I put “decentralized stack” in quotes because many of Vitalik’s suggestions are in reality significantly more centralized than the “Traditional stack”. The most egregious example is “DAOs”, for example. Almost all DAOs today are dystopian plutocracies where a handful of whales and VCs control everything. Even the ones trying to break free are light years away from traditional organisation structures - particularly cooperatives - which are infinitely more decentralized and democratic than any DAO will ever be. Heck, even the largest public corporations are answerable to democratically elected regulatory bodies, e.g. Apple being forced to adopt USB-C standard, allow sideloading apps, alternative app stores etc. But I digress…

OK, I didn’t want to make this post about criticizing Vitalik’s list, but I write these blog posts improvisational style, so it’s important to note that the “Banking system” is a much larger superset of what the “Decentralized stack” offers. Vitalik notes that we still need banks for loans, but it’s not just loans, it’s really anything subjective - subjective money with monetary policy, dispute resolution for scams, etc. In reality, there’s an overlap between “Banking system” and “ETH, stablecoins, L2s for payments, DEXes etc”, but they also offer different services, so the two stacks are not comparable, and you’ll need both. Yes, you can add the traditional stack on top of the crypto stuff, but then it’s just part of the traditional stack. Finally, I’ll also note that Vitalik is not a fundamentalist and readily acknowledges that most people will mix and match both stacks.

With that said, ETH and native crypto assets absolutely, 100% need strict global consensus! Indeed, this is the dominant usecase, accounting for >90% of crypto’s $ value.

Some of the other usecases - it depends. Stablecoins don’t necessarily need strict global consensus, but strict global consensus has allowed them to frontrun regulations and come to market early. But eventually, there’s no reason a US CBDC cannot offer exactly the same service as a USDT or USDC, but in a more decentralized and efficient manner, cutting out the middlemen (i.e. Tether, Ethereum, L2s etc.). I’ve also argued using ZKPs CBDCs can be fair and democratic - but of course, bad implementations of CBDC can also be dystopian.

Now, even in that world, there’ll be demand for blockchain-based stablecoins, so yes - it does require strict global consensus, but perhaps more as a niche long term.

DEXes - yes, they do need strict global consensus, but also it’s a niche. A vast majority of trading happens on CEXs, and will continue particularly with regulated products like spot ETFs now available. I’ll add DeFi into this camp, and yes - DeFi absolutely needs strict global consensus for its unique properties. Obviously, it’s a niche too.

I’ve already railed against DAOs. I’d argue that a vast majority of DAOs don’t really need strict global consensus. They do need it for making their tokens globally accessible, but that’s about it. Or, if the DAOs are crypto protocols, and the proposals and voting lead to smart contract execution. But a vast majority of things that a vast majority of DAOs do - they don’t need strict global consensus.

Yes, ENS does require strict global consensus! You need a global registry that everyone agrees with. At the same time, I don’t know if this is the endgame, and it’s possible see new identity solutions emerge beyond what public blockchains can do.

I’ve grouped all of these things together, because by and large, most of these do not require strict global consensus, and can be achieved peer-to-peer.

The one exception is Sign in with Ethereum. Though it doesn’t need to write to strict global consensus, it may need to read from strict global consensus. Also, some of these may need strict global consensus for one or two aspects, even if they are mostly done without.

To conclude, we keep coming to the same conclusion, and growing our evidence base - strict global consensus enables objective money and objective identity. Applications requiring these two things can benefit from the strict global consensus property of public blockchains. For nearly everything else (niche usecases aside) - just use peer-to-peer or other traditional methods.