7e97ba6bf3
Summary: We’re not convinced that using TFJS at this time is worth it, for two reasons. First, our matrix computations can be expressed using sparse matrices, which will improve the performance by orders of magnitude. Sparse matrices do not appear to be supported by TFJS (the layers API makes some use of them, but it is not clear that they have much support their, either). Second, having to deal with GPU memory and WebGL has already been problematic. When WebGL PageRank is running, the machine is mostly unusable, and other applications’ video output is negatively affected (!). This commit rewrites the internals of `basicPagerank.js` while retaining its end-to-end public interface. We also add a test file with a trivial test. The resulting code is not faster yet—in fact, it’s a fair amount slower. But this is because our use of `AddressMap`s puts JSON stringification on the critical path, which is obviously a bad idea. In a subsequent commit, we will rewrite the internals again to use typed arrays. Paired with @decentralion. Test Plan: The new unit test is not sufficient. Instead, run `yarn start` and re-run PageRank on SourceCred; note that the results are roughly unchanged. wchargin-branch: pagerank-without-tfjs
SourceCred
Vision
Open source software is amazing, and so are the creators and contributors who share it. How amazing? It's difficult to tell, since we don't have good tools for recognizing those people. Many amazing open-source contributors labor in the shadows, going unappreciated for the work they do.
As the open economy develops, we need to go beyond commit streaks and follower counts. We need transparent, accurate, and fair tools for recognizing and rewarding open collaboration. SourceCred aims to do that.
SourceCred will enable projects to create and track "cred", which is a quantitative measure of how much value different contributors added to a project. We'll do this by providing a basic data structure—a cred graph—into which projects can add all kinds of information about the contributions that compose it. For example, a software project might include information about GitHub pull requests, function declarations and implementations, design documents, community support, documentation, and so forth. We'll also provide an algorithm (PageRank) which will ingest all of this information and produce a "cred attribution", which assigns a cred value to each contribution, and thus to the people who authored the contributions.
Principles
SourceCred aims to be:
-
Transparent
If it's to be a legitimate and accepted way of tracking credit in projects, cred attribution can't be a black-box. SourceCred will provide tools that make it easy to dive into the cred attribution, and see exactly why contributions were valued the way they were.
-
Community-controlled
At the end of the day, the community of collaborators in a project will know best which contributions were important and deserve the most cred. No algorithm will do that perfectly on its own. To that end, we'll empower the community to modify the cred attribution, by adding human knowledge into the cred graph.
-
Forkable
Disputes about cred attribution are inevitable. Maybe a project you care about has a selfish maintainer who wants all the cred for themself :(. Not to worry—all of the cred data will be stored with the project, so you are empowered to solve cred disputes by forking the project.
Roadmap
SourceCred is currently in a very early stage. We are working full-time to develop a MVP, which will have the following basic features:
-
Create: The GitHub Plugin populates a project's GitHub data into a Contribution Graph. SourceCred uses this seed data to produce an initial, approximate cred attribution.
-
Read: The SourceCred Explorer enables users to examine the cred attribution, and all of the contributions in the graph. This reveals why the algorithm behaved the way that it did.
-
Update: The Artifact Plugin allows users to put their own knowledge into the system by adding new "Artifact Nodes" to the graph. An artifact node allows users to draw attention to contributions (or groups of contributions) that are particularly valuable. They can then merge this new information into the project repository, making it canonical.
Community
Please consider joining our community.