Cross-Chain Order Flow Auctions

In 2022, crypto infrastructure advanced significantly. Blockspace is now cheaper, more abundant, and more connected than ever. Zero knowledge technology continues to improve scalability, transaction security, and user privacy. The Ethereum Merge exemplified how strong economic incentives can successfully coordinate major technical upgrades, conducted mid-flight without any negative impact to the end user experience. 

Amongst this infrastructure innovation, one area I’m particularly excited about is order flow auctions. Order flow auctions (OFAs) are the financial plumbing of the crypto ecosystem: they coordinate diverse sets of stakeholders to route transactions from the point of origin (e.g. a user’s wallet) to final inclusion in the chain. Along the way, inefficiencies are extracted (often called “maximal extracted value” aka MEV). The more complex the routing, typically the more sizable the MEV opportunities. 

This piece will focus on one particularly emergent category of OFAs: cross-domain markets. Cross-chain OFAs are difficult because of asynchronicity between chains. When two chains commit blocks at different times, conducting transactions between those chains involves latency. That results in subpar execution, additional opportunities to extract MEV, centralization risks, and overall greater market inefficiencies.

So far, there seem to be two paths for dealing with this asynchronicity: slot auctions and pricing premiums. Both answer the question “how do we reduce friction stemming from chains with different block speeds” – but one tackles it by trying to (almost artificially) sync two chains, while the other suggests that buyers will be able discover the right price to finalize their transactions on both chains during the limbo period.

Slot Auctions

Most existing OFAs are block auctions: builders bid with determined block constructions on the right to include those contents as the next block in the chain. A builder’s goal is to build the most profitable block given all existing information (i.e. knowledge of the transactions in the mempool and the current state of the chain). These auctions are considered just-in-time auctions because they occur just before the block gets committed to the chain.

Slot auctions, in contrast, auction off the right to build a block at a future point in time. The slot is purchased without a pre-commitment of block contents. Importantly, slot auctions encompass block auctions – slots can be reserved or bought as “blank” to be conducted as just-in-time auctions at some time in the future.

Slot auctions also expand the set of buyers who might bid in the auction. The following are some examples of different buyer personas and their potential motivations:

  • App layer projects: one example could be an NFT project buying slots well in advance of a mint to improve the probability of successful transaction executions (with the goal of improving user experience). 
  • Builders: a major incentive for builders to purchase slots is the potential for multi-block MEV. When builders control back-to-back blocks, they can strategically manipulate the state of the chain in the first block and capitalize on the opportunity in the second. Detailed examples can be found here and here. I imagine these personas are likely to be larger, more sophisticated builders, who possibly run their own searchers (and thus also have experience finding, creating, and capitalizing on arbitrage opportunities).
  • Validators: validators may buy a slot far in the future with the intent to auction it off as a just-in-time block auction (with the goal of receiving the most profitable block built). They could benefit both from selecting the most profitable block at the time of commitment to the chain, as well as from a time arbitrage (i.e. buying a slot far in the future for relatively cheap with the expectation that demand for that blockspace will exceed what’s being currently priced in; somewhat analogous to buying real estate in an underdeveloped area with the expectation that it will be a much “hotter” area in the future). 
  • L2s: L2s could buy L1 blockspace as a way to build competitive advantage – investing heavily in guarantees of L1 blockspace could crowd out other L2s’ access, which in turn makes one L2’s blockspace comparatively more attractive to projects. 

In cases where the buyer is not explicitly a builder, they would likely need to either contract with a builder or operate their own builder. In fact, it’s possible one way we see this play out is for builders to purchase slots and auction off the rights to those blocks to third parties (apps, L2s, etc), thereby still effectively making these third parties buyers of the slots (just indirectly so).

Importantly, slot auctions are useful for building cross-chain blockspace markets. Purchasing slots for the estimated overlap between chains presents a mechanism to temporarily “sync” the two chains. It’s like if two runners are running on a track at different speeds, occasionally they overlap. Knowing their respective speeds allows an observer to calculate the times at which they’ll be at the same spot. The same idea extends to blockchains: if a buyer knows the speed at which two chains commit blocks, they can predict the points at which those two chains will be in “sync” and purchase the requisite slots on each.

One instantiation of this idea is the Interchain Scheduler, a cross-domain blockspace market for Cosmos chains. Another is Polkadot’s parachain slot auctions (although these aren’t used for cross-chain activity). Slot auctions are still in their infancy. Many open questions remain, from how to price slots to whether slots should be auctioned individually or in bundles to determining if slots can/should include partial blocks. It’s an area of OFAs I find especially interesting and plan to keep a close eye on in the year(s) ahead.  

