Buterin, arguably the most important person in crypto after the mysterious Nakamoto of Bitcoin, understood that a blockchain could enable more than just a digital store of value like Bitcoin. Rather, it could function more like an iPhone, where all kinds of applications — in this case, smart contracts — could be written to it. And with that realization, the DeFi revolution began.

An Explosion of New Financial Primitives

As it turns out, many financial “primitives” — i.e., core financial services — can almost be entirely replicated using a combination of smart contracts and overcollateralization. And thanks to open-source software, once this caught on, the growth of DeFi primitives was explosive. DeFi burst onto the scene in 2018, and by the beginning of 2022, the top-100 protocols had reached a market capitalization of $120 billion (though it’s fallen off considerably since then).

Take borrowing, for example. Anyone in the world with connectivity can borrow money in DeFi without a credit score or even having to reveal their identity. Credit risk is mitigated for lenders because loans are more than 100 percent collateralized and because smart contracts automatically liquidate that collateral in seconds if the collateral value drops to risky levels. Borrowers can thus access liquidity without having to divest their assets. But you do need to “mind your collateral” as the crypto adage goes, because there will be no warning call from a broker, no grace period for being a good client, and no extra time for you to sort things out.

Arms-length trading in financial assets can also take place in a decentralized manner through a decentralized exchange (DEX). Although the concept of a DEX gets complicated very quickly, essentially central order books are replaced by automatic market makers that provide liquidity in one or more asset pairs. (If you really need to know, an automatic mechanism adjusts the relative amounts of the two currencies so that the product, in most cases, of their two prices remains constant. See? Told ya’ you didn’t want to know.)

And the innovations continue from there. In “atomic swaps,” smart contracts execute multiple legs of a trade across multiple assets simultaneously; if any leg fails to deliver, the trade doesn’t execute at all. In “flash loans” — atomic swaps on steroids — a loan can be taken out and paid off in the same transaction. (This might sound crazy, but flash loans do have useful applications for financial arbitrage and for refinancing trades.) Users can now do things like aggregate liquidity across platforms, manage risk or find leverage through options or perpetual futures trades, engage in prediction markets, and automate insurance — all through interaction with code, not humans.

There needs to be some mechanism of protocol governance, of course, to decide things like the level of fees and rewards, etc. But even here, the ethos of DeFi is to delegate these decisions to the relevant community at large rather than to a top-down decisionmaking body. Being crypto, the vehicle for this needs to sound techy and futuristic, so these protocols are called decentralized autonomous organizations, or DAOs. But, in essence, DAOs are mechanisms for community voting through pure referendum and/or committee. DAO members are typically paid in the protocol’s governance token to participate in voting or other activities. And because of the fact that token holders have to do work, the idea is that they have a much more immediate relationship to the protocol than, say, equity shareholders would have to a corporation in which they have a stake.

In addition to governance, protocols can issue tokens to pay for the provision of liquidity or for providing services (liquidating collateral, for example) to the protocol. And crypto investors engage in “yield farming” across protocols, looking for those that pay the highest interest rewards or have the best prospects for capital appreciation. The quantity of staking, lending and liquidity provision on a DeFi protocol is known collectively as the protocol’s total value locked (TVL). And TVL is one widely recognized indicator of the protocol’s community engagement, level of activity and overall health. The TVL on the most popular DeFi protocols runs into the billions of dollars.

Arguably, though, the more that DeFi talks to the physical world and the more it seeks Main Street adoption, the more it will need to make compromises with centralization in order to protect its users and secure its own viability.

Skeptics argue, with some justification, that the scope of DeFi has been broadly selfreferential to date: one borrows against coins primarily to buy some sort of other coin and one trades from one coin to get access to another coin. Moreover, the experience with DAO governance to date has been mixed at best, with some real success but also with many examples of short-term thinking, neglect and an inability to dig their way through contentious, complicated issues. Imagine your neighborhood online discussion group with a bank account attached to it, and you can grasp the idea.

Nobody realistically thinks decentralized finance is going to completely replace the traditional financial system in either scope or in volume. Nor is DeFi well suited for a range of important financial applications. Most fundamentally, perhaps, overcollateralization, almost by definition, does not allow for the credit creation that occurs through fractional reserve banking. DeFi also has problems with transactions that need to take place over time or that rely to any degree on “good faith” or “reasonable behavior.” And, of course, there are many times when transactions need to be reversed — in the case of legal invalidity, for example, or to correct a “fat-finger” error in entering data.

But DeFi protocols that talk to the physical world are growing in both range and sophistication. Trade and supply-chain finance is one application where funds can be easily held through smart contracts until goods are delivered (verified through internet-of-things scanning), and the “tokenization” of illiquid financial assets is another. A rental property investment, for example, could be tokenized into bite-sized chunks for retail investors so that they could earn some portion of a much larger rental income stream. These kinds of applications are showing real signs of traction in a number of interesting areas. And institutional investors are watching these uses closely as a potentially good bet for DeFi’s next round of explosive growth.

Arguably, though, the more that DeFi talks to the physical world and the more it seeks Main Street adoption, the more it will need to make compromises with centralization in order to protect its users and secure its own viability.

Things Fall Apart; The Center Holds?

Way back (lol) in crypto’s early days, when it was even more difficult to manage crypto assets (bitcoin, really), 70 percent of global trading volume took place on one Tokyo based exchange called Mt. Gox. In 2014, Mt. Gox was hacked and hundreds of thousands of bitcoin were lost, with a dollar value in the hundreds of millions. For the crypto purists, the event was a watershed, and the lesson was fundamental: not your keys, not your coins. That is, it is an individual’s own responsibility to keep the very long and wholly unmemorizable string of characters known as a “private key” somewhere safe.

