Thread by @lightclients

In the last month, users spent $23,830,482 on ERC-20 approvals. EIP-3074 would reduce that number by at least 30%, saving users millions while freeing up block space (thread).

Anyone whose nonce is greater than ten has likely experienced the approve+transfer flow that defi applications use to interact with ERC-20 tokens. The approve method is a relic of the original ERC-20 design.

@DuneAnalytics dashboard for above tweet: duneanalytics.com/lightclient/US…

There have been many attempts to alleviate the burden of token approvals, but none have succeeded. We believe EIP-3074 will resolve the issue because it addresses the two problems that have beleaguered other approaches.

Issue #1: there is a long tail of ERC-20 tokens that can't/won't upgrade to a new token standards. EIP-3074 allows msg.sender to be set artificially based on a signature, so it is perfectly compatible with old tokens.

Issue #2: previous attempts at native support were simply too complicated to integrate into the protocol. EIP-3074 is *very* simple to implement in clients (we've already done it in geth & openethereum).

How does EIP-3074 save users gas? Today, an ERC-20 approval typically costs around 43,000 gas and a transfer costs 37,000 gas. Of course, if you're sending a transfer you'd just do that without the approval, but let's pretend it's a placeholder for something more interesting.

All in, that's 80,000 gas for the operation. However, only 38,000 gas is doing relevant computation. Over half is the "intrinsic cost", equaling 21,000 per tx. If you could do approve+transfer in the same tx, you'd only have to pay the 21,000 once!

This is exactly what EIP-3074 allows users to do. Suppose there is a contract written specifically to facilitate approve+call interactions. It would take in an ERC-20 token address, the amount, the subsequent call, a signature, and it would cost around 7,000 gas to execute.

Adding it up:

21,000 intrinsic cost
7,000 EIP-3074 contract
1,200 EIP-3074 extra calldata
38,000 approve+transfer
---------------------------
67,200 gas

Compared to 80k above, we saved 16% overall & 37% on the approval operation (i.e. disregarding the intrinsic cost).

Scaling up, what if we approve+transfer ten different tokens? Normally that would be 800,000 gas, but with EIP-3074 it's:

21,000 intrinsic cost
30,000 EIP-3074 contract
1,200 EIP-3074 extra calldata
380,000 approves+transfers
---------------------------
432,200

46% better!

How can this be? Well, ten approve+transfers requires 20 txs. That's 20*21,000 = 420,000 gas just for the intrinsic cost. Because EIP-3074 *batches* operations together into a single tx, the 21,000 cost is only paid once.

Although these numbers are rough estimates, and there is a lot of nuance in EIP-3074, the potential gas savings for users is significant.

Only exploring the gas savings EIP-3074 provides doesn't do it justice though. It will drastically simply UX for users and allow complex interactions that were not feasible before.

https://twitter.com/lightclients/status/1371911245561917441?s=20

Want to help us (@_SamWilsn_ + @adietrichs + me) bring EIP-3074 to mainnet?

Todo list:
i) retweet thread head for awareness
ii) join community call 4/26 @ 16:00 utc
iii) review the EIP text & if you support it, signal your support publicly
iv) dm me!

Community call info:

https://twitter.com/poojaranjan19/status/1384943167292776450?s=20