The point _may_ be that oranges can be evaluated based on a few simple variables (eg., richness of flavor, juiciness, etc.), where there's no common set of variables that can be used to evaluate all the different types of books (ie, what makes a good history book differs to what makes a good philosophy book, etc.) and if they can't be compared, a 'best' can't be established.

That said, as the author alluded to, 'best' in this context is vague and can mean what we want it to, eg., person a may consider best to mean the book that has had the widest influence, where person b may consider it to mean the book that they personally found the most insightful, etc.

I feel like it would be hard to choose the best of a commodity, so this is even more perplexing to me.

Easier because directly comparable, harder because one can't tell the difference?

Right. That’s why it’s helpful to get recommendations.