The lesson was clear, and raw. Your private key is your only access to your money; entrust it to a third party, and you are at its mercy. You can keep your private key yourself. But if you lose it, there’s no 1-800-bitcoin customer support desk to help you get it back. Lose it to someone else, and the money is theirs.

The issue that emerged to disastrous effect last summer was that centralized and opaque crypto-adjacent players made claims and offered services similar to those offered by traditional financial players — but without being regulated as such.

But enterprising market participants at the time also recognized that more userfriendly mechanisms to hold and exchange assets were going to be essential for crypto to grow from its nascent stage. And the widespread adoption that has taken place since 2014 is primarily due to the emergence of centralized exchanges that took the opportunity to offer the same kind of security safeguards, risk profiling and financial reporting that investors enjoy in traditional finance. These exchanges have made crypto investing much easier and safer for retail investors on one level because they serve as custodians for user funds and operate on a more familiar password — and password recovery — system.

Some, like Coinbase and Kraken, both based in the U.S., became regulated (as required), with the same kind of investor safeguards demanded of traditional exchanges. Others, operating offshore, did not. FTX, for example, though easily one of the most widely recognized, was not regulated, and it imploded over the course of five days with, at the time of this writing, little likelihood of customers recovering funds. FTX was a joy to use, until it wasn’t.

Even at the blockchain base-layer where, by definition, the ledger is meant to be immutable and tamper-proof, there have been more than a few instances of centralized decision- making. Just last October, the bridge associated with the Binance blockchain was hacked to the tune of $560 million. Binance shut down the blockchain entirely and halted withdrawals in order to thwart the ability of the hacker to fully withdraw the stolen digital assets. By doing so, Binance was able to limit customer losses to $100 million — but this action also caused an uproar in the crypto community. If the founder of the Binance blockchain could flick a switch and turn off the chain this one time, that meant that they could do it any time they wanted to.

The most famous example of blockchain meddling, though, occurred with the Ethereum investor DAO back in 2016, and it is notable both for its size at the time and for the draconian nature of the remedy. The Ethereum DAO raised $150 million of Ether (then 14 percent of the total supply of Ether, or ETH). But less than three months after its launch, it was hacked, and $60 million worth of ETH was stolen. There was a lot of handwringing about what to do because there was a window for action before the hacker could convert the stolen ETH to cash. And at the time many claimed that the hack was an existential threat to the entire Ethereum experiment. In the end, the Ethereum Foundation hard-forked the chain — that is, it created a new version of the blockchain starting right before the hack. By doing this, it was able to carry on as if the offending transaction had never happened at all on this new version. Investors were made whole, but the move was so controversial that the old, pre-fork version of the blockchain also continues today.

These kinds of maneuvers are not possible on truly decentralized blockchains like Algorand. The trade-off with decentralized chains though is, of course, an inability to intervene in the advent of emergency. But on the whole, crypto is a caveat emptor kind of place. And the overall ethos of decentralized finance, in particular, is that it is obviously scalding hot, so if you are uninformed it’s best not to pour it in your lap.

Three Arrows, Celsius and Voyager advertised themselves as crypto on the one hand, and bank-like on the other. That has turned out to be a very dangerous middle ground for all concerned
Mayhem in the Middle

Because the almost iron-clad corollary to the ideal of decentralization is the admonishment that you are responsible for yourself at the end of the day, there has always been a far more pronounced “do your own research” mantra in DeFi than in traditional finance. But it’s important to recognize that this kind of model also derives from (and is made possible by) the fact that on a public blockchain both the transaction history and the source code of the protocol itself are transparent — open and available to everyone. It’s not always easy to DYOR in DeFi, but it is possible.

In traditional finance, by contrast, opacity is the rule. And because of the fundamental fact that it’s not actually possible for an investor to investigate the books of, say, Credit Suisse, government regulators play a critical role in investor protection. The issue that emerged to disastrous effect last summer, though, was that centralized and opaque crypto-adjacent players (Centralized finance, or “CeFi” to the hipsters) made claims and offered services similar to those offered by traditional financial players — but without being regulated as such.

Michael Hsu, the acting U.S. Comptroller of the Currency, highlighted this issue as one of his biggest regulatory concerns in his opening remarks at a recent fintech symposium in Washington, D.C. One of the most important issues in thinking about regulating crypto, he argued, is that CeFi entities use terminology that consumers associate with regulated banking. Words like “custody,” “ownership” and “APY” have specific meaning in a regulated environment, and CeFi investors too often have expected (and been fooled into thinking) that, say, “yield” means a stable yield or that “custody” means that their assets are ring-fenced. Three Arrows, Celsius and Voyager really are (were!) centralized entities, controlled by individuals. But they advertised themselves as crypto on the one hand, and bank-like on the other. That has turned out to be a very dangerous middle ground for all concerned.

But here’s the other thing about this year’s crypto meltdown: while these centralized but unregulated lending platforms and exchanges made vaguely worded almost-promises that led to countless millions of dollars in losses for unsuspecting customers, the purely decentralized protocols executed exactly as advertised. On DeFi platforms, the loans made to Three Arrows Capital and their like were overcollateralized. And when the value of the collateral went into free fall, smart contracts liquidated it in seconds without lawyers or bankruptcy courts to make lenders whole. In the pure world of decentralized finance, there was nowhere to hide and no Interpol to hide from. The code was the law.