Hyperliquid's Bold Vision for DeFi | Jeff Yan | S9 E3
Jeff Yan walks through Hyperliquid's origins, its engineering of a high‑performance L1, novel staking economics, MEV mitigation, and a community‑first strategy that balances vertical integration with open‑source composability.
1h 28m·Guest Jeff Yan·Host Monae·
Centralized vs Decentralized Exchange Trade-offs
Centralized exchanges win by abstracting away the heavy technical burden, but they require users to place blind trust in a single entity. Hyperliquid seeks to give users the same frictionless experience while retaining the trust-less guarantees of DeFi. The discussion highlights why users care about custody, innovation speed, and systemic risk.
Centralized exchanges succeed by off‑loading technical complexity from users
- Centralized platforms can cut corners on infrastructure, compliance, and latency, delivering a seamless trading experience.
- By handling validator coordination, order‑book management, and settlement internally, they let users ignore the underlying engineering challenges.
- This model creates a high‑speed, low‑friction product that attracts volume at the cost of user custody.
- The trade‑off is a reliance on a single corporate entity for security and continuity.
Quote
00:00:00“That's why centralized exchanges have been so successful so far is that you get to basically cut a lot of corners or or I guess you just kind of like get to not think about a lot of these problems.” — Monae
Quote
00:38:40“That's why centralized exchanges have been so successful so far is that you get to basically cut a lot of corners.” — Monae
Users must trust a single company with custody and ongoing innovation on centralized platforms
- When you trade on a centralized exchange you are entrusting your funds to one corporate entity.
- That trust extends to the exchange’s ability to keep the platform secure, compliant, and continuously improve.
- Any failure—technical, regulatory, or financial—directly impacts users because there is no fallback decentralised layer.
- This concentration of risk is a core reason many traders look for trust‑less alternatives.
Quote
00:37:02“With the centralized exchange, you're really trusting this one company.” — Monae
Quote
00:37:08“You're trusting them with your money.” — Monae
Hyperliquid aims to combine trust‑less guarantees with a frictionless user experience
- Hyperliquid’s design focuses on delivering the same instant, low‑latency trading experience that users expect from CEXs.
- By building a permissionless L1 with on‑chain settlement, users retain custody of their assets at all times.
- The platform’s architecture eliminates the need for a trusted intermediary while preserving the UI/UX quality of centralized venues.
- This hybrid approach is positioned as the “best of both worlds” for DeFi traders.
Quote
00:24:05“Smart people naturally are attracted to hard problems and we'll solve them all.” — Monae
Quote
00:24:07“I would definitely bet on the global financial system being on a decentralized platform like owned by the people.” — Monae
Hyperliquid Origin: From FTX Collapse to L1 Vision
Jeff Yan explains how his background in traditional finance and crypto prop-trading led to the founding of Hyperliquid after witnessing the failure of FTX. The collapse highlighted the need for a truly decentralized, user-focused exchange, prompting the team to evolve from a simple perp DEX into a purpose-built L1.
Jeff’s background in traditional finance and crypto prop‑trading seeded Hyperliquid’s founding
- Jeff studied math and CS at Harvard, then worked at Hudson River Trading, gaining rigorous market‑making experience.
- After moving into crypto prop‑trading, he saw a gap for a high‑performance, trust‑less exchange.
- This blend of traditional market discipline and crypto agility shaped Hyperliquid’s engineering ethos.
- The team’s small size reflects a “lean startup” mentality rooted in his prop‑trading past.
Quote
00:02:26“I lead Hyperlid development and we're a super small team.” — Jeff
Quote
00:02:34“I grew up in the US, went to school at Harvard, studied math and CS and after school went to do market making in Tradfi, so at a firm called Hudson River Trading.” — Jeff
The FTX collapse exposed the need for a decentralized, trustworthy platform
- The sudden failure of FTX showed how centralized custodians can jeopardize user funds.
- Jeff observed that crypto’s value proposition turned antithetical when trust was placed in a single entity.
- This event catalyzed the decision to build a platform where users retain custody and the protocol enforces solvency.
- Hyperliquid’s mission became “a decentralized platform owned by the people.”
