Let’s walk through the scalability roadmap for the Ethereum ecosystem, but from an end user’s point of view. There’ll certainly be plenty of minor updates, but I’ll cover the major ones here. I go into details of how each upgrade will impact the user experience, but feel free to skip and see the Tl;dr for each. I think there’s still a lot of misconceptions in this space about how “People will move to Ethereum 2.0 which will lower gas fees, but it’s years away”. Lastly, all my content is in the public domain, please feel free to share, adapt, copy as you wish. I have removed all links, so the post is not deleted — you may have to do some searching to find out more — apologies.
Now: Smart contract rollups
Interestingly, the two biggest shifts for the end users are happening now. The biggest of them all is, of course, the shift from transacting on Ethereum L1 to transacting on smart contract rollups.
Optimistic Ethereum has been live on mainnet since January 2021 with Synthetix staking, and now has two other applications (apart from token transfers): Uniswap V3 and Kwenta (Synthetix dex). With Chainlink now deployed, we should expect plenty of projects to deploy on OE in the coming weeks. OE is open to all — try it out for yourself! Fees are 1/10th for a Uniswap V3 trade, and 1/50th for a Kwenta trade and paid in ETH itself. Arbitrum One has taken the opposite approach, by having projects deploy first, then open to users. It’s “ALMOST ready” (quoting A.J. Warner from Offchain Labs on Saturday) too, and I’d expect it to open to everyone in August. I suspect they are waiting for some final key pieces of infra like Etherscan to integrate.
You can check it out for yourself, of course, but I’ll describe the experience. For example, if you head to Kwenta.io, you click on “Switch to L2”. Metamask (or other compatible web3 wallet) will ask you if you want to switch, and with one click, you’re in Optimistic Ethereum! From there, it’s exactly like using Ethereum. Except, gas fees are 1/50th that of Ethereum, and transactions are confirmed instantly. No more waiting! (This is true of all rollups with sequencer models) I expect all frontends and rollups to offer a similar experience. You’ll need to bridge your tokens from Ethereum to OE. Currently, there’s only one option, using the Optimism Gateway (which has the downside of 7 day delay in withdrawals) but we’ll see multiple bridges deployed over time with instant withdrawals. Indeed, multiple projects are hard at work in making bridging between Ethereum and rollups (and between rollups) as seamless as possible. You’ll also have CEXs and wallets enable direct withdrawals and deposits from/to rollups — I believe OKEx has pledged to support Arbitrum One and Coinbase OE. Binance has recently released direct withdrawals/deposits to Polygon — it’ll work similarly. This way, we’ll see a new class of users that’ll never use Ethereum L1 — they’ll have direct fiat ramps to/from rollups, and all of their transactions will happen purely on rollups. I can’t emphasize this enough!
At the end of the year, or early next year, we’ll see zkSync 2.0 and StarkNet release, with Hermez following in Q2 2022. These programmable zkRollups enhance things further over optimistic rollups like Arbitrum One, Optimistic Ethereum and OMGX, with faster withdrawals without requiring bridges (crucial for NFTs) and better compression and privacy techniques.
It’s important to note that the transition to a rollup-centric industry is a gradual one, and this space will take a couple of years to mature. Currently, most solutions are MVPs with “training wheels” like single sequencers, transaction rate throttling and multi-sig L1 contracts. They are also brand-new tech and will take some time to prove themselves. The beauty of rollups is that each chain will have their own innovations, at a much rapid pace than any L1 ever could. For example, all different rollup chains will have different MEV mitigation techniques: MEVA (OE), FSS (A1), timelock encryption (zkSync 2.0, StarkNet), VDFs (StarkNet), fully private (Aztec). Bunch of meaningless acronyms, I know, but the point is they will be tackling hard problems head on, and innovative solutions will emerge. This means that over time, the experience of using one rollup can differ from others in notable ways, though I expect in the long term most rollups to converge on the best solutions.
There are challenges with cross-L2 and L1 <> L2 interoperability, but as I mentioned above, multiple projects like Biconomy, pNetwork, Celo, Celer, Connext, Hop, Chainbridge and Witnet are hard at work. We even have rollup developers like StarkWare and Loopring collaborate to release innovative solutions like dAMM which mitigates liquidity fragmentation by leaving liquidity on L1 accessible to multiple zk-L2s.
