Gamedev here. Sigh.
One, the measurement of novelty is bad: it's basically the measure of whether the game exhibits a novel mashup of mechanics according to the BoardGameGeek ontology of mechanics.
This is a terrible definition of novelty, as it ignores all of the other aspects of game design, both in gameplay (interesting systems, challenges, loops) and non-game play (fiction, setting, presentation, etc). Who measures novelty as "mashup of mechanics"?
Two, the measure of knowledge diversity is based on BGG reports of how many genres the designers have worked in. This is, again, not only questionable, but also leads to weird effects: imagine knocking Salman Rushdie, JRR Tolkien, or Umberto Eco for having low knowledge diversity because Amazon says they write books in only one or two genres!
But fine, even assuming that, the actual effects are not strong. Looking at the figures, the scatter plots are all over the place, and the trend lines fit poorly.
And finally, the main problem is: sure, if you compare hobbyists (who make what they want for fun) with professionals (who are making a product that needs to sell more than a small handful of copies), of course you're going to find that hobbyists are more free to experiment with wild permutations of mechanics. But in doing so, they conflate personal ability with situational constraints. What about professional designers if they were able to work without commercial constraints, like hobbyists?