Conversation

Want: • Scrape popular self-referential blogs (, LessWrong sequences, , maybe ?) • For each, construct dependency graph: what old posts do new posts reference? Simplify by hand Find guiding abstractions. May suggest entry points for new readers
9
84
Completely unsurprisingly if you've read a bunch of , but if you just count the number of self-referential links on then "A Big Little Idea Called Legibility" comes out on top by far! Haven't done dependency graph yet which may be more interesting. Raw counts:
Image
3
21
The code to generate the dataset is a total kludge of ruby scripts, but I actually just generated a better CSV of in-blog references which you can use: gist.github.com/backus/fc62897 (cc you may also enjoy just exploring that table).
2
2
Replying to and
I bet you'd be better off just starting from scratch. I'd love to learn from what you make. Things I wanted: • Less lag + no chaotic initial render • Detect subnetwork clusters, separate more • Nicer labeling. Avoid text overlap? Imagine you see much more due to better taste
2
1
Replying to and
Stretch goals/ideas I was curious about: • See what happens if we treat "rust age" (2007 - 2012) as separate author from (2012 - 2018) • Play w/, detect, visualize "chains" of citations? Ex: Maybe legibility node should be even bigger if posts that cite it are freq cited?
1
1
Replying to and
Nuance. A terminology citation is a different beast than a prereq citation. Not sure how to tell those apart structurally. If I were less sloppy I’d be linking “legibility” citations mostly to glossary except for ones where I actually extend Scott (eg GP3, ‘status illegibility’)