Subscribe to Bankless or sign in
Vitalik Buterin just published a new research post, "Hyperscaling By Creating New Forms of State."
The gist: 
Ethereum needs to massively scale execution, data, and state in the years ahead. zkEVMs can do this for execution, peerDAS can do this for data, but state is still in search of a breakthrough.
vbuterin
How state works today
State, of course, is Ethereum's memory. It's essentially a giant database that every block updates and that every node needs to agree on.
This system tracks account info, like ETH balances, and smart contract storage info, like ERC-20 approvals, NFT ownership, DeFi positions, and etc. Additionally, every storage slot is assumed to exist forever unless overwritten.
Ethereum's current model here, then, is "everything is always available." All state is always immediately retrievable, synchronously readable, and composable with every other contract.
This is a powerful and unique offering, but it's also why state is becoming a problem. Every new storage slot increases the size of Ethereum's global state, in turn making nodes more demanding to run.
Why new state forms could work
Ethereum’s current state model works at today’s scale, but left unchanged, it won’t keep pace with the dramatic gains coming to execution and data.
To address this reality, Vitalik just proposed a new paradigm: keep today’s permanent state as a premium storage tier, and add new, cheaper, more restrictive state tiers alongside it.
Toward this vision, Vitalik sketched out what he considered to be the most viable options going forward, namely:
- Temporary storage
A separate storage tree that would be “zeroed out” periodically (e.g. monthly), after which any data stored here could be "resurrected" with proofs. - UTXOs
Instead of “ownership = a permanent slot,” contracts would create discrete records that go directly into history. Ethereum itself would just track minimal permanent “spent/unspent” markers.
It's not practical to replace Ethereum's existing permanent storage system entirely, so the idea would be to keep it as a more expensive premium tier and to complement it with a much more affordable temporary storage or UTXO system, or some sort of mixture of both.
The comparative expensiveness of full, always-on storage would mitigate superfluous state bloat, while temporary storage would give Ethereum the flexibility it needs to support a true 1,000x scaling surge.
What this means for tokens
As Vitalik noted in his post, fungible (ERC-20s) and non-fungible tokens (ERC-721s, ERC-1155s, etc.) "make up the bulk of Ethereum’s state [...] today."
So what would a tiered storage landscape, if it comes to pass, look like for Ethereum tokens?
In that paradigm, the vast majority of state drivers would migrate away from permanent storage, and "NFTs and ERC20 token balances would be [among the easiest things] to move into UTXOs or temporary storage."
What would stay in permanent storage is stuff like user accounts and smart contract code, but most if not all of your favorite tokens would opt to adopt temporary storage.
These migrations won't be driven by alignment so much as plain economic reality. With temporary storage, tokens and the apps around them will be able to enjoy very cheap transaction fees. Why pay more when you can pay less for effectively the same UX?
Indeed, end users won't notice much different besides lower costs or perhaps having to occasionally "resurrect" old token balances or NFTs, which will be as simple providing a proof through your wallet.
What about fully onchain NFTs?
By fully onchain NFTs, I mean NFTs created with no dependencies outside of their smart contract code on Ethereum. Through this code, these NFTs' media and metadata are maintained and generated always-already by Ethereum's runtime.
It's a powerful, unprecedented medium that has become a wellspring of creative advances, e.g. artworks like Terraforms.
The good news is that in a tiered storage system, fully onchain NFTs would continue to operate exactly as they do now, and indefinitely so, since existing permanent storage isn't going anywhere.
Also, creatives will be able to continue to create fully onchain NFTs using permanent storage. The gas costs will just be higher compared to temporary storage so that full, always-on activeness is reserved for things that truly warrant it.
Accordingly, the state of fully onchain NFTs is, and will be, like a glass display case that stays lit forever and will always be immediately viewable and capable of being interacted with.
Conversely, the state of regular NFTs (i.e. PFP projects and so on) that move to temporary storage will become like receipts to an archive: you prove and manage ownership with your records, but the underlying assets won't be kept in the main showroom forever, so to speak.
Zooming out
Ethereum learning how to forget (what it can) may just be the final piece of the puzzle that lets onchain culture use cases break into the mainstream in the years ahead.
For now, this tiered state idea is just a proposal, so it may take a long time to come to fruition if it does so at all. Perhaps some alternative breakthrough will sprout up that the Ethereum community will eventually opt for instead.
But the bottom line is that this vision is practical and sensible and could realistically be a preview of what the future of storage looks like on Ethereum. Better to start wrapping your head around it now as such!
