This post is taking part in the Kiwi L2 Writing Challenge
Blockchains can feel like cities: they inhabit a built environment which captures a culture. Like cities, blockchains can keep going even if the city is destroyed because of the relationships that were embedded in it. People can move in and out of cities based on their needs, leaving behind a history, compressed like an asphalt tag into the city’s streets.
Urbanism can give us some useful metaphors and frameworks for thinking about how to design blockchains, especially as it pertains layer 2 scaling solutions.
A popular city is usually dense in its center: diverse, general, and costly. It has commercial, entertainment, residential, offices, etc at its core. For city dwellers, despite it being costly, this diversity is desirable because of its ease. This is the layer 1 of a city. The layer 2 parts of a city is its expanded neighborhoods: whether zoned for residential or industrial or other uses.
A common form of zoning is single-use zoning, otherwise known as exclusionary zoning or Euclidean zoning named after the town of Euclid, Ohio where its constitutionality in the US was first established in a 1926 Supreme Court Case. It is characterized by explicit separation of uses within a city, requiring residents to commute between them. It states (broadly) that: you can *only* have one use in this zone and nothing else.
Another form of zoning that I personally adore is nuisance-based zoning that’s common in Japan. It states that a new neighborhood is zoned based on its maximum nuisance level. Anything else is allowed in that neighbourhood if you can tolerate the nuisance to your liking. For example, you can have a commercially zoned area, allowing shops and retail, and you would be allowed to build homes there, next to your shop if you are willing to tolerate loading trucks at 5am in the morning.
Nuisance-based zoning is often seen as desirable in urbanism because it gives flexibility to citizens, requires less public transport or commuting (you live close to schools, offices, retail, etc), and requires less up-front planning to ensure its success. It’s more self-contained and gives the people in it, the freedom to shape their neighbourhood.
Euclidean Zoning:
vs Nuisance Zoning:
Zoning design and L2 scaling have characteristics in common:
- Both ask the question of how to expand a costly (and diverse) center.
- Both ask the question of whether to design for specificity or generality.
- Both ask the question of how to incentivize people to move away from the center and into an expansion such that all economic and cultural activity is buffeted by this process.
- Both ask the question of how to connect centers and satellite conurbations, if at all.
Generally, all L2s are cheaper, both because of its economic distance to its center (less demand initially), but also due to explicit design tradeoffs.
However, if you were to design L2s and take metaphors from zoning, you’d go about in different forms:
Euclidean L2s:
- focus on specificity at cost of generality. In some instances they technically explicitly exclude certain technical possibilities which is a hallmark of a Euclidean L2. zkSync Lite is an example.
- are more reliant on the L1 as the specific use relies on users willing to commute (bridge) from the center (L1) to the neighbourhood (L2). More often than not, this L2 doesn’t have its own token that might increase complexity. Gnosis Chain is an example as it uses xDAI stablecoin for fees.
- more often than not it relies on its L1 to provide supporting services. A Euclidean L2 likely uses the L1 for data availability in the same way that a neighbourhood might borrow policing, energy, and other resources from its city center to sustain itself.
- A Euclidean L2 likely relies on optimistic fraud proofs (it relies on a nearby court and doesn’t have its own judicial jurisdiction).
Nuisance L2s:
- focus on generality up to a certain scale. It’s as close as possible to a full, diverse, compute environment. In some instances, the L2 might be subsidized (eg, Optimism funding public goods), but it doesn’t exclude other activity explicitly.
- thrives by proximity to its L1, but aims for a certain degree of self-sufficiency. eg, it might have its own token that lives in parallel to its closest L1. For example, OP (for Optimism’s governance) or ARB (for Arbitrum).
- more often than not, it focuses on providing its own supporting services (a local police, firefighting, energy, water, etc). In other words, its data availability might be managed on its own.
- more likely rely on zk-esque validity proofs (equivalent to having its own district court).
Which type of L2?
This graph from Vitalik speaks to choosing along the axes of scale-demanding & security-demanding. That’s one way. But, a key value-prop (for me at least) of a blockchain is its ability to gather history. The problem with more centralized components (like validiums doing off-chain data availability), is that you hamper the long-term viability of it. If we want more cross-pollination and collaboration between L1 and L2 there has to be a longer-term belief that it would remain intact indefinitely.
Thus, I’d lean more towards L2s being zoned for nuisance than euclidean.
YIMBY Urbanists (pro development) often prefer nuisance-based zoning because it generally favors a denser city where citizens can have a variety of their needs in their vicinity.
For example, in the US, when moving to a new suburb, it’s often desirable because it’s cheaper, quieter, and perhaps safer than urban centers. But it comes at cost of genericity: people, shops, culture, and commute often by car. Nuisance-based zoning tries to merge these two into one, allowing citizens to live closer to many diverse mini-centers. In essence, nuisance zoning is more decentralized.
More general, it can be categorized as the choice between planned specificity vs unplanned generality. A well executed planned specific L2 *can* outperform an unplanned general L2. But, blockchains, like cities, are supposed to thrive indefinitely, taking in activity and letting it go again. They *will* change in different ways and over the long-term, it becomes differentiated by its history. While blockchains don’t have to consider geography, it does create a built environment through transactions. Liquidity, for example, can be seen as a city’s cultural and historical network effect. If you’re into book publishing, it helps to be close to New York City. If you’re in film, it helps to be close to LA, regardless of the historical reasons that made those cities destinations for those industries.
Likewise, tokens, tools, protocols, NFTs, and all manner of activity result in that blockchain maintaining a specific momentum. It’s also the case that no matter how much cross-chain UX and technicals improve, it’s still easier to interface with local neighbors on your own chain (and by metaphor, your own city).
With this in mind, if we want more nuisance-zoned L2s, what can help foster that?
To reiterate the trade-offs:
- Design for unplanned generality vs planned specificity.
- Consider incentives for people to build out their own neighbourhood character (like, a separate token).
- Ensure some self-sustaining sufficiency (eg, its own data availability).
- More ZK roll-up than Optimistic roll-up.
This does come across as being more of an L1 than an L2, but I think the key part is what I would propose next. Maybe something for the next 5-10 years of design.
L1s should incentivize and subsidize diversity. L2s should enable diversity, but subsidise specificity.
Part of the value prop of expensive city centers is that it’s cheaper to cross-pollinate in all forms. Thus, if L1s can subsidise transactions where it’s comparatively cheaper to do many things at once, it will attract usage that needs proximity, complexity, and atomicity. Think: transactions that interface many protocols at once. Simpler transactions are still possible, but it might be costlier.
L2s on the other hand, should zone for specificity while allowing generality. In other words, simpler transactions around certain protocols can be cheaper such that it attracts usage in the same way that some neighbourhoods might be general, but have incentives to join: eg, attract students that graduate in the nearby university to start firms in the area. Those young people still need healthcare, food, etc, and so you shouldn’t encourage exclusion. Thus, its nuisance zoned to its maximally desirable nuisance (or it’s maximally desirable complexity).
This means that L2s generally become self-sufficient but work more in tandem with L1s, where L1s become more maximally general and L2s become general, but foster subsidised specificity. It’s the same way that neighbourhoods attract their own culture, industry, and residents that agglomerate and benefit from economies of scale.
Either way, blockchains function very similarly to cities and there are metaphors, tools, and frameworks that can help reason out design choices to ensure long-term cooperation and expansion of use.
If you enjoyed this, please consider subscribing to my weekly newsletter: https://sceneswithsimon.com/