We’ve had application-specific rollups like Loopring, zkSync 1.x and DeversiFi for over a year now. It’s likely many of these will transition over to smart contract rollups, but we’ll also see application-specific rollups like Reddit’s Arbitrum chain or Sorare continue to exist standalone. Finally, users are free to continue transacting on Ethereum, but I suspect the average users will move to rollups and Ethereum L1 will become a playground for rollups to post their transaction data and proofs, whales, financial institutions, and governments over time.
Tl;dr: You’ll be transacting on different chains, much lower fees, instant confirmations.
This week: EIP-1559
The other big shift happening very soon is EIP-1559. There are two perspectives to this, Ethereum L1 and rollups:
On Ethereum, you’ll no longer have to select a gas price. Your wallet will simply give you a gas price and accepting it will mean confirmation in the next block for most transactions. This means in most cases you’ll only have to wait an average of ~7 seconds for confirmation! No more trying to figure out what gas price to bid, no more waiting around. Optionally, advanced users will get to select priority fees.
There are some edge cases where you’ll need to wait longer, like a sudden spike due to said degens aping into some new NFT. Speaking of spikes, these will be smoothed over. For starters, the network’s throughput will temporarily double. For example, the Stoner Cats sale took over a majority of the network and took 36 minutes to complete. Post EIP-1559, this will be completed in approximately half the time, i.e. 18 minutes. (Sidenote to NFT creators: please use L2s like zkSync 1.x and Immutable X for mass drops.)
Will EIP-1559 lower gas prices? It’s hard to say, but the most likely answer is — mildly. The good news is we’ll have a much better idea by this time next week! On a general day-to-day, it’ll prevent people from overbidding, so we should expect a slight drop in gas prices. When there are sudden, short-term spikes (couple of minutes), the gas prices will be smoothed over significantly. However, for sustained spikes like the Stoner Cats drop lasting over 10 minutes, we’ll still expect high gas fees as degens overbid on priority fees and the base fee ramps up. Fortunately, it’ll be over in half the time, so that’s still a net 2x benefit.
Moving over to rollups — interesting, both Arbitrum One and Optimistic Ethereum already have EIP-1559-like mechanisms, so you’re already getting these benefits already. With EIP-1559, it’ll be significantly easier for rollups to estimate L1 gas costs, so the rollup end user can expect lower gas fees. For example, OE currently has a 50% overhead floor to account for gas fee volatility. Post-1559, and with competition from other rollups, I fully expect this to drop significantly.
Tl;dr: On L1, you won’t need to guess gas prices and most transactions will confirm within 15 seconds. On L2, lower gas fees.
Late 2021: Social recovery smart contract wallets on L2
This one’s a bit of a dark horse, but I expect the smart wallets to proliferate as rollups gain prominence. Social recovery smart contract wallets (quite a mouthful — I’ll just call them smart wallets) have the potential to abstract out a lot of the complexity from the end user.
Imagine wallets where you deposit fiat to a wallet and get started. It could have a “Simple mode” with DEX and DeFi aggregations in-built. Want to buy some buy an ERC20 token? Just one click away. Want to earn interest on your USD? The wallet simply deposits it to Yearn or its own aggegator/pool. All of the heavy lifting — convert USD to USDC, convert USD to ETH to pay for gas, approve transactions, select the right protocols etc. — is done behind the scenes. With the move to rollups and L2s, all of this can be cheap enough to be viable. Of course, this wallet can also be used to log in to other protocols not built directly into it. As Login with Ethereum matures, it can be used as a universal wallet for all things Web3. Argent is a great example for a smart wallet — they are integrating zkSync later this year.
As Vitalik pointed out at the EthCC talk, such wallets are not only much better software and hardware wallets for the average user, they also have significant advantages over centralized solutions. Who would you rather trust? 5 out of 10 family and close friends? Or Facebook or Wells Fargo? The latter will surely take much longer to process your recovery requests, too.
As an aside, EIP-3074 brings some of these benefits to Ethereum L1 users with EOA wallets, but with the move to rollups and smart wallets, I’m more excited by smart wallets on L2.