Pricing Premiums

Another way of dealing with asynchronicity in cross-chain blockspace markets is to use price as a lever for improving the probability of execution. 

This tackles the asynchronicity problem from a different angle than slot auctions: rather than trying to identify the moments at which chains may temporarily sync, pricing mechanisms recognize that one chain will finalize before the other and seek to discover the price at which the bidder is willing to pay to guarantee finality on the second chain once the transaction has been finalized on the first chain. These still act as just-in-time auctions, where step one is to win execution on the first chain, and step two is paying enough to finalize the transaction on the second chain.

The difficulty is finding the right price to pay. For the arbitrageur to be successful, the cross-chain MEV profits need to be greater than the sum of their bids on the two chains. Along that logic, the max bid they should be willing to pay on the second chain is the expected profit minus the bid they paid on the first chain (0 < bid < arbitrage profit – cost on the first chain).1 

Interestingly, this market structure seems to privilege chains whose finalization occurs second. It’s similar to the holdout problem in economics: completion relies on the holdout and, in this case, the second chain (or last chain, if there are more than two) is the holdout. Thus, validators on the last chain to finalize are in the most advantageous position to extract profit.

This opens up a slew of interesting questions: what should the bidder pay for execution on the first chain, knowing they’ll likely have to pay even more on the second chain? When is a pricing mechanism favorable to slot auctions? How does variation in the finalization gap impact the pricing auction? What new centralization vectors does this introduce? And more.

Importantly, cross-chain pricing mechanisms and slot auctions can – and likely will – coexist. Pricing mechanisms bridge the gap during slots in which two chains are out of sync, thereby solving a problem area beyond the solution set presented by slot auctions. As a result, I expect that cross-chain pricing mechanisms will continue to be an exciting area of innovation in the months/years to come.

How Cross-Chain Markets Change Market Dynamics

Cross-chain blockspace markets are interesting, in part, because they’re both inevitable and an unsolved problem. We’re seeing maturation of new chains (both L1s and L2s) with expectations for the emergence of many more (such as L3s). Bridges are improving and apps are going multi-chain. And because blockspace sits at the foundation of the ecosystem, new market structures – like cross-chain markets – create new dynamics throughout the entire stack. The following are a few ways we may see market dynamics shift as cross-domain blockspace markets develop.

#1: Order flow auctions shift from supply-driven to demand-driven markets

Currently, I view blockspace as a supply-driven market, where validators (i.e. those who control the blockspace / supply) can exert outsized influence on stakeholders throughout the rest of the stack. That’s because historically blockspace has been the scarcest resource in the stack. 

But with abundant, accessible blockspace – especially in a world where users may be indifferent to or unaware of what underlying blockspace they’re using – power dynamics might shift to the top of the stack (i.e. the stakeholders where orderflow originates, such as wallets and applications). Attention and user activity may become the new “scarce” resources.

One major implication could be a trend toward exclusive order flow. I’ve already met over a dozen early-stage teams that intermediate user transactions cite interest in “MEV as a revenue source.” Exclusive order flow – where a transaction originator sells private order flow to limited builders down the stack, rather than making it public and letting a large group of builders compete for best execution – isn’t a new concern. In the status quo, however, the risk that the builder doesn’t win the next block (or the next, or the next) acts as a major disincentive. In other words, the risk that a builder’s block won’t be included in the chain means it’s in the wallets’ interest to share the order flow with as many builders as possible to increase the probability of timely execution. Cross-chain markets change this dynamic in two ways. First, they increase access to supply by integrating more chains. More blockspace means a higher probability that the builder can access blockspace somewhere. Second, slot auctions allow third-parties to more accurately forecast whether a certain builder’s block will win inclusion in the chain. Both make exclusive order flow more attractive.

Interestingly, there might be some silver linings to this market structure. One benefit, for instance, is that if wallets capture more value from the MEV supply chain, they could end up redistributing value to users via mechanisms like gas subsidies. Another benefit is that the combination of slot auctions and exclusive order flow could serve to help decentralize the builder network: if smaller builders have the certainty of receiving order flow at certain points in time, it may increase the economic viability of both buying slots and operating a builder. In this world, exclusive order flow may exist, but it might look more like wallets selling order flow to fifty builders rather than five. 

#2: New sources of network effects emerge for L2s 

In a world in which blockspace is abundant, the question for chains becomes how to differentiate their blockspace from another’s. This is especially true for L2s, where many seem to be converging on similar tech stacks for dimensions like scalability, throughput, and privacy. 

Cross-chain markets may create new strategies for L2 defensibility. The first I already outlined above: if slot auctions emerge, we could see L2s buy up significant portions of L1 blockspace in order to crowd out other L2s’ access. I don’t know the extent to which this is economically viable long-term, but it could at least be used as a short-term strategy to attract activity. 

