AMA with Ben Livshits and Gonçalo Pestana

Welcome to the latest post in our series of BAT Community-run AMAs. 

The ongoing AMA series is an opportunity for our users and fans to get to know the Brave team, and, as of recently, guests from Brave’s partners.  

Ben Livshits (Chief Scientist) and Gonçalo Pestana (Sr. Research Engineer) from Brave participated in the latest Ask Me Anything on r/BATProject. The pair fielded both pre-submitted and live questions from Redditors about the THEMIS protocol, the ongoing THEMIS RFC&C event, and the team’s research and plans for the progressive decentralization of the Brave Ads and Rewards infrastructure. 

Ben Livshits is Chief Scientist for Brave Software. He is also a Reader at Imperial College London and an affiliate professor at the University of Washington. Ben is the author of over 100 academic papers, dozens of patents, and multiple tech transfer awards for bringing research into practice. 

Gonçalo Pestana is a Senior Research Engineer on the Brave Research team. His current work is focused around the THEMIS protocol and R&D for privacy-preserving antifraud and reputation systems for BAT.

Highlights from the event can be found below, with a link to the full AMA at the bottom of this post.

ReallyNewHere: To keep up with the THEMIS solution one has to be really technically competent in that field. Simple question, will users be able to use BAT for something like staking, similar to PoS?

Ben: We have been thinking about this but haven't yet arrived at a decision regarding staking; we appreciate that people want to use their BAT in interesting ways, however, and want to enable other ways in which BAT can be utilized.

someonetwoforgett: Can you explain THEMIS like you are explaining it to a 5 year old?

Gonçalo: I assume a 5 year old already knows about (and uses) browsers and ads, so let me take a stab at this one.

Imagine your browser will give you candy based on ads you see. Now, in order to get your candy, you need to tell Brave how many you should receive. However, you want to keep the ads you saw as your secret, so you won't reveal them. THEMIS will make sure that, even though you keep your secrets, you'll be able to receive your candy. Ah! And you don't try to lie and ask for more candy than you deserve—THEMIS will magically know that you are lying and you won't get any candy :O

(This one was harder than I thought. :))

Kingflares: The THEMIS RFC&C deadline was extended a few weeks on discord to allow for more teams. How does this affect the overall timeline of THEMIS?

What is the new deadline or next update for THEMIS and when can we expect it to be done? What is the tram working on after THEMIS?

Ben: We wanted to widen the THEMIS participation to more teams, because they asked us to do so—it seemed appropriate to be more open—this means that we'll wait a bit longer, until the month's end, to receive their submissions. You can read the latest update here.

Gonçalo: Thanks for the questions! It is true that the deadline for the RFC&C submission has been extended a couple of weeks—the new submission date is the 31st of March. This should not affect our plans considerably but will allow teams to have stronger submissions. Once we get all the submissions from the teams, we will review them and plan the next steps regarding which technologies to use and how to plan the development and rollout of the protocol.

Taney: Could you discuss the evolution of THEMIS' consensus mechanism? THEMIS first depended on Proof of Authority, but I understand that is no longer the case. What consensus mechanisms right now seem the most promising?

For the consensus mechanisms under consideration, how might verifiers be rewarded? Would anyone with the appropriate knowledge and time be able to set up a node and receive rewards for verifying?

Gonçalo: Correct, the THEMISv1 protocol relied on a PoA sidechain to calculate the rewards in a privacy-preserving way and to verify and orchestrate the reward payments of Brave Rewards. Due to the drawbacks of sidechains, we designed the THEMISv2 protocol to offload the reward updates and calculation from the sidechain to the users and Brave. In THEMISv2 the users, Brave and advertisers use zero knowledge proofs to verify that the protocol is running correctly. The proofs may be verified on-chain to provide public visibility of the correctness of the protocol. However, THEMISv2 is agnostic to the blockchain used and underlying consensus mechanism. We’ve been working in the RFC&C with L1 projects to find the best option considering decentralization, costs and scalability.

In THEMISv2 there may be the need to batch the proofs before verifying them on a blockchain in order to reduce the on-chain costs. This could open the possibility to create an incentivize network of “batchers”, who would provide computation resources necessary to batch the proofs and be rewarded for that work. However, this is a long-term scheme (if it ever happens) and initially Brave will be the entity batching THEMISv2 reward proofs. Thanks for the questions!

McJvck: Are you looking for young talents to do research in this area? Are you open for Ph.D. positions or plain hiring someone with a masters degree in Cyber Security?

Ben: We are open to people with a variety of research-oriented backgrounds, with degrees in computer science.

Gonçalo: You can check our careers page here /careers/. We're always looking for great people!

JustinMurray03: What are the primary metrics you're looking at when approaching a move to a L2 or other blockchain? TPS? Do we have a public comparison of metrics we can access? (of the teams that have made their interest public)

What are the goals for Themis scale-wise, such as how many transactions per second do you expect will be needed from the infrastructure EOY 2021 vs EOY 2022?