My worry is that a centralized player like Venmo or Square will figure this out first, minus some of the decentralization.
Tl;dr: No need to mess about or worry with private keys; much better UX.
Early 2022: The Merge
This is the big one everyone’s talking about, when proof-of-work mining turns off and Ethereum’s execution layer (current eth1) is driven by the beacon chain.
Curiously, this is a complete non-event from the end user’s perspective. There’ll be a mild 10% increase in throughput given block times are a consistent 12 seconds instead of a variable average of 13.5 seconds with proof-of-work, but that’s about it.
If you made a transaction an hour before the Merge, the next an hour after the Merge, and weren’t aware that the Merge happened, you wouldn’t notice.
Of course, proof-of-stake will bring other less tangible benefits: your transactions will now consume 99+% less energy and there’ll be a different nature of security and decentralization (most will agree it’ll be superior, but the debate is still ongoing).
Finally, for users running and verifying their own nodes, there may be a change. I couldn’t find the answer to this one, so feel free to add — would the regular user need to run both execution and beacon nodes? I had always assumed this would be the case, but I’m probably wrong.
Tl;dr: Non-event with some niceties.
Mid 2022: Staking withdrawals
Don’t have much to add, this will be when users will finally be able to withdraw their 32 ETH stake and earned staking rewards. Interestingly, the transaction fees & MEV you earn post-Merge will be readily accessible on the execution chain.
There may be other minor improvements in this fork, such as moving to SSZ transaction encoding and changing how execution chain syncs, but these are unknowns at this point.
Tl;dr: Stakers can withdraw.
Late 2022 / Early 2023: Data shards
By now, I expect the rollup ecosystem to start maturing, and majority of activity will be happening on rollups. Data shards will increase throughput on rollups by 18 times. Simple transactions like token transfers will scale up to 100,000 TPS on zkRollups. Each data shard will have its own gas market, and with data decoupled from execution, we’ll see gas fees plummet as rollups start using data shards.
Data shards are an evolving solution. Sometime later in 2023, we’ll see an upgrade in security model to data availability sampling. Over the years, as the protocol matures, Moore’s and Nielsen’s laws kick in, more efficient erasure coding techniques emerge, and the network becomes more decentralized, we’ll see more shards added, with each shard offering greater data availability. I’ve previously estimated that by the year 2030 rollups all combined could be doing 15 million TPS thanks to data shards.
On the L1 side, nothing much changes — gas prices will continue to remain high, though developers may find ways to leverage the data shards for L1 smart contracts.
Tl;dr: On L2: throughput multiplies by over an order of magnitude, gas fees plummet. On L1: Mostly a non-event.
2023: Stateless clients
After pursuing stateless clients for years, we’ve finally seen significant breakthroughs this year, with a proto-EIP published. For the user running nodes, you’ll suddenly no longer require an SSD, and will only require a few MBs to run and verify Ethereum! However, your bandwidth requirements will go up by 2x-5x as you’ll be required to download witness data instead. It’ll still be quite mild, though.
With stateless clients and state expiry due (see below), we can be less conservative with gas limits, and let state grow while being more decentralized. As a result, we could see mild bump in throughput on the L1 chain by 3x or so. However, given most activity will now be on rollups, will it matter? Perhaps, if the execution chain starts becoming the bottleneck for state diffs and proofs, rather than data shards.
Tl;dr: Very easy to run an Ethereum node, no longer requires SSDs. For users that don’t run their own nodes, increase in L1 throughput by 2x-3x.
2024: State expiry
Roughly one year after stateless clients, state expiry will kick in. This is very much a complementary system, that’ll further make state size growth sustainable. For active users, this will be a non-event, but there may be some implications for less active users. If you want to interact with expired (2+ year old) state, you won’t be able to. It’s too early to say how this will play out, but the user will likely you’ll have to provide a witness from a block explorer to resurrect expired state. I’m sure user-friendly solutions will be developed.
Tl;dr: Too early to say, but inactive users may have to provide witness data to resurrect their relevant state.
Of course, there are other upgrades and improvements planned, but for the medium term (next couple of years) that should cover the major upgrades and how they’ll impact the average Ethereum user.