External Models
Principle #9: Systems build models of external systems
The ninth principle of systems science states that systems build models of other systems that exist externally in their environments.
This piece will explore the role that models play in helping systems achieve their goals by looking at how blockchain-based stablecoins model aspects of the traditional financial system in an attempt to bring stability to the notoriously volatile world of cryptoassets.
Models
A model is an abstract representation of a real system.
All systems develop models of their environment or aspects of the environment with which they interact. These models help systems interact with other systems in their environment in a way that allows them to fulfill their purpose.
The models don’t represent reality, they represent simplified versions of reality. For complex adaptive systems, like cryptoeconomic systems, having detailed and elaborate models of external systems is essential for survival.
Stablecoins
A stablecoin is a cryptocurrency whose value is tied to another asset. One unit of a stablecoin that is pegged to the US dollar is (ideally) always worth $1. Stablecoins have emerged as one of the most popular applications of blockchain technology. They’re used for remittances, trading, savings, and they serve as a core building block of decentralized finance.
One key question for stablecoins and those who design them is, what is the best way to maintain stability?
Let’s explore the differences in how centralized, collateralized, and algorithmic stablecoins model aspects of the external traditional financial system in order to maintain their pegs to the U.S. Dollar.
Centralized PayPal USD
PayPal USD (PYUSD) is a centralized stablecoin which launched on the Ethereum blockchain in August 2023. PayPal believes that “mainstream adoption of future digital experiences will require a stable digital instrument that is crypto-native and easily connected to fiat.”
Like other centralized stablecoins, PYUSD is backed by a reserve of real world assets. Its value is derived from “dollar deposits, US treasuries, and cash equivalents” held by PayPal. The cryptocurrency’s stability is tied to the banking system, regulatory environment, and the ability of users to redeem these stablecoins for real dollars.
PYUSD contains a direct model of the U.S. dollar because it is directly backed by and redeemable for it. Users can exchange PYUSD directly with PayPal for the equivalent amount in fiat currency.
PYUSD also models a key regulatory aspect of the U.S. financial system in the form of a “freeze contract” contained in its code. This allows PayPal to freeze or destroy user funds to ensure the company can comply with know your customer and anti-money laundering laws in the United States.
Centralized stablecoins like PYUSD benefit from leveraging the reputation of tightly regulated traditional institutions. Their primary challenges revolve around trust (is the money really there?), regulatory compliance, and the centralization risks associated with banking institutions.
Any loss of trust in the financial institution issuing the stablecoin can impact its price. For example, when Silicon Valley Bank (SVB) was on the verge of collapse, the USDC stablecoin lost its peg to the dollar because the cryptocurrency was partially backed by dollars held in SVB.
Collateralized DAI
DAI is a decentralized stablecoin which launched on the Ethereum blockchain in December 2017. It was created with the intent of “offering economic freedom and opportunity to anyone, anywhere,” at a time when the world hadn’t yet seen a successful and widely adopted decentralized stable cryptocurrency.
Unlike centralized stablecoins like PYUSD, there is no traditional corporation issuing DAI. DAI is not backed by real world assets held in a bank account. Instead, its stability relies on a complex system of collateral assets, price feeds, and liquidation mechanisms to maintains its peg to the US dollar.
DAI has an indirect model of the dollar because each unit of DAI is backed by crypto-collateral rather than dollars. Users can redeem DAI directly for quantity of cryptoassets such as Ether that are worth one dollar, but they can’t redeem it directly for traditional dollars held in the banking system.
To create one unit of DAI, anyone can lock up cryptoassets as collateral in the Maker Protocol. Crucially, each unit of DAI must be over-collateralized. To create one DAI you must lock up $1.50 worth of Ether in a vault. Over-collateralization ensures that even if the value of the collateral drops, there's still enough value locked up to back the issued DAI. DAI is worth $1 because you can always redeem a single DAI for $1 worth of the underlying assets that were used to generate it.
Price feeds regularly push real-time price data to the Maker protocol. This allows the system to constantly monitor the value of the collateral backing each unit of DAI in terms of USD to ensure it remains above a certain threshold.
If the value of collateral backing DAI in a vault drops significantly, and approaches the value of the borrowed DAI, the system’s liquidation mechanisms are triggered to maintain the peg. The collateral is liquidated, or automatically sold off by the protocol to ensure the DAI remains fully backed.
Decentralized collateralized stablecoins like DAI benefit from decentralization. Anyone can use them without asking for permission, no one has the power to freeze user assets.
Their primary challenges revolve around maintaining the peg through market fluctuations, ensuring the security of smart contracts, and managing the complexities of decentralized governance. DAI briefly lost its peg during a 2020 crypto crash when the value of collateral assets in the Maker Protocol dropped significantly. In response the protocol introduced a feature allowing users to use centralized stablecoins as collateral.
Algorithmic Terra
TerraUSD (UST) is an algorithmic stablecoin which launched on the Terra blockchain in 2020. It was created with the intent of creating a dollar-based stablecoin that was “sufficiently scalable to help blockchain ecosystems scale.”
Unlike centralized and collateralized stablecoins, there are no external assets backing each unit of UST. UST can’t be redeemed for dollars in a bank account or crypto-collateral that exists outside of the Terra ecosystem. Instead, it relies on algorithms and smart contracts to programmatically adjust its supply in response to market dynamics, aiming to keep its price stable.
UST has a highly abstract model of the dollar because its value is tied to market behaviors rather than assets or reserves.
Like DAI, UST relies on price feeds to determine if the currency is maintaining its peg below its peg. Unlike DAI, UST aims to maintain its stability by automatically adjusting its supply based on market conditions.
When demand for UST is high and its price floats above $1, the system increases the supply of circulating UST tokens in order to push the price downwards. When demand for UST is low and the price drops below $1, the system decreases the supply of circulating UST tokens in order to push the price upwards.
At the heart of this system is a mechanism which makes it so that Terra users can always swap the LUNA token (the non-stable native cryptoasset of the Terra blockchain) for UST, and vice-versa, at a “guaranteed” price of $1.
In theory, the lack of need for collateral or direct backing might be a beneficial for a money system. In practice, algorithmic stablecoins don’t have a great track record. Terra experienced an epic collapse last year when UST lost its peg, LUNA hyper-inflated, and trust in both tokens evaporated. UST is currently worth about one cent.
The primary challenges for algorithmic stablecoins revolve around the accuracy of price oracles, the effectiveness of their algorithms in diverse market conditions, and the potential for feedback loops that could destabilize the coin.
The direct models of the dollar embraced by centralized stablecoins have proven to be effective, but they come at the expense of giving up some of the permissionless qualities of decentralized cryptoassets.
Collateralized stablecoin’s indirect models have also been highly successful, although the systems aren’t perfect by any means.
Algorithmic stablecoin’s abstract models have yet to demonstrate their long-term viability.
I’ll have much more to say about models and modeling in the future. For now, remember that models aren’t just something that exist in the minds of humans, or that we create in order to make sense of the world. They exist in all complex systems, and the nature of models held by systems determine how successful they are when interacting with the world.