Another interesting source may simply stem from integration into these cross-chain markets. The more connected an L2 is to other chains, the more arbitrage opportunities that may emerge. That in turn may attract more searchers and builders, which can result in better/less extractive execution (because competition reduces the take rate). The better the execution, the more likely users are to want to use the chain. More activity creates more MEV, which in turn drives even more searcher / builder interest. This also makes it harder for new forks of chains to compete with their incumbents; the tech might be replicable, but integrations (and the networks that build around each integration) are much harder to copy.

#3: An increase in centralization pressures

Finally, cross-chain markets risk creating a variety of new centralization vectors. 

One vector is a reduction of competition within each part of the stack. Consider, for instance, that cross-chain arbitrages require storing inventory on multiple chains. The more chains involved, and the more distributed the opportunities, the more capital required. In these scenarios, the larger searchers – i.e those with significant amounts of capital – are likely well-positioned to extract profitable opportunities. And the more they profit, the larger they get. Furthermore, the more smaller searchers struggle, the more the market concentrates toward a few large searcher operations. A similar dynamic could be true of validators, where those with the ability to operate on multiple chains are in a more advantageous position.

Another is centralization throughout different parts of the stack. I expect we’ll see projects start to test different types of vertical integration: 

  • App layer projects could run builders, with the goal of minimizing fee extraction for users
  • Chains could contract searcher-esque parties to run arbitrage strategies to capture MEV for the protocol (such as what Skip is doing for Osmosis and Sei)
  • Builders will likely run searchers in an attempt to create multi-block MEV and/or to hedge against increasing competition in the builder market (running a searcher may help ensure that bundles are directed to their block vs. a competitor’s)

But that’s not to say that vertical integration should be viewed as inevitable. These all involve trade-offs: apps running exclusive builders may be expensive to operate and could result in less frequent execution for users (e.g. a builder may only win inclusion 20% of the time). Capturing MEV as protocol revenue might reduce rewards to validators, which in turn may disincentivize small validator operations and weaken the security (or censorship resistance) of the chain. Nonetheless, possibilities for vertical integration within decentralized networks is a fairly new design space and I expect there to be many experiments and surprises along the way.

Final Thoughts

We’re still in the early innings of blockspace markets. Each new turn of the cycle adds another element to the design frontier: new capital formation (e.g. NFTs) created new types of MEV, the Merge introduced proposed-builder separation (in doing so defined a new segment of stakeholders), and the emergence of multi-chain apps may create more sophisticated mechanisms to internalize MEV. 

I’ve learned a lot from following the progress and research pioneered by teams like Flashbots, Skip, the Ethereum Foundation’s Robust Incentives Group, Informal Systems, and more. If you’re interested in MEV, cross-chain markets, and cryptoeconomic mechanism design, I highly recommend following all of these teams. And if you have thoughts or feedback on this piece, you can always contact me at [email protected] or on twitter.

Thank you to Jesse Walden, Geoff Hamilton, Caleb Shough, Li Jin, Mason Nystrom, and Ibrahim Yusufali for reading drafts of this piece. You can find more of my writing on both the Variant website and my blog.

1I’m actually not sure about this point. Consider the case where the arbitrage profit is $10 and the bid on the first chain is $3. On the second chain, a breakeven bid (assuming the arbitrageur wins the profit) would be $7. If they don’t bid, they lose $3. But there may be a scenario in which they think they need to bid, say, $8 to win – in that case, they either lose $3 or lose $1 ($10 profit – $3 bid – $8 bid). I think this type of scenario – where the searcher knows they are bidding for a loss, but they are bidding to minimize their expected losses – is in reality unlikely to occur because the searcher would need to know the winning bid ahead of time. Nonetheless, it’s still an interesting thought exercise.


Disclaimer: This post is for general information purposes only. It does not constitute investment advice or a recommendation or solicitation to buy or sell any investment and should not be used in the evaluation of the merits of making any investment decision. It should not be relied upon for accounting, legal or tax advice or investment recommendations. You should consult your own advisers as to legal, business, tax, and other related matters concerning any investment. Certain information contained in here has been obtained from third-party sources, including from portfolio companies of funds managed by Variant. While taken from sources believed to be reliable, Variant has not independently verified such information. Variant makes no representations about the enduring accuracy of the information or its appropriateness for a given situation. This post reflects the current opinions of the authors and is not made on behalf of Variant or its Clients and does not necessarily reflect the opinions of Variant, its General Partners, its affiliates, advisors or individuals associated with Variant. The opinions reflected herein are subject to change without being updated.