LengRank

LengRank is Library of Leng's article-importance score. It uses the article-to-article links preserved during conversion to find writing that other writing points toward.

Algorithm

The converter turns same-archive article links into /articles/<id> URLs. LengRank scans converted Markdown for those links, builds a directed graph where each article is a node, and runs PageRank with the standard damping factor of 0.85. The score shown in search and on the home page is the final probability mass assigned to each article.

Articles with no outbound links still participate in the graph. Their rank is redistributed evenly during each iteration, matching normal PageRank handling for dangling nodes. Iteration stops once the total rank movement is below 1e-8, or after 200 rounds.

Edge Filters

LengRank ignores links that are unlikely to represent editorial citation: self-links, repeated targets from the same article, links to articles outside the generated manifest, and links between articles with the same normalized author ID.

Draftsim also gets a site-specific filter: Draftsim-to-Draftsim links do not count. That site aggressively autolinks common Magic terms to evergreen internal pages, so those links are better treated as site SEO structure than as article recommendations.

Where It Is Used

Popular article lists sort by LengRank, then by publication date and title for stable ties. Search blends full-text relevance with LengRank so exact query matches still matter, while heavily cited articles get a small boost.