Talking with FirstBatch, the Project Aiming to Personalise Web3 Experience
We had the opportunity to talk with a new entrant in the Arweave ecosystem: FirstBatch. From the moment you enter their website, you’ll be greeted by a prompt that manages to encapsulate the sheer scale they want to achieve in just two words: “Personalized everything”. Pair that with a pinch of ZK (zero-knowledge), which promises to achieve this personalisation without creating any privacy breach whatsoever, and one can say that you are in front of quite an explosive mix.
Without further ado, let’s find out what’s FirstBatch all about from themselves:
Q: So, from what we’ve gathered, you are trying to build the future of social networking and its inherent personalisation on Web3 in a decentralised, transparent, interoperable, and anonymised manner. Can you explain why having personalisation in Web3 is important, the Universal Interest Graph (UIG), and how your vision differs from what already exists in Web2?
A: While social networking is an important vertical for FirstBatch, we are building out personalisation for any web environment or platform in Web2 or Web3. Experiences in web-based social networks are a good place to start because it is here that we see the deepest UX personalisation via social graphs and algorithms and a depth of data available to understand user personas.
In the centralised interest graphs of Web2, users are able to experience platform-specific personalisation such as custom feeds, friends, content, product recommendation, and more – but at two costs. The first is that users must forfeit ownership of their data rights and privacy in order to receive personalisation from platforms. Second, since behavioural data is not self-sovereign, a user’s personalised experience is owned by the platform’s servers and not interoperable.
Personalisation and UX is an important factor in Web2 for onboarding and retention of users, which in turn attracts and furthers development. While the current Web3 UX is usable for people who are familiar with blockchain, wallets, keys, governance, tokens, etc, it is a completely new landscape for most web users who will be migrating from Web2. The seemingly endless FUD around Web3 in recent years makes it difficult for new users to want to engage with and ultimately trust Web3 platforms.
Personalisation is also shifting to become more community-focused, nuanced, and complex. Rather than just considering the preferences of individual users, personalisation mechanisms must now also consider the values and beliefs of communities to which those users belong. This shift is partly driven by the increasing importance of social media and other online platforms that allow people to connect easily with like-minded individuals and form communities based on shared interests.
Decentralised, privacy-preserving personalisation and recommendation are essential for new users to understand and trust the new interactions and environments made possible by innovations in Web3. Having the Universal Interest Graph, DANNY, and other core components of FirstBatch protocol available on Arweave ensures that developers have the tools they need to build community and scale any platform from NFT Marketplaces, Social Commerce Apps, Exchanges, or whatever comes next in any web environment.
Q: When you started building your protocol, what made you choose Arweave over other networks? Is there any other blockchain on which your protocol could be built?
A: We considered EVM- based structures, but it quickly became unfeasible to set up an EVM-based structure due to data storage costs and computation problems. Arweave is attractive for our needs as it is designed to be scalable, low-cost, and energy-efficient due to its consensus mechanism.
SmartWeave contracts offer us a few key advantages, such as Lazy Evaluation. SmartWeave contracts move the computation to the caller’s computer, unlike EVM. Since computation is done locally and queries to UIG do not change state, there are no gas costs for users. This provides unlimited scalability to UIG contracts. Since the Arweave network relies on a distributed network of users to store data, storage cost is drastically lower than EVM and generally cheaper than traditional centralised storage solutions.
Compared to other EVM-based solutions, the Arweave protocol is designed to be highly scalable, with the ability to handle a large volume of transactions and data storage without experiencing performance degradation. Since Arweave utilises Lazy Evaluation, the computation can be distributed among computers of your choice while using the shared state, ensuring the validity of results and faster response times like in Web2. The same property allows people to contribute to the scalability of UIG by running a node that queries the smart contract and connecting through an API.
Q: We read about your plans for the FirstBatchID (FID) and are curious to learn more about it. Will the FID be available cross-chain, and do you envision it to be the ZK ID to govern all other Web3 IDs in the future?