Quote
00:03:21“When FTX collapsed that was the sort of impetus to start thinking about what ultimately became Hyperlid.” — Jeff
Quote
00:03:31“We saw the problems with FTX firsthand and witnessed a mental shift where people realized that crypto is all fun and games until something bad happens.” — Jeff
Hyperliquid started as a Perp DEX and evolved into its own L1 to meet performance needs
- The initial product was a permissionless perpetual exchange (Perp DEX) built on existing chains.
- Limitations in throughput, latency, and global settlement forced the team to build a custom L1.
- The L1 provides a “trusted financial layer” for all on‑chain finance, not just derivatives.
- This evolution reflects a strategic shift from a single market to an infrastructure platform.
Quote
00:08:31“Hyperliquid kind of grew from just a Perex to financial infrastructure powering a Perex, but also other stuff.” — Jeff
Quote
00:07:41“We basically never backed down from the challenge. We're just instead of compromising like okay let's just drill deeper like can we build the next layer of infrastructure below the one we're currently building.” — Jeff
Engineering a High-Performance L1: Hyper EVM and Low Latency
Hyperliquid built a purpose-made L1 to deliver sub-millisecond block times, high throughput, and global accessibility. By adopting the EVM, the team tapped into existing tooling while creating a tightly-coupled core that links order-books, perps, and spot markets. Layer-zero integration enables seamless asset bridging.
A custom L1 was required to achieve the high‑throughput, low‑latency demands of a global DeFi exchange
- Existing public ledgers (Ethereum, Solana, Bitcoin) could not meet the sub‑millisecond latency needed for a professional trading experience.
- Hyperliquid built its own consensus and networking stack to guarantee fast finality and global order‑book consistency.
- The architecture prioritises deterministic execution, eliminating the variability seen in L2 roll‑ups.
- This foundation enables both spot and perpetual markets to coexist on the same chain without bottlenecks.
Quote
00:07:41“We basically never backed down from the challenge. We're just instead of compromising like okay let's just drill deeper like can we build the next layer of infrastructure below the one we're currently building.” — Jeff
Quote
00:07:45“We basically never backed down from the challenge. We're just instead of compromising like okay let's just drill deeper like can we build the next layer of infrastructure below the one we're currently building.” — Jeff
Adopting the EVM gave Hyperliquid instant developer mind‑share and composability
- The EVM provides a massive ecosystem of tooling, libraries, and developer familiarity.
- By building a “Hyper EVM,” the team could reuse existing smart‑contract patterns while extending functionality for on‑chain order‑books.
- This decision also allowed seamless integration with layer‑zero bridges and other EVM‑compatible chains.
- The result is a platform where DeFi builders can plug in without learning a brand‑new language.
Quote
00:09:19“We decided the EVM is the way to go. There's a lot of developer mind share and culture around building applications that interface with other applications through smart contracts.” — Jeff
Quote
00:10:02“Not just to tack on an EVM so people could play around with it. It was a super important component into how users would ultimately end up interacting with these natively built primitives like order books, perps, swap trading, etc.” — Jeff
Layer‑zero integration enables seamless USDT bridging and native asset minting on Hyperliquid
- Hyperliquid partnered with Layer‑Zero to launch USDT on its L1 without traditional bridges.
- The integration leverages the shared EVM compatibility, allowing trustless minting and burning directly on Hyperliquid.
- Users can interact with USDT without ever moving assets off‑chain, reducing latency and bridge risk.
- This demonstrates the platform’s ability to become a hub for cross‑chain liquidity.
Quote
00:10:51“The USDT launch which you know was in collaboration with layer zero so two very big crypto protocols and kind of like coming in and just like plugging into Hyperlid so seamlessly because they're familiar with EVM chains and adding it as another EVM chain in their mesh network.” — Jeff
Quote
00:11:18“They can do all the trustless bridging and minting natively on Hyperliquid without bridging risk and then let users on Hypercore directly interact with their asset in a way that they don't even need to touch because that part is built into the protocol itself.” — Jeff
Staking as a Multi-Utility Tool: Fees, Governance, and HIPP3
Hyperliquid's native token HYPE is used for fee discounts, validator security, and as a governance lever. Staking directly reduces trading fees, while also securing the network and incentivising honest oracle updates. The upcoming HIPP3 framework will enable permission-less perpetual markets that rely on staking for market-making rights.
Staking HYPE directly reduces a trader’s fee rate on Hyperliquid
- Users who lock HYPE tokens receive a proportional discount on their per‑trade fees.
