Aug 10, 2023

Recommendation engineX

Been digging alot into data networks and identity layers lately and still stunned with how decentralised social graphs combined together can create more richer identities that then enables whole world of use cases for other developers to leverage.



1/4


2/4 The current web2 model


Web2 platforms have always functioned like walled garden where their recommendation engines have been tuned on platform specific behaviour

Facebook: based on interests, demographics, interactions with posts and pages

Spotify: Listening history, collaborative filtering, genre based suggestions

The same goes for netflix But the main point, all of their recommendations are tuned on these parameters happening within their own walled gardens.


3/4

web3 scenario:

distributed ledgers enhances the diversity of data that is tied to an identity and thus creating more richer profiles.

Social graph, various attestations, NFTs and SBTs, Identity solutions like Galxe ID, Bright ID, Gitcoin Passport etc can all be aggregated down to one common identity which then devs can leverage to gather info from the outer world to power their own engine. Protocols like are now even working on bringing offchain data and attaching them to wallet profiles.

4/4

Use cases

  • Social networks: offer betters suggestions and a more curated feed by pulling aggregated interests from activities across the blockchain.

  • Growth tools: A major problem with questing networks is that they couldn’t identify power users because of no proper incentivization mechanisms in place. A unified identity layer can help these questing or ad networks offer better ROI to Brands/Advertisers.

  • Games: Recommendation engines building large data sets based on surveys are an outdated model. Social behaviours and interests change daily. Games can segment these aggregated data to drive new features. A challenge here is that alot of the data is still offchain (I think majority of the games are doing basic NFT shit while a smaller portion of them are bringing slow moving data like leaderboard and other stats onchain while a very small portion of them building some complex logic onchain.

  • What else?