A: FirstBatch ID will be available to all chains and Web3/Web2 applications to have personalisation and align with interest-driven communities in any web platform environment or experience. However, for the sake of user privacy, we do not verify KYC or personal information such as address, citizenship, or demographic data. Our focus is directed toward aligning and connecting social communities and interests across Web2 and Web3 platforms.
Q: Considering that the FID will have a personalisation vector attached to it containing the interests and activity of the user, how do you plan on protecting the privacy of the user? Will it be just through ZK IDs, or will you also protect the personalisation data?
A: The first step for FID creation is to create a ZK ID by registering as a FirstBatch user. Next, AI associates interests to a ZK ID by conducting an analysis of a user’s on and off-chain social sources. This happens in a decentralised manner on the client side, which reduces vulnerability to attack. We do not store or handle any keys, passwords, or user IDs – only authorisation that a user has given FirstBatch permission to analyze their public data. The vectors associated with each FID represent a broad set of interests that can belong to many users and communities across the web. A user’s persona is the unique association of vectors that can only be constructed via FirstBatch Inference API. Even if persona vectors were to be exposed, they lack contextual meaning without FirstBatch and are not associated with any particular identity, thanks to ZK.
Q: How will you protect the personalisation data? And wouldn’t any project that would integrate with the FID need to interpret the personalisation data in order to provide the relevant content to the user?
A: (connected to answer above) Projects that integrate with FirstBatch will not need to conduct any interpretation of the data. We will analyze and contextualise the data that can be accessed by partner platforms for personalisation services via the API Key and $BATCH token.
Q: To my knowledge, anonymous advertising IDs are already used across the Web2 industry, so that my Personal Information (Name, email, phone number) is not linked to my preferences and activity. Of course, this is a system with inherent flaws which compromise privacy, but let’s imagine that the advertising ID really does grant anonymity. Today, advertisers are able to identify a person just based on the wealth of activity, preferences, and purchasing data they have access to, without ever getting their personal information. Wouldn’t users be confronted with the same risk when the Universal Interest Graph (UIG) grows to contain enough data about them?
A: Our approach with the UIG is to contextualise data, and we are not focused on the identification of individual users, nor do we allow third parties to make this association. The traits represented on UIG can be thought of as clusters with their own “gravity”.
As we mentioned above, the interest data available on UIG can be associated with many people, but each individual has their own unique association and pulls to the “gravity” of a given interest group.
Since persona vectors only have a context within the UIG, all third parties wishing to access the UIG will need to register for API access through FirstBatch. Next, any queries to the UIG for IDs that correspond to a given “gravity” of traits will not reveal any identifiable information, thanks to ZK, about a user, such as a name, email, phone, etc, which we do not analyze or vectorise in the first place. Interest data is protected in the same way by ZK – it is not possible for third parties to associate a given interest with a non-ZK ID or person, nor is it possible for them to track the source of interest data. FirstBatch also does not sell interest data to third parties for profit.
Our main focus is to provide personalisation based on the public traits of an individual relative to a community or idea, different from ad-models that focus on the association of an individual to their private ID data. Additionally, there is a lack of autonomy for users and accuracy for data collection with ad-based personalisation models. Unlike ad-based models, FID allows users to select which parts of their identity will be available or not for querying. It is also nearly impossible to generate really accurate data without users’ own consent on their public data, which is something that FID receives but mass advertising does not.
Q: Last but not least, we’re very excited to see what the future of personalisation will look like. When can we expect to see your technology in action?
A: We are planning to launch FID in early May alongside our ready-to-use decentralised infrastructure deployed on Arweave. In addition, we will launch a user-facing application where FID holders can access decentralised semantic search, personalised feeds and content, and ID management. FID holders can have personalisation on Lens, Mirror, Farcaster, many Arweave platforms, Immutable, and 10000s of web2 platforms (e-commerce/social commerce, socials, short-term rentals, news, video, gaming etc).