- The discount is calculated on‑chain, ensuring transparency and preventing hidden rebates.
- This mechanism aligns the interests of high‑volume traders with the protocol’s revenue model.
- It also creates a tangible utility for the token beyond speculative value.
Quote
00:13:13“Staking is an input into trading fees.” — Jeff
Quote
00:13:39“You can stake and you have lower trading fees.” — Jeff
Staking secures the network and incentivises honest validators and oracle updaters
- Validators must stake HYPE to be eligible for block proposal, aligning their financial interest with protocol health.
- Oracle updaters are also required to stake, providing a bond that can be slashed for malicious data feeds.
- The slashing mechanism deters double‑signing, false price feeds, and other consensus attacks.
- This dual‑use of staking creates a self‑policing ecosystem without relying on external enforcement.
Quote
00:15:00“Staking also keeps deployers and oracle updaters honest.” — Jeff
Quote
00:15:27“If you act maliciously you can be slashed during the unbonding period.” — Jeff
HIPP3 will launch permission‑less perpetual markets that rely on staking for market‑making rights
- HIPP3 is a framework that lets anyone spin up a new perp market without a central gatekeeper.
- To gain the right to provide liquidity, a builder must stake a minimum amount of HYPE, aligning incentives with the protocol.
- The staking requirement also creates a financial barrier against spam and low‑quality markets.
- This design pushes the responsibility of market curation onto token holders rather than a centralized team.
Quote
00:14:05“HIPP3 was announced recently and there's still a lot of community input going into it and testing on testnet.” — Jeff
Quote
00:14:13“It's a permissionless per, and staking is a big input to this because with per unlike spot like spot order books are well defined and most exchanges start as spot exchanges and only later venture into derivatives.” — Jeff
MEV Mitigation and Symmetric Taker Treatment
Hyperliquid tackles Miner/Maximal Extractable Value by treating takers symmetrically and deprioritising toxic flow. This reduces the incentive for market makers to widen spreads, protecting retail traders. The approach may lower raw volume but improves overall market quality.
Hyperliquid treats takers symmetrically to blunt MEV extraction
- All takers receive the same execution priority regardless of their size or latency advantage.
- By eliminating preferential routing, the protocol removes the profit motive for front‑running or quote‑stealing.
- This design forces market makers to quote tighter spreads, benefiting the broader user base.
- Symmetry also simplifies the fee model and reduces the need for complex anti‑MEV mechanisms.
Quote
00:27:10“We treat takers symmetrically across the board which is that their flow is deprioritized.” — Jeff
Quote
00:27:31“Less volume overall but the volume is sort of better because both sides are kind of always happy.” — Jeff
De‑prioritising toxic flow improves market quality even if total volume drops
- By giving each participant a second to decide whether to trade, the protocol reduces aggressive sniping.
- The net effect is a modest reduction in raw throughput but tighter bid‑ask spreads.
- Retail traders experience less slippage and lower execution costs.
- This trade‑off aligns with Hyperliquid’s goal of a fair, user‑centric market rather than pure volume maximisation.
Quote
00:27:31“Less volume overall but the volume is sort of better.” — Jeff
Quote
00:27:44“Retail traders aren't screwed; the design protects them from widened spreads.” — Jeff
Retail traders are protected from widened spreads caused by MEV and toxic market makers
- When market makers can pick off stale quotes, they widen spreads, increasing costs for small traders.
- Hyperliquid’s symmetric taker policy forces market makers to keep quotes tight, limiting the need for them to widen spreads.
- The result is a more predictable execution price for retail participants.
- This protection is a core part of the platform’s value proposition for everyday users.
Quote
00:26:21“Market maker is basically like a user and that user gets screwed either way and from their perspective the reaction is to widen their quotes.” — Jeff
Quote
00:27:44“Retail traders aren't screwed; the design protects them from widened spreads.” — Jeff
Permissionless Builder-Deployed Perps and Market Curation
Hyperliquid's protocol is permissionless, allowing anyone to launch a perpetual market after staking. Builders gain market-making rights through a bidding process, while core markets remain curated for stability. This hybrid model balances openness with quality control.
The protocol allows permissionless deployment of perpetual markets
- Anyone can submit a market specification and, after staking the required HYPE, have it deployed on‑chain.
