Framemogging and Demoing

Hey fellas, long time no see.

Frame transactions

The drama I've been writing about seems to be resolved with a pretty expected outcome. Frames[1] are SFI'd. ethrex runs a devnet with frame transactions already. And you can even try some of the early or current-state-of-the-art things at pq-sap.gwei.domains[2]. (Don't forget to switch the chain and request the funding from their faucet[3].)

Kurtosis

The testnet is powered by the Nethermind frames implementation as well as ethrex. I think it was the first time I interacted with a devnet like that, and I had only been reading about Kurtosis setups in the context of EIP-8304 work.

DFI'd 8304

This one is quite sad, and regardless of V's shill of new state types on Ethereum (8304[4] is such a new state type), most client teams seem to agree that this EIP would not be included in Hegota. However, a lot of EL SSZfication would probably happen in Hegota+1, so for me it doesn't really make sense to consider such a data storage type.

Native UTXOs

Upcoming research on native UTXOs[5] together with 8304 is relevant for us. Instead of Alice using Bob's 0x123...321 on the transaction envelope, Alice would actually derive the key from the published meta-address key. That allows Bob to not use the main pubkey 0x123...321, but derive an arbitrary one for each spend. The ciphertext can also include additional information about the UTXO needed for spending that only Bob would be able to decode.

Back to basics

Thanks to myself in the past, I've explicitly stated that the scope of this research would be primarily about the key-exchange part, and that spending would remain a moving part on the protocol level.

Gas benchmarks

If you'd find a few minutes to try out the pq-sap demo[2] and multiply a few values together to get the actual spending costs of the demo, you'd quickly realize that none of the suggested methods are really within the realm of practicality — showing a whopping ~$200 to verify a signature.

EIP-8288

As always, V's on top of all the good stuff and has already proposed mempool signature and proof aggregation[6], which at the moment seems like the only considerable option. And due to the design of it, spending can be extremely cheap and even usable in the real world, yay! By now the biggest improvements are: SPHINCS⁻ — ~(200k | $20) — and the Fireblocks ML-DSA verifier[7] — ~(4m + 1.1m | $150).

Ephemeral accounts

Arguments for ephemeral accounts and implementation approaches[8] is a relatively fresh piece of research that went off my radar, but it carries quite an interesting tool for us. Ideally stealth addresses should not be re-usable, and they may be discarded after a spend — but there was no mechanism to enforce that. Sounds really interesting; need to dive deeper.

ERC draft

I started cleaning up the AI agents' work and emptied the ERC draft to start producing it during the next week.

Current documentation is available at skanislav.github.io/pq-sap[9].

Have a nice week!