I tend to think of "machine learning" as the use of some automated algorithm to learn a ruleset as opposed to manually programming that ruleset. More often than not this algorithm requires some external dataset from which to learn the rules but in this case the algorithm is using another ruleset configured by Jansen to learn the rules. In that sense, since there was an automated algorithm that generated a "model" that abided by a set of externally provided rules, I would class this as machