Last week, crypto derivatives exchange dYdX announced that it will be leaving the Ethereum ecosystem and launching its own blockchain within the Cosmos ecosystem. According to dYdX’s founder, a new chain will allow the platform to provide the best possible experience for its users – enabling the platform to more easily customize things like fee structures and transaction speeds.
The new chain will replace dYdX’s current platform built atop StarkWare, an Ethereum scaling solution platform that uses ZK-rollup technology to allow for quick and cheap transactions.
Even as layer 2 networks like StarkWare are expanding Ethereum’s capabilities at a rapid pace, updates to the core Ethereum protocol are lagging, and competition from other smart contract ecosystems is growing fiercer by the day.
DYdX’s decision to leave Ethereum has been viewed by some as evidence that the original smart contract network simply is not moving fast enough to accommodate the demands of a maturing crypto ecosystem.
DYdX’s path – which saw the platform outgrow Ethereum’s layer 1 blockchain, move to StarkWare, and then leave Ethereum altogether – provides insight into two competing visions for the future of crypto: the app chain versus the global computer. It’s also a case study into the weaknesses of Ethereum layer 2s, which are generally viewed as a saving grace for a network that has famously struggled to scale.
The decentralized order book
The advantage of decentralized finance (DeFi) is that it enables users to transact without any intermediary. In the case of a decentralized exchange (DEX), this means users can buy and sell assets without a bank dictating prices and taking fees. DYdX might not have the same name recognition as DeFi giants like Uniswap, but it has quietly grown into a major force within DeFi due, in part, to its ability to facilitate large trades without slippage.
Slippage is a quirk of automated market makers (AMMs) – the go-to technology that powers decentralized exchanges, such as Uniswap and Sushi, behind the scenes.
AMMs were one of the early DeFi innovations enabling users to exchange currencies without middlemen. With AMMs, buyers and sellers do not dictate token prices. If a user wants to swap one token for another on an AMM-based exchange, they are linked up to a liquidity pool containing a mix of both currencies. To swap, say, USDC for ETH, a user drops some ETH into a USDC/ETH pool and is given an equivalent amount of USDC from the pool in exchange.
A simple mathematical formula determines the exchange rate based on how many tokens of each type sit in the pool.
Slippage happens when a swap is large enough to throw the ratio of currencies in a pool way out of whack – distorting the exchange rate. While AMMs are generally useful for retail traders, slippage can render them useless for some institutional-sized swaps.
DYdX avoids all of these issues by using a more traditional order book model to facilitate swaps, directly linking buyers and sellers of tokens and contracts. This method removes slippage, which means dYdX has proven better for institutional-sized trading.
On the negative end, maintaining an order book (a list of buy and sell orders) and directly matching up counterparts can require more computation (and therefore higher fees) than the simpler AMM-type systems.
We won’t wade into the debate of order books versus AMMs here – the order book also has its disadvantages.
The most important takeaway is that order book DEXs like dYdX are particularly maladjusted to the slow speeds and high gas fees of networks like Ethereum. These sensitivities are what originally drove dYdX away from Ethereum’s layer 1 mainnet to the Ethereum layer 2 StarkWare.
Beyond layer 2
Ethereum layer 2s like StarkWare were supposed to come to the rescue for platforms like dYdX.
In general, layer 2s expand the capacity of blockchains like Ethereum by processing transactions on a separate blockchain. These separate chains execute transactions, bundle them up, and hand them back down to the layer 1 chains where they get officially recorded onto the ledger.
DYdX explained why it initially moved over to the platform in a 2020 blog post: “Ethereum can process around 15 transactions per second (TPS), which is not enough to support the hypergrowth of DeFi, NFTs, and more. While Ethereum 2.0 will theoretically boost network speeds to 100,000 TPS, base layer scaling is still a while away. In the meantime, Layer 2 scaling solutions – in the forms of Rollups – free up Ethereum’s base layer by offloading execution, leading to reduced gas costs and increased throughput without increasing network load. StarkWare’s dYdX integration combines STARK proofs for data integrity with on-chain data availability to ensure a fully non-custodial protocol”
When they hand the transactions back down to the layer 1 chain, layer 2s tend to use fancy math and other strategies to prove that the transactions are “true” – meaning they haven’t been faked or tampered with. DYdX’s layer 2 solution of choice – StarkWare – leverages a technology called STARK proofs to significantly decrease fees and increase speeds.
