The THEMIS RFC&C event, which started early February, has brought independent researchers and over 10 companies (e.g. Ava, AZTEC, Gluon, IMDEA Institute, O(1) Labs, Polkadot, SKALE, Solana, and StarkWare among others) together with the Brave research team to comment and propose directions for the future of the Brave Ads protocol. At this stage, the participating teams and Brave are actively engaging in technical discussions about the THEMIS protocol and implementation trade-offs. The teams have been actively proposing improvements to THEMIS, ranging from addressing design decisions (e.g. replacing signature schemes, which underlying blockchain to use, etc) — all the way to a complete revamp of the protocol design.
The current design of the THEMIS protocol offloads the reward computation to end-users. This design decision gives users control over their BAT balance and does not require users to trust that Brave calculates their rewards correctly. One of the expected challenges of running THEMIS protocol on the client is that proving the reward calculation based on black-box accumulators (BBAs) may require too much computational resources from constrained devices, such as mobile phones. To tackle this, participants have been sketching alternative strategies. Some teams are proposing to change the signature scheme of the BBAs to use non-pairing curves, since the computational complexity of generating the zero-knowledge proofs that require signature verification of pairing-friendly curves is quite high. Other teams have proposed proof schemes that may make it practical to use bilinear-friendly curves. In addition, other teams have proposed to leverage scalable blockchains to offload the computation from the users and to replace the BBAs with on-chain commitments and to leverage high throughput blockchains to implement the broadcast channel necessary for the Transparent Ad Attribution system.
Some teams are also proposing schemes that would allow the users to redeem their BAT automatically once their proofs have been verified on-chain. These schemes would enable the protocol to work without relying on Brave to release BAT reward payments to users.
We are looking forward to the remaining weeks of the RFC&C and to receiving the final submissions of the teams. The ongoing discussions have been highly productive so far, and we believe that by the end of the RFC&C event, we’ll have strong submissions that will greatly improve how THEMIS is deployed in the future.
During the remaining time, we will give higher importance to submissions that have clear performance measurements and we encourage all teams to devote more time to realistic performance evaluations. While we appreciate design trade-offs, direct measurements are often the most objective way to compare things side-by-side. We would be delighted to review these measurements together during the remaining time. We thank all the participating teams for their dedication and for the excellent input we’ve seen so far.