State of the application layer

In early 2022, I had this phase where I was focusing on discussing application layer innovation. Right before then, in late 2021, my main focus was scalability, so it was apt to make amends and point out all of that scalability would be pointless if there wasn’t orders of magnitude higher user demand for applications.

Nearly a couple of years later, regrettably, very little has changed. There’s now an overabundance of L2s and L1s with barely utilized blockspace, yet people are still obsessed about infrastructure (understandable, plenty of L1/L2 bagholders).

One of my biggest peeves about the crypto space is this constant sense of entitlement and hopium, misguided on irrelevant analogies. No, just because you build blockspace does not mean it’ll be used - it’ll only be saturated if there’s meaningful applications that offer a 10x improvement over incumbents, and there’s significant user demand for it. There have been dozens of smart contract blockchains live for 5-7 years now - an absolute eon in the digital age - almost all of which have negligible non-spam activity. We’ve had multiple Stage 1 decentralized app-specific Ethereum L2s and even smart contract (Arbitrum One) for over 3 years now - likewise, very far from being saturated. At this point, there’s overwhelming evidence that blind entitlement and hopium will not lead to blockspace saturation.

I’m always open to evolving my framework of where blockchain applications make sense, but currently, all evidence points to the reality that using blockchains has a very narrow window where it makes sense. Specifically, it must meet all of these criteria:

  1. Peer-to-peer operation: this includes P2P software like IPFS, BitTorrent, multiplayer games, messaging apps, and yes, blockchains. It’s important to note that if you just want to eliminate a centralized entity and decentralize an app, you can simply go P2P without using blockchains in most scenarios.

  2. Strict global consensus: this is the unique property blockchains bring to the table - strict global consensus, so every peer in the world agrees on exactly the same outputs.

  3. Objectivity: however, a limitation of blockchains is it can only parse objective results.

I’ve further surmised this means there are two major categories where blockchains make sense - objective money and objective identity. Now, of course, you can build various different applications that are otherwise non-blockchain, but use blockchains only for objective money and/or objective identity, hybrid applications. You can also use governance to add subjectivity.

Farcaster is a great recent example of a hybrid application. It uses blockchains for objective identity (identities) and objective money (fees) - while everything else (to the best of my knowledge) is coordinated outside of blockchains. As I’ve discussed before, there’s a spectrum, and we may see applications which are mostly centralized, but with one blockchainy element.

Examining the application layer today, outside of very few exceptions (Farcaster) - it has barely progressed. At the same time, the existing applications are establishing themselves. We know alternative store-of-value is still the #1 application. Stablecoins have solidified themselves as the #2 usecase, with $25+B being transacted every day through the bear market. It’s pretty obvious by now these are the two dominant usecases of crypto, and will most likely account for >90% of the economic value for the foreseeable future. A very distant #3 is gambling and speculation, though #1 has an element of it too.

A couple of years ago, I was 50/50 on improving the proven product-market fits, while continuing to push on application layer innovation. As time has gone on, it seems pretty evident that we should actually be spending more time and capital on proven things like improving stablecoins, DeFi, store-of-value etc. For example, outside of Bitcoin and Ethereum, USDT on Tron is the dominant application. I have written a post about how we can improve on this usecase. In short, my recommendation today is to abandon most infrastructure projects, only pursue new applications that actually make sense, while direct a majority of investment into expanding usage of stablecoins, store-of-value and such.

To be clear, there will be application layer innovation, but it’s also important to acknowledge the reality that most of the applications where blockchains made sense have already been built.

I’ll end the post by highlighting one exception - fractal scaling. Fast chains - L1s, L2s doesn’t matter - can do a few hundred to a few thousand TPS (depending on the complexity of each transactions). However, for certain applications you need hundreds or even thousands of these chains. Think about traditional applications sharded or meshed across thousands of servers. One example of this is a fully onchain multiplayer game. If one hits a scale of millions of users, you’re going to need thousands of today’s fastest chains parallelly. Today, the infrastructure for this does not exist, but it is well under development by multiple teams. We know the best way to do this - zk rollups/validiums with aggregated proofs. Whether these applications see product-market fit, I don’t know, but it’s certainly worth trying. Most other infra projects that only shill scalability - we’ve already wasted dozens/hundreds of billions of dollars on (relative) ghost chains. It’s years too late to time to move on.

Subscribe to polynya
Receive the latest updates directly to your inbox.
This entry has been permanently stored onchain and signed by its creator.