- This opens the platform to a diverse set of assets, from niche tokens to experimental derivatives.
- The permissionless nature encourages rapid innovation without a central gatekeeper.
- The on‑chain deployment ensures transparency and auditability of market parameters.
Quote
00:51:27“Permissionless protocol... people can deploy per.” — Jeff
Quote
00:52:04“Builders can create markets without capital, just by staking.” — Jeff
Builders must stake HYPE to earn market‑making rights, aligning incentives
- To obtain the right to provide liquidity on a new perp, a builder must lock a minimum amount of HYPE.
- The staking amount is bid against other interested parties, creating a market for market‑making slots.
- Successful bidders receive the exclusive ability to post orders, ensuring they have skin in the game.
- This mechanism discourages spam markets and aligns builder profit with protocol health.
Quote
00:52:30“You have to stake to get the right to make a market.” — Jeff
Quote
00:53:24“The intention is not to have a competition... it's a permissionless protocol where people can deploy per.” — Jeff
Core markets stay curated while permissionless extensions expand the ecosystem
- Hyperliquid maintains a set of core perpetual and spot markets that are carefully selected for liquidity and stability.
- Permissionless builder markets coexist alongside these core markets, offering niche exposure without compromising overall user experience.
- The protocol can adjust fee structures or staking thresholds for builder markets to ensure they meet quality standards.
- This hybrid approach provides both reliability for mainstream traders and openness for innovators.
Quote
00:51:27“Permissionless protocol... people can deploy per.” — Jeff
Quote
00:53:24“The intention is not to have a competition... it's a permissionless protocol where people can deploy per.” — Jeff
Vertical Integration vs Modular Ecosystem Design
Hyperliquid deliberately integrates only the layers it deems essential--order-books, perps, and the Hyper EVM--while leaving higher-level applications to the community. The team believes that a resilient DeFi stack requires multiple independent options at each layer, fostering robustness and innovation.
Hyperliquid selectively integrates core components while leaving higher‑level apps to the community
- The team builds the order‑book, perpetual engine, and Hyper EVM in‑house to guarantee performance.
- Everything above that layer—wallets, UI, analytics—is expected to be built by third‑party developers.
- This reduces the engineering burden on the core team while encouraging a vibrant ecosystem of complementary services.
- It also ensures that the protocol’s most critical code paths remain tightly controlled and audited.
Quote
00:54:54“Vertical integration is obviously super good as a business decision.” — Jeff
Quote
00:56:29“We never seriously entertained building everything... we focus on coordination through protocols not through a 10,000 person company.” — Jeff
Decentralization across stack layers creates a more robust financial system
- When each layer (consensus, execution, UI) has multiple independent implementations, a failure in one does not cripple the whole system.
- This redundancy mirrors traditional financial infrastructure where backup systems exist at every tier.
- Hyperliquid’s philosophy is that the “weakest link” determines user experience, so diversifying options mitigates systemic risk.
- The approach also encourages competition and innovation at every layer.
Quote
00:55:31“DeFi is more resilient if there are multiple possible choices at each layer of the stack.” — Jeff
Quote
00:57:06“Coordination through protocols not through a 10,000 person company.” — Jeff
Community‑first growth is a strategic choice over building a massive in‑house product suite
- The team believes that a small, hungry team can iterate faster than a large organization with bureaucratic layers.
- By empowering external developers, Hyperliquid leverages the global talent pool without the overhead of hiring.
- This strategy aligns incentives: builders earn fees, the protocol gains liquidity, and users get diverse products.
- The ultimate goal is to “grow the pie” rather than capture a fixed slice.
Quote
00:58:00“We want to grow the pie.” — Jeff
Quote
00:37:44“We have zero marketing department. It's completely organic and community run.” — Jeff
Token Economics: HYPE vs Stablecoins and Buybacks
Hyperliquid prioritises USDC as the primary trading pair while the HYPE token serves as a utility for fee discounts, governance, and buy-back funding. The token is not positioned as a money-like asset; instead it underpins network incentives and aligns stakeholder interests.
USDC is the core trading pair; HYPE is a secondary utility token
- Hyperliquid launched with USDC‑denominated spot markets because users demand dollar‑stable trading.
- HYPE is introduced later to provide fee discounts, staking rewards, and governance rights.
- This separation keeps the primary liquidity source stable while allowing the native token to accrue value through protocol incentives.
