In a recent research paper published in the Federal Reserve Bank of St. Louis Review, I argued that “DeFi [decentralized finance] has the potential to create a truly open, transparent and immutable financial infrastructure.”
The unprecedented speed of financial innovation in DeFi is exciting, and the space is home to some of the smartest people I know. But in the midst of this excitement, it is important not to forget about the risks.
When I say risk, I am not referring to the usual suspects, such as smart contract vulnerabilities or economic attack vectors. I am also not talking about centralization creep, admin keys, oracle dependencies or highly concentrated governance token holdings. All of the above are important aspects, may cause some havoc and certainly deserve our attention.
What I am mainly concerned with is an inherent risk, based on one of the fundamental properties of DeFi. I am talking about the dark side of composability, the vast differences in theoretical and actual transparency and the potential for the amplification of systemic shocks.
In theory, DeFi is highly transparent. The rules (smart contract code) and actions (transactions) are publicly available, and some additional meta-information (events) may help an observer to make sense of the data. This is a great advantage over the traditional financial system, where much of the data is scattered across various proprietary databases and, in many cases, not accessible or to be analyzed in any meaningful way.
I used to be quite optimistic that DeFi could improve on the status quo and have spent significant portions of my early DeFi talks elaborating on why the cryptographic transparency may shape the future of finance. Today, I am still excited and hopeful but less enthusiastic.
Theoretical transparency does not necessarily correspond to actual transparency. Just because the data is publicly available does not mean that people – or machines, for that matter – can make sense of it. In many cases, it is tough to find answers to seemingly straightforward questions and the risk of data getting misinterpreted is considerable.
Let us start with a very simple example. Assume you want to know the current ownership distribution of a given DeFi token. You could easily request the current account balances from the corresponding ERC-20 token contract and use this data to compute ownership statistics.
However, simply looking at the account balances of an ERC-20 token contract would be highly problematic. In many cases, a large portion of the tokens is controlled by other protocols. If, for example, 1,000 tokens are locked in a staking protocol, we must find a way to proportionally assign these 1,000 tokens to the accounts that own a share of the staking pool. Some of the staking pool owners will likely be other protocols. Consequently we have to assign the tokens once again to the owners of that protocol, who may potentially be further protocols.
Each step increases the complexity, introduces additional dependencies and makes the token a little harder to analyze. We end up with schemes that create tokens, on top of tokens, on top of tokens. Once we include negative balances (short selling through lending markets), various sorts of financial derivatives, and realize that there is a potential for recursive effects, we get an understanding of how this seemingly straightforward analysis may pose severe challenges.
In a recent research paper I have written together with Matthias Nadler, we analyzed these adjustments, computed ownership tables for many of the most popular DeFi tokens, and introduced the term “wrapping complexity,” which measures the average number of adjustment – or “wrapping” – steps for a given base token. Whenever this token (or one of its wrapped variants) is locked in a protocol, the wrapping complexity increases.
Although the analysis is just a first step, we are confident that the wrapping complexity can provide valuable insights and be an additional puzzle piece to quantify and observe dependencies and risk aggregation. Moreover, it may give some context to TVL (total value locked), a number that still receives a lot of attention, despite the widespread view that it is somewhat problematic and must be interpreted with great care.
But token composability is just the tip of the iceberg or the peak of the “Misty Mountains of DeFi.” In addition to the vast areas that are covered in mist, there is an extensive yet largely unexplored system of caverns below the mountains. To really understand how everything is connected, one would have to look far beyond the transaction level, analyze internal calls and aggregate the dependencies in a macro model.
I am afraid that we have barely scratched the surface. Considering that the DeFi ecosystem is getting more complex and interconnected each day, our lack of knowledge may be problematic.
Let me be clear: Composability has great effects on financial innovation, and it is one of the defining properties of DeFi. So, I am in no way arguing against composability or openness. What I am saying is that composability in the absence of actual transparency can have devastating consequences.
You don’t have to look back too far to find examples of what can happen when complexity gets in the way of transparency – and I truly believe we should learn from what went wrong in traditional financial markets.
I am concerned that we are in the process of creating a system with a few systemically relevant protocols, some of which have severe external dependencies and are heavily centralized kill switches that may potentially be a threat to the entire DeFi ecosystem.
See also: Jesus Rodriguez – When DeFi Becomes Intelligent
It seems like the speed of financial innovation is much greater than what data analysts can possibly keep up with. This may be a general problem; however, in the DeFi space it is more pronounced. Composability essentially supercharges the effect and transparency is always lagging behind.
On the bright side: All the data is there. It is freely available for everyone to analyze, and there is an increasing number of individuals and organizations trying to make sense of it. So essentially, this is a call to action.
We need more researchers in the space. We need additional tools, better metrics, and we must get rid of our fixation on one-dimensional and potentially problematic numbers, such as TVL. This will help us to navigate the “Misty Mountains of DeFi” and fulfill the promise of a more robust and transparent financial infrastructure.