The author seems to get carried away, emphasizing that microbiomes (measured by host health) do not form a totally ordered set. I think that's a caricature of what microbiome hackers are claiming. A totally ordered set would mean there's one "best" microbiome. But there might be one or only a few "best" microbiomes for a particular person with a particular diet.

In one section, the author criticizes microbiome hacking based on the fact that the experimenter does't have c. difficile. Does the author really believe that c. difficile, or other acute health problems, are the only possible results of an unhealthy microbiome?

It seems most plausible that dietary health is going to be a function health(microbiome, diet/environment, genetics). I doubt anyone thinks it will be a simple function, but along with acutely unhealthy microbiomes, there will be some obvious (given enough long-term data) problematic microbiomes across wide genetic and dietary variations, that don't quickly cause acute disease. And I don't think there will be as much variation as the author believes in healthy microbiomes once you start controlling for genetics and diets.