What is the Builder Score?

Builder Score is a numerical value that represents a user's reputation as an onchain builder.

The Builder Score aims to distinguish builders from other types of crypto users, by valuing both the quantity and quality of their contributions. It reflects both "potential" (future growth) and "achievements" (proven track record).

Users can leverage their Builder Score within Talent Protocol's growing partner ecosystem, while developers can easily add the Builder Score to their apps.

Apps like Basenames or Etherscan already integrate with the Builder Score to bring more context to their users and to curate the best builders in crypto.

The "Builder Score" Scoring Framework

All Talent Protocol scoring systems are composed of a selection of Data Points, and assigning to each data point a:

1. value conversion

2. signal strength

3. max score

4. multiplier

1. Value Conversion

For some Data Points it's helpful to normalized their numerical value before applying the scoring function, but this is an optional step.

Examples:

• Convert twitter_account_age from a timestamp to the number of years the account has.

• Normalize base_out_transactions with a sqrt function to reduce the impact of large values, making the scoring system more balanced and fair, and ensuring big contributors don’t completely overpower smaller ones. This means that your first contributions matter more than later ones.

• Invert farcaster_social_capital_rank value, so that the #1 ranked user has a higher value (and more points) than the last ranked user.

2. Signal Strength

Defines how strongly a given data point proves that the user is a real builder, and can have 3 possible values: weak, medium, strong.

Examples:

• Winning a ETHGlobal hackathon is a strong signal that someone is a real builder.

• Simply having a GitHub account is a weak signal that someone is a real builder.

3. Max Score

Defines the maximum amount of points each Data Point can contribute to the overall Builder Score. We use the signal strength to help define the max score of each Data Point:

• strong (up to 40 points)

• medium (up to 20 points)

• weak ( to 8 points)

4. Multiplier

In the Builder Score we use a multiplier to automatically convert the numerical value of a Data Point into a score, measured in "points".

The multiplier is needed in order to standardize very different scales of Data Point values.

Summary

The Builder Score represents 3 core decisions:

1. What Data Points are relevant to assess the reputation of a builder.

2. How strongly does each Data Point signal that the user is a real builder.

3. What should be the multiplier to convert each Data Point value into Builder Score points.

The final Builder Score is the sum of all the points.

Builder Score doesn’t have a maximum cap, so it’s likely to increase over time, as we add more integrations and possible credentials to the protocol.

A Scoring System assigns numerical scores (measured in "points") to a group of Data Points.

Talent Protocol intentionally separates Data Points (objective data) from Scoring Systems (subjective data). Both are autonomous and composable protocol primitives.

Scoring Framework

A Scoring System is composed of:

1. a selection of Data Points to be considered for this particular score

2. a value conversion (optional) to normalize Data Point values (ex.timestamp_to_year or sqrt)

3. a signal strength (strong, medium, or weak) attributed to each Data Point

4. a maximum score (in "points") that each Data Point can contribute to the overall score

5. a scoring function or condition that converts the current value of a Data Point into a numerical score (measured in "points")

Scoring systems are absolute, meaning they don’t include any form of normalization or percentile-based approach. The final score should be a simple sum of all Data Point individual scores.

This framework intends to keep every scoring system simple, transparent and scalable.

Calculating a Score

To calculate a score, Talent Protocol will follow the steps below:

1. Fetch new data for each account and store them in the respective Data Point.

2. If the Data Point value changed, we create an Event (coming soon).

3. We then apply the scoring framework to all Data Points with a value change.

   1. If there are multiple sources for the same data point (example: a user with multiples wallet addresses), we pick the one with the highest value.