Github recently added a new "Used by" metric which attempts to find projects that use the given repo. I really find this metric useful (along with actual download stats of course). But it does raise some interesting questions...

Stars are one metric that I've used in the past to help decide whether a project is dependible, popular, and "should I invest time using it" but it seems as if this is a completely useless number. Perhaps the used by metric is more helpful?

Let's see...

Google's flatbuffers project on github at the time of this writing has 12k stars, almost 2k forks, and used by 253? What? I find this interesting... why is it so low? I'd expact a project with 12k stars to be used by MORE by than the star count as many people don't bother to star projects.

Let's compare it to another project in the same category, ie msgpack.

Looking ONLY at msgpack-python's on github it has 8.5k used by, 1k stars and 162 forks... how is it so different?

If we compare actual download stats from pypi it's indeed true that many more people/projects are indeed downloading msgpack as opposed to flatbuffers (for python anyways).

- msgpack downloads per week

- flatbuffers downloads per week

Most likely the the used-by count is just messed up for flatbuffers but in any event, the used by metric seems to indicate usage much better than the useless "like button". I wish Github would just remove the star count altogether as it's more deceptive than it is helpful.