[philipodonnell]

The point has been made before that npm downloads are a poor measure of framework popularity because the architecture of the framework and the typical distribution channels can skew the number of npm pulls in a way unrelated to the number of projects utilizing it.

[mlsarecmg]

These numbers track actual, daily usage in production environments. You can't explain these numbers away, not if they're both competing in the same field. Both are distributed in the same way (npm, unpkg, various cdns), both websites show you how, both do not need build tools to function, both are better off with build tools. It would be pretty disingenuous to claim suddenly the only actual numbers we have are skewed.