- The design mirrors other L1s that bootstrap with fiat‑pegged assets before expanding token utility.
Quote
01:01:10“Stablecoins vs HYPE token... we started with USDC at the core because that's what our customers wanted.” — Jeff
Quote
01:04:46“We do have an exchange generating a ton of fees so we're going to do buybacks.” — Jeff
HYPE token is used for fee discounts, governance, and buy‑backs rather than as a primary medium of exchange
- Holding and staking HYPE lowers a trader’s fee rate, providing a direct economic benefit.
- Governance proposals are voted on by HYPE holders, giving the community a voice in protocol upgrades.
- A portion of exchange fees is allocated to buy‑back and burn HYPE, aligning token scarcity with platform revenue.
- Jeff emphasizes that “you can’t really build for money,” positioning HYPE as a tool, not a speculative asset.
Quote
01:03:14“You can't really build for money. It's just not actionable.” — Jeff
Quote
01:04:46“We do have an exchange generating a ton of fees so we're going to do buybacks.” — Jeff
Staking HYPE creates a virtuous cycle of lower fees and increased protocol participation
- Traders who stake HYPE enjoy reduced fees, encouraging higher trading volume.
- Increased volume raises fee revenue, part of which funds further buy‑backs, boosting token price.
- The staking model also locks value into the protocol, improving security and aligning long‑term holder interests.
- This feedback loop is central to Hyperliquid’s sustainable economic design.
Quote
00:13:13“Staking is an input into trading fees.” — Jeff
Quote
00:13:39“You can stake and you have lower trading fees.” — Jeff
Community-Driven Organic Growth
Hyperliquid's rapid ascent is attributed to a tiny, highly motivated team and a passionate community that markets the product organically. The lack of a formal marketing department forces the project to rely on word-of-mouth and developer evangelism, which Jeff believes is a competitive advantage over centralized exchanges.
Growth is driven by an organic community rather than a traditional marketing department
- Hyperliquid has no dedicated marketing team; user acquisition comes from community advocacy and social media.
- The team’s small size (≈10 people) forces a focus on product excellence, which naturally attracts users.
- Community members actively promote the platform, creating a grassroots network effect.
- Jeff cites this as a “blessing in disguise,” giving the protocol authenticity and credibility.
Quote
00:37:44“We have zero marketing department. It's completely organic and community run.” — Jeff
Quote
00:37:49“The community does an amazing job, better than the marketing departments of all these centralized exchanges.” — Jeff
A tiny, highly skilled team enables rapid feature deployment and iteration
- The core team consists of roughly ten engineers and non‑technical staff, allowing fast decision‑making.
- This lean structure lets Hyperliquid ship new products (e.g., USDT launch, HIPP3) in months rather than years.
- Jeff describes the team’s size as a “blessing in disguise,” giving them agility that large exchanges lack.
- The approach aligns with the ethos of building a “fun thing to solve” and attracting smart people to hard problems.
Quote
00:41:15“We're a super super small team. We're only 10 people.” — Jeff
Quote
00:24:05“Smart people naturally are attracted to hard problems and we'll solve them all.” — Monae
The vision of a decentralized global financial system resonates with the community, fueling participation
- Monae bets on the global financial system being owned by the people, a sentiment echoed by many listeners.
- This ideological alignment motivates developers to contribute code, liquidity, and marketing effort.
- The community’s belief in DeFi’s long‑term potential creates a self‑reinforcing loop of adoption and improvement.
- Jeff sees this as a core driver for Hyperliquid’s early success and future growth.
Quote
00:39:03“I would definitely bet on the global financial system being on a decentralized platform like owned by the people.” — Monae
Quote
00:24:05“Smart people naturally are attracted to hard problems and we'll solve them all.” — Monae
Future Scaling: Multiple Proposers, Fast Blocks, and Latency Trade-offs
Hyperliquid is investigating multiple concurrent proposers to cut latency and further diminish MEV. Faster block times reduce the window for front-running, but they increase network overhead. The team believes hardware advances will keep the system scalable.
Multiple concurrent proposers are being explored to lower block latency
- By allowing several validators to propose blocks simultaneously, the protocol can achieve sub‑10 ms finality.
- This design mirrors research from Salana and other L1s that aim for ultra‑low latency.
