SourceCred

Introducing the CredSperiment

The CredSperiment is our plan to dogfood SourceCred. Thanks to generous funding from Protocol Labs, we’ll be giving out several thousand dollars to SourceCred contributors based on their cred in the coming months.

As the name suggests, this is an experiment. We want to learn more about how SourceCred works in practice.

  • How well does cred hold up as an incentive when real money is involved?
  • What features do we need so that we can represent the different kinds of contributions happening?

I also think of this as a game. We’re creating new kinds of points and new roles for players. How can we make contributing to SourceCred really fun, engaging, and rewarding?

An Overview

The CredSperiment is built around three quantities:

  • Cred, which represents how much someone has contributed to the project
  • Grain, which represents how much someone has supported the project
  • Dollars, which can be used to buy coffee and pay rent

Financial backers, like Protocol Labs, will provide dollars to fund the experiment, and receive grain as a (non-monetary, non-exchangeable) record of their support.

Then, every week, we’ll pay out a certain amount to contributors based on their cred. The contributors can choose to get paid in dollars, or they can ‘re-contribute’ their earnings back to SourceCred, in which case they are acting as supporters as well as contributors, and they receive grain instead.

Closing the loop, supporters will be able to use grain to “boost” contributions, creating new cred. This means that supporters can use their grain to guide attention within the project. Supporters can even earn cred if they are wise with their boosts.

To go in-depth on the system dynamics, read the system overview.

Pre-requisites

We have some work to do before we’re ready to launch the CredSperiment. Among these pre-requisites:

  • Specify initial procedures for the CredSperiment (e.g. procedure for calculating cred each week, see notes in the system overview)
  • Finish merging the Discourse plugin
  • Build identity resolution between Discourse + GitHub plugins
  • Build initial tooling for tracking cred and grain
  • Circulate the initial scores and weights for community feedback before committing

Timeline

I initially wanted to launch this by the start of September, but looking at the unsatisfied pre-requisites, it’s not realistic. I’ll aim to launch by end of September instead, which should be quite achievable.

[edit note: I replaced instances of “mana” with “grain”. See here for context]

8 Likes

An update on the timeline and prioritization:

I’m thinking about the CredSperiment in distinct phases:

Phase 1: Payments based on Cred

  • Every week, we allocate rewards to contributors based on cred. Those contributors can either take the reward in dollars (paid via OpenCollective) or mana (which is tracked and stored)
  • The mana and payments may be tracked in an ad-hoc manner (e.g. a Google sheet)

Phase 2: CredSperiment Infrastructure

  • We’ll build the infrastructure needed to track cred, payments, and mana in a more robust way
  • This might involve some significant changes to SourceCred (building out server infrastructure and actually having a backend for the website)

Phase 3: Mana and Boosting

  • We’ll use the infrastructure from phase 3 to add the “boosting” mechanic, where participants can use mana to influence cred and “invest in” boosting contributions.
  • See "Cred Historians" and Curators

My goal is to ship Phase 1 by the end month, i.e. we pay based on cred in the first week of October.

3 Likes