What would a move off of ethereum to a different L1 blockchain such as Solana look like?

Gonçalo: Hi, thanks for the thoughtful questions! We have released a tech report about THEMIS for the RFC&C that may help answer your questions. We expect THEMIS to support up to 5M concurrent users (with potential for more) and to verify 50M rewards per month, including on-chain verification. In terms of costs, we’re setting the bar at up to $0.1 per reward computation and verification, with on-chain verification and costs for batching proofs (if necessary). In addition, we want to make sure the load on the client is as low as possible—the performance and battery consumption of the protocol should be negligible. These numbers help to steer the protocol design and technology decisions we are currently making.

bat_account: What is the likelihood that BAT switches to a different L1 blockchain with THEMIS? And if that happens, do you foresee the current distribution/breakdown of BAT's ERC-20 being transferred over to that L1?

In short, is it possible that there will be "new BAT" and that "old BAT" holders will get diluted? I understand BAT is not a security.

Ben: That's a good question—should we decide to migrate we plan to ensure that that doesn't negatively influence current holders of the ERC-20 token; there are technical schemes that allow for that, although no good examples of such transitions that I am personally aware of.

Gonçalo: This is a great question, and as Ben answered in this thread, cross-chain token migration is a very hot topic nowadays and I believe that will be the case in the future.

McJvck: Will THEMIS be an on-chain or second-layer solution? Also, once THEMIS is out, will Brave Software be required to engage with the ad network or is it 3rd-party free?

Gonçalo: Thanks for the question! The current THEMIS protocol design relies on both L1 and L2: while the reward calculation is performed off-chain and verified using zkproofs, we also plan on verifying the batch of those proofs on a scalable L1 layer.

Currently we're assuming that THEMIS users are using the Brave Browser—as it is the case with Brave Ads nowadays.

MrSaturdayAMcoffee: Love the work you guys are doing. Given the number of teams participating in the RFC&C, do you see a future with many chain systems flourishing or a consolidation toward a few big players?

Besides performance, what other factors are considered in your evaluation of the proposals?

What other applications of THEMIS-like system would you like to see?

Ben: Thanks—that's a very interesting question for BAT as well as other established tokens. As more ecosystems develop, should they consider migrating the token? While we've seen interesting tokens on Polkadot and BSC, say, we haven't seen established ones migrating across chains. What that cross-chain space looks like a year from now time will tell—we're seeing a lot of interesting works in centralized and decentralized bridges, for instance.

Gonçalo: Regarding evaluation of RFC&C proposals, we're considering several different dimensions as outlined in the Section 4 of the RFC&C tech report. Performance is one of them, for sure, both from the user perspective (how much energy and resources will take for the user to compute rewards and prove their correctness?) and from the reward batching and off-chain verification. Besides performance, on-chain costs, decentralization, complexity and moving parts of the protocol are also important metrics to consider.

robbaflockaflame: What are your short term and long term goals in terms of research? What is there to discover?

Ben: That's a great question! At Brave Research we focus on a broad range of topics and blockchain is one of them... Have a look at /research/ to find out a lot more, including many of the papers we've written, etc.

Gonçalo: Some examples of research questions we keep working on:

- How to build a decentralized ad reward protocol where Brave, users and advertisers can verify that the protocol is running correctly? How to do it in a privacy-preserving manner? (THEMIS is an attempt at answering these questions, by the way.)

- How to distribute Brave assets (e.g., ads catalog, Brave today content, etc.) in a privacy preserving way, having scalability in mind and while reducing bandwidth overhead?

- How can we leverage DeFi and other Web3 technologies in the browser? How could BAT improve the Web3 experience of the Brave users?

- How to improve privacy-preserving ad targeting on the client?

- How to detect and block fraud given the privacy guarantees we provide to the users?

These are just a sample. 🙂

HenryfromCC: IMO Brave is a project that shows real thought leadership, all around.

Do you find it motivating to work on a project that has so much potential to improve people’s privacy and wellbeing on the internet, generally?

As a privacy advocate myself... I feel like it would be extremely motivating and fulfilling to work on a project like Brave/BAT, and I just wanted to thank you guys personally for everything you are doing. It’s great that the project has attracted such talented people!

Ben: Thanks for saying that—yes, we get a lot of positive feedback as people perceive us as standing up for their privacy! Thanks for your kind words.

Gonçalo: Absolutely! The cherry on the cake for me is that working on privacy projects is technically challenging and we are constantly thrown at problems that have hard constraints and tradeoffs. This inevitably forces us to work cool and green technology, which is very intellectually rewarding IMO!

Read the full AMA here.

See our recent AMA with Brian Bondy, Co-founder and CTO of the Brave privacy browser & Dietrich Ayala, IPFS Ecosystem Lead, here. 

Follow the BAT Community’s Updates here: https://www.reddit.com/r/BATProject/

Related articles

Ready for a better Internet?

Brave’s easy-to-use browser blocks ads by default, making the Web faster, safer, and less cluttered for people all over the world.