But layer 2 solutions also have their weaknesses.
One of the oft-mentioned shortcomings of layer 2s is that they rely on a single node operator – or sequencer – to coordinate whatever activity gets passed down to the layer 1 network.
Sunny Aggarwal, the co-founder of the Cosmos-based Osmosis exchange, explained to CoinDesk, “Almost every rollup platform right now is all single operator – whether it’s Arbitrum, Optimism or StarkWare.” In other words, one company or computer, rather than a distributed network of node operators, is responsible for bundling up the transactions that eventually get passed down from the layer 2 chain to the mainnet.
According to Aggarwal, “Rollup systems give you safety – like, you can trust the execution of the code is correct – but they don’t give you liveness nor censorship and front-running resistance.”
Aggarwal’s main contention is that layer 2s – while they have undeniable speed and cost advantages relative to Etheruem’s mainnet – tend to suffer from a single point of failure as a result of their reliance on sequencers.
If, say, StarkWare decided to censor certain transactions before passing them down to Ethereum’s mainnet, they could do so. A layer 2 sequencer can also go offline, and (if they are malicious) front-run transactions – previewing buy/sell activity to squeeze out better deals for itself.
While these theoretical concerns may not seem especially prescient to everyone, Aggarwal noted that with Osmosis, “our thesis has always been that eventually, applications are going to get big enough that they’re going to want to go on to their own chains.”
Safety vs. sovereignty
The dYdX move underscores a difference between Cosmos and Ethereum, ecosystems which posit distinct visions for the future of crypto.
Ethereum positions itself as a kind of global computer. Anyone can build programs that run on this computer, and the security of the entire system extends to each of these apps.
Cosmos’ vision of the future is one of so-called app chains: blockchains that are purpose-built for specific use cases. Rather than one specific blockchain upon which many apps are built, Cosmos is a family of distinct blockchains that easily communicate and swap assets back and forth.
The primary advantage of a system like Cosmos is customizability. By spinning up your own Cosmos chain, you can set parameters specific to a given use case.
One way dYdX plans to take advantage of this customizability is by making platform fees scale with transaction size, which is more like how a centralized exchange works. Today, gas fees are based on network traffic and computational complexity – they don’t increase or decrease depending on how much money is at stake. DYdX also plans to customize how blocks are issued in its new system. By moving onto a new chain, dYdX says its core technology will be better optimized to handle its order book exchange model as it grows.
The primary disadvantage of the app chain model proposed by Cosmos tends to be security. On Ethereum, thousands of computers compete to add blocks to the chain and validate transactions. This heavily decentralized security model is one of Ethereum’s core selling points.
Most Cosmos chains, on the other hand, have dozens – not thousands – of nodes keeping things secure. Cosmos chains tend to have vibrant governance communities, but issues have, from time to time, slipped past the Cosmos validators responsible for chain upgrades and security.
By moving to Cosmos, dYdX is charting its path in a world where sovereignty – rather than absolutist guarantees around security – are prized above all else.
Dan Edlebeck, the founder of Cosmos-based Sei Network, told CoinDesk that on Cosmos, “You can guarantee you are in more control of your chain itself.”
While centralized security levers may sound like sacrilege to decentralization maximalists, Edlebeck describes it as a feature rather than a bug. As he explained, “You can make customizations at the validator level – whether it’s their geographic location, or whether it’s the technical specifications that they need to have to be able to run your validator – you can customize your chain for your own needs.” According to Edlebeck, these security customizations allow chains to run more efficiently without sacrificing decentralized security in its entirety.
Other visions for the future of crypto (and crypto security) exist as well: Polkadot has a unique, multichain “hub-and-spoke” model for sharing security between distinct blockchains. Solana takes a more centralized approach to security, but it claims to provide a more streamlined experience for app developers and users.
While each vision has its maximalists – it seems likely that, at least for the foreseeable future, they will need to coexist. But we should expect to see more projects leaving (and joining) Ethereum as crypto’s uncertain future continues to take shape.