Sustainability can simply be defined as the protocol remains online, resilient to attacks, and usable under all conditions. Arguably, it also needs to be relevant and keep up with contemporary demands.
Different types of projects have different requirements to attain this
The most challenging, of course, is L1s with proof-of-stake/work consensus
Economic: a) subsidies > network-wide operational costs; b) demand for base asset (demand drivers can be multifarious) ≥ subsidies
Technical: cost of operation grows linearly; cost of verification is ~constant
Social: multiple development teams, plenty of organized individuals that can enforce change
Social sustainability is, obviously, the most important. A strong social layer can overcome technical and economic unsustainability challenges. But a momentarily strong economic & technical project can fail as conditions change without adequate social engagement.
Beyond sustainability, networks can thrive if e.g. demand is greater than supply - this leads to higher economic security. Or, cost of verification decreases over time. Or, it becomes incredibly socially diverse to the point large portions of the developer ecosystem can leave without having a noticeable impact. But, minimum sustainability should be a basic requirement.
IF an L1 can attain sustainability across all three areas, across all possible scenarios, then that opens the door wide open to different types of protocols. Let’s call it sustainability escape velocity.
You can have immutable smart contract infrastructure like WETH which completely inherit sustainability from Ethereum.
The next layer comes immutable smart contracts like Uniswap, but they do have two additional challenges:
a) You need frontends to be sustainable, and have a variety of them
b) It needs ongoing (technical & business) development to be relevant
Both a) and b) can potentially be achieved with a token, though UNI is obviously a poor example of this at this time with no clear sustainable demand drivers. This is where you also need social sustainability - for a motivated community to work on make the UNI token sustainable.
Of course, other projects like Aave, Lido, Maker, ENS etc. are doing better. Perhaps they need to, because they are not immutable, and now you also need your token to be secure enough. To be clear, the need for economic security is not as important as an L1, as you have additional checks and balances possible such as timelocks, strict approval/voting thresholds etc.
There’s a large variety of dapps which will have different demand drivers and different operational expenses & subsidies. Either way, the point is they need to ensure demand matches or exceeds supply.
As a side note, while I’m talking about sustainability here, once again, crypto protocols should be encouraged to be profitable - not just survive, but also thrive. There’s a nefarious crabs-in-a-bucket mentality in crypto where profitable protocols are frowned upon. Without antitrust regulations and strong economies of scale especially for financial dapps, crypto will always tend towards forming oligopolies - this is just a side-effect of crypto’s open nature. Ideally, we should have social layers strong enough to counter oligopolist pressures socially (effectively implicit antitrust measures), although this is probably unrealistic.
Now, let's consider L2s:
An immutable L2 pretty much only needs one sequencer & one honest prover to be live. In reality, though, most L2s are not going to be immutable, some will have multiple sequencers, and are going to spend substantial amounts subsidizing protocol growth (and some like Optimism on other things like public goods [sic] funding). Fortunately, L2s can experiment with different revenue models, particularly application-specific L2s/L3s etc. We already have interesting examples like Immutable X or dYdX who have zero gas fees, but are profitable through trading fees; or Sorare with a direct application-specific business model. For general-purpose L2s, they of course need to consider user transaction fees and MEV, but also look towards novel models, e.g. turnkey L3s on top, dapps directly contract out blockspace etc.
Either way, the same rules apply to L2s: the greater their revenues, the more they can grow and thrive
There’s of course many other types of projects - oracles, public infrastructure etc. Then there’s also a big topic about how restaking can propagate the sustainability escape velocity forward (I won’t cover it because enough people already are) - but I’ll end this post here
Summing it all up, at the end of the day, crypto projects don’t bend the basic rules. Like businesses and nations that came centuries and millennia before, there’s no such thing as conjuring value out of thin air*. It must be earned, through different means appropriate to the protocol.
*To be clear - inflation is fine, as long as it generates greater demand growth