I’d like to see distributions of age of citations in papers aggregated by fields. Ie if a paper in field X is published in 2022 with 10 refs, how many are 1y-5y old, 6-10y old,…, 50-60y old etc. Take such data an aggregate by field. Then break that out by influence.
Conversation
I’d guess stem fields have a barbell bias: lots of recent papers and a few old ones. I’d guess HSS papers have an exp decay pattern.
I’d also guess influential papers in any field have anomalous distributions.
I think you could distinguish stale vs flourishing fields this way
3
7
I’m actually interested in general discourses, not academic work, but can’t think of a way to test this kind of hypothesis with them. Blog links and twitter quotes would be similar data. The bot picks up on old links for eg.
Track it via the arguments. Always works with books, nothing is ever really new after all.

