Ways to get estimates of how popular a module is:

Popularity estimates

• use StackOverflow's search API (proposed enhancement to MetaCPAN)

Other metrics that are somewhat related

• development activity

My own proposals

• ancestor count
  • the metric itself:
    • It's sort of like a citation impact factor calculation — it measures how respected the module is within the community.
    • But, instead of citations, it analyzes dependents — other modules which use this module.
    • It's multi-generational — if another module is a small wrapper around this module, and that module gets used a lot, then that improves this module's metric.
    • Granularity would probably be at the CPAN-distribution level, but could possibly be down to the CPAN-module level.
  • various ways to use this metric:
    • Bacause BackPAN exists, it's possible to calculate how this metric varies over time.
      • This is FAR more useful than just being able to make a pretty line graph. Rather, it can be used in a Google-Zeitgeist-like way (or Digg-front-page way) to show "rising stars" — modules which are rising in popularity most quickly. For instance, it would have identified the rapid uptake of things like Moose and Plack.
    • An individual author could be given a 'author ancestor count' which is the sum of the ancestor count of all of their distributions.
    • Related-to:
      • It's probably possible to use fungibility analysis (determine when one module stops being used, and another module starts being used instead, in the same version), to determine related-to.
      • There's probably a lot of other ways to guess related-to metrics, but I don't have really concrete ideas right this second.
    • It might even be possible to use this in a GoogleRank-like way, so that when a highly-respected author uses another module, it increases that module's rank more than if an unknown author uses it.
  • tweaks
    • It might be a good idea to add weights to generation levels, so that the first-generation dependents impart a higher score than second-generation dependents, and so on.
  • TODO
    • create this, and release it as a "CPAN Zeitgeist" webpage somewhere (note: I did a trademark search at the USPTO, and it looks like Google has not trademarked the word 'Zeitgeist' for themselves