- Jeff references a paper by Max Resnik and Anatoli discussing the feasibility of such a system.
- The approach promises faster order matching and reduced MEV extraction opportunities.
Quote
01:18:13“Multiple concurrent proposers... could actually quote or you can do blocks faster somewhere around 20 milliseconds is their idea.” — Jeff
Quote
01:18:22“Low latency ecosystem... it's a world where networks are super fast you don't care about additional hops.” — Jeff
Faster block times dramatically shrink the window for MEV exploitation
- When blocks are produced every millisecond, a front‑runner has almost no time to reorder transactions.
- The reduced time between block proposals limits the profit potential of both on‑chain and off‑chain latency‑based attacks.
- Jeff notes that with 1 ms blocks, “you can't really mess with it” because there’s no wiggle room.
- This creates a more level playing field for all traders, especially retail participants.
Quote
01:33:26“If blocks are one millisecond for example then it doesn't really matter how dishonest you can be in a block.” — Jeff
Quote
01:32:00“Fast blocks reduce MEV because the amount of MEV in a block scales with the time between blocks.” — Jeff
The latency gains come with increased network overhead, but hardware scaling mitigates the cost
- Introducing multiple proposers adds extra communication steps, slightly raising bandwidth consumption.
- Jeff argues that modern networking hardware can handle the extra traffic without a performance penalty.
- The trade‑off is acceptable because the security and fairness benefits outweigh the modest latency increase.
- This perspective aligns with Hyperliquid’s philosophy of solving engineering problems rather than accepting compromises.
Quote
01:31:22“There will be a constant factor latency increase and probably increases the sort of networking overhead of the network as a whole.” — Jeff
Quote
01:32:31“Hardware just scales super well. So I feel like this should just happen.” — Jeff
⚙ Agent-readable JSON index — click to expand
{ "memcast_version": "0.1", "episode": { "id": "xHGKv0Cx7ao", "title": "Hyperliquid's Bold Vision for DeFi | Jeff Yan | S9 E3", "podcast": "Bell Curve", "guest": "Jeff Yan", "host": "Monae", "source_url": "https://www.youtube.com/watch?v=xHGKv0Cx7ao", "duration_minutes": 89 }, "concepts": [ { "id": "centralized-vs-decentralized-exchange-trade-offs", "title": "Centralized vs Decentralized Exchange Trade-offs", "tags": [ "centralized-exchange", "user-experience" ] }, { "id": "hyperliquid-origin-from-ftx-collapse-to-l1-vision", "title": "Hyperliquid Origin: From FTX Collapse to L1 Vision", "tags": [ "prop-firms" ] }, { "id": "engineering-a-high-performance-l1-hyper-evm-and-low-latency", "title": "Engineering a High-Performance L1: Hyper EVM and Low Latency", "tags": [ "latency", "layer-2", "evm", "custom‑L1" ] }, { "id": "staking-as-a-multi-utility-tool-fees-governance-and-hipp3", "title": "Staking as a Multi-Utility Tool: Fees, Governance, and HIPP3", "tags": [ "global-governance", "staking", "hip‑3" ] }, { "id": "mev-mitigation-and-symmetric-taker-treatment", "title": "MEV Mitigation and Symmetric Taker Treatment", "tags": [ "toxic-flow", "mev" ] }, { "id": "permissionless-builder-deployed-perps-and-market-curation", "title": "Permissionless Builder-Deployed Perps and Market Curation", "tags": [ "staking", "market-curation" ] }, { "id": "vertical-integration-vs-modular-ecosystem-design", "title": "Vertical Integration vs Modular Ecosystem Design", "tags": [ "ecosystem", "decentralization", "modular-design", "systems-integration" ] }, { "id": "token-economics-hype-vs-stablecoins-and-buybacks", "title": "Token Economics: HYPE vs Stablecoins and Buybacks", "tags": [ "stablecoin", "compute-economics", "ai-hype", "buybacks" ] }, { "id": "community-driven-organic-growth", "title": "Community-Driven Organic Growth", "tags": [ "community-collaboration", "organic‑farming" ] }, { "id": "future-scaling-multiple-proposers-fast-blocks-and-latency-trade-offs", "title": "Future Scaling: Multiple Proposers, Fast Blocks, and Latency Trade-offs", "tags": [ "latency", "mev" ] } ] }