Andrew Hong’s primer on web3 scoring and algorithms
Sep 12, 2022 . 3 min read . 357 views
title: 'Andrew Hong’s primer on web3 scoring and algorithms' publishedAt: '2022-09-12' summary: 'Web3 is here to stay and it is taking online reputation systems to the next level. With web3, a users digital identity is not tied to a particular website...' category: '#ReputationSystems #ScoringAlgorithms #Web3'
Web3 is here to stay and it is taking online reputation systems to the next level. With web3, a user’s digital identity is not tied to a particular website or company.
The reputation accrued on one platform could be used to unlock services/goods on another platform.
Imagine, getting invited to free bagels at a new breakfast place near you based on your Starbucks rewards level. Coz Forrest says Coffee and Bagels are like Peas and Carrots.
In his writeup, Andrew shares some of the main categories of web3 user scores today:
- Creators/Curators: Scoring someone’s ability to create or curate content
- Defi/Credit Management: Scoring someone’s ability to deploy and manage capital
- Contributors: Scoring someone’s contributions to a protocol/DAO and their skill set
- Rule-based: Scoring someone based on a defined set of actions taken
- Sybil (Identity): Scoring someone based on how likely they are to be human
To better understand these scores Andrew categorizes them along two axes:
-
Contextual vs. Generalized:
- How generalizable is a score across the communities it’s used in?
- e.g: Your ranking in PubG is more contextual to the game than your level of education, which is more generalized and has more use cases.
-
Elo Scores vs. Levels:
- is the score constantly adjusted and normalized, or is it a number that only goes up?
- e.g: Your socioeconomic status (combination of education, income, health) is an Elo score vs your work experience is an example of levels that only goes up.
Some examples of scores from web3 world:
- Contextual + Elo Score: jokeDAO, Mirror $WRITE airdrop
- Contextual + Levels: Metropolis, Roblox
- Generalized + Elo Score: Yup.io, lens.xyz
- Generalized + Levels: Rabbithole, Worldcoin
Right now, most of the scores are being used in tiered protocols access and leaderboard/ranked feeds, however, more powerful curation algorithms are around the corner.
Andrew believes we’ll likely see Sybil scores that combine with credit scores for enhanced access in Defi, or contributor scores that combine with rule-based scores for prioritized onboarding to DAOs.
There are teams already working on different parts of this problem to make it possible.
Web3 scoring algorithms tech stack
Data Providers: These provide APIs to access data. e.g ZORA provides data on NFT mints.
Data Mappers: These allow visualizing all on-chain data. e.g Dune Analytics
Score Aggregators: Unlike data mappers, data aggregators ingest a subset of data depending on their interested protocols.
Upon ingesting they perform feature engineering and more advanced data modelling on the raw score. e.g:
- Chainlink flux aggregator: for on-chain score deployment, you might want to take a weighted Sybil score from multiple sources.
- Gitcoin passport: for social apps, you’ll likely want easy access to multiple types of scores to enhance user profiles. Ceramic fits here too.
- Orange Protocol: for complex protocols (i.e. defi) you might need an ensemble model approach where you have complete control of inputs, models, and output variability.