When I finally wrapped my mind around crypto, what I found surprised me. I saw a new computing platform—the so-called Web3—that consumes and generates digital tokens—cryptocurrencies—that incentivize its operation.

What further surprised me was an inconsistency in the way people talked about these tokens and their relationship to this new computing platform. Here’s the dilemma.

The promise of Web3 drives speculation in cryptocurrencies, but speculation in cryptocurrencies should inhibit the adoption of Web3. So either Web3 is a ruse to drive a bubble in cryptocurrencies, or Web3 can’t take off until the bubble bursts.

You can’t have crypto as a speculative asset class and believe in Web3’s future at the same time.

I’ve written this post to explain my thinking and to invite responses. Since this paradox gets lost in the confusion about what blockchains actually are, we start with a (hopefully!) accessible overview of them. From there, we’ll then unpack the dilemma and its implications.

What is a blockchain… actually?

It took me a long while to wrap my mind around what a blockchain fundamentally is. But many years, whitepapers, and youtube videos later, a relatively simple image emerged: a blockchain is like a single, shared, very slow, virtual computer, which no one person really owns, that we pay a cryptocurrency to run programs on and that sometimes pays us in that cryptocurrency to help run them.

I was honestly surprised that that’s what a blockchain was. Notice I haven’t said anything about distributed systems, cryptography, or blocks or chains or proofs of work or stake or anything like that. Just a computer we all share, like a mainframe from 50 years ago.

Something that makes it hard to learn about crypto is that the common terminology exists on different levels of abstraction. A “smart contract” is like a program on this shared computer. Easy enough. But “blockchain” only makes sense if you think of a computer as a state machine, where each “block” is a state transition, and the “chain” is the history of each transition. But now we’re far away from the language of “contracts”. We’ve gone from law (contract) to programming (computer program) to abstract model of programming (state machine) to what that abstract model kind of looks like when you draw it on a whiteboard (blockchain). We’ve taken three conceptual hops!

There’s more. “Distributed ledger” refers to the fact that this computer’s data is held in many places. The “crypto” part of “cryptocurrency” refers to the cryptographic algorithms that underlie these systems that help align incentives to keep it running. You don’t really need to know these terms to understand crypto, but they’re usually the first ones you encounter.

It’s hard to use these terms in the same sentence–it’s like referring to knives by the chemical makeup of their blades in a cooking tutorial. They’re weirdly in the weeds, mostly unhelpful and distracting to people learning about it for the first time.

Web3

Tuning out the noise, one cool thing about blockchains is that they enable these public, shared, ownerless computing platforms. That’s the vision of Web3: an internet that runs on these platforms.

I don’t know if I believe Web3 is the future. I certainly think it’s very cool, but I also thought this was cool so I’m predisposed to liking weird computing things. I’m also guessing there were plenty of people that thought the internet was cool in the early 90s but didn’t think it was worth taking seriously at the time.

Let’s say you believe in Web3. Or even you’re willing to make a bet that it could be huge, so that you don’t feel regret if it blows up and you didn’t invest. But invest in what? The conventional wisdom says you should buy cryptocurrencies. You believe in Web3? Buy ETH.

But here’s the dilemma. Our world is increasingly dominated by computation because computers are cheap, abundant, and getting cheaper and more abundant every year. The cheaper computers get, the more worthwhile applications we can build on them. A bet on Web3 should follow a similar story, that computing on Web3 gets gradually cheaper, so we can build great new applications on it.

But speculating on cryptocurrencies has the exact opposite effect. The more demand for ETH, the higher its price, and the more expensive it is to use the Ethereum blockchain. The more expensive it is, the less it’ll be used, and we’ll have fewer great Ethereum applications. A bet on the rising price of cryptocurrencies is a bet against Web3.

You have to pick though: is ETH a speculative asset class? Or is a token that lets you buy computing on the Ethereum shared computer? It’s this semantic problem I was talking about earlier: because there are so many conceptual leaps to get between the idea of a cryptocurrency and the virtual computer at the heart of the blockchain, it’s easy to miss how deeply connected they are. Crypto’s unapproachable terminology obfuscates the dilemma (by design?). You can’t have it both ways. Either a currency which always increases in value or the universal adoption of distributed computing platforms.

The value of a cryptocurrency stems from its power to purchase computing time on a blockchain. If you believed in the traditional cloud, you wouldn’t hoard AWS credits and watch them go to the moon. Computing getting cheaper is a good thing for more computing. So why would appreciating cryptocurrencies be anything but injurious to Web3?

Postscript: do inexpensive cryptocurrencies help Web3?

AWS stays cheap because as demand for AWS cloud computing goes up, Amazon goes and buys more servers, increasing supply.

But blockchains can’t increase “supply” of computing in the same way. The cloud is highly parallel, millions of computers independently interacting with one another. A blockchain is like a single computer. No single one can handle all the world’s information exchanges. So maybe cryptocurrency appreciation is inevitable, since either price or the blockchain’s physical ability to perform computations will be the bottleneck on its use.

How do we get around this problem? Maybe we’ll add more blockchains to handle increased demand. Or maybe we’ll do more computation “off chain”–on the regular computers we use today–and reserve the blockchain for the important computations we want public records of. Or is this all just a really fun toy that is inherently incapable of scaling? Or is there another solution that I don’t know of yet, or one which is still undiscovered?