Metrics based on issues / commit activity are certainly higher fidelity.
As you indicate though, they require more effort to adjudicate. Are issues from core team members? Are commits meaningful? Is community activity meaningful? I wish GitHub would give allow us to parse things like this more easily.
My use of star count is generally a binary indicator. 1k+ is probably a legit project and below is probably still early. Beyond that, it's probably too noisy.
As you indicate though, they require more effort to adjudicate. Are issues from core team members? Are commits meaningful? Is community activity meaningful? I wish GitHub would give allow us to parse things like this more easily.
My use of star count is generally a binary indicator. 1k+ is probably a legit project and below is probably still early. Beyond that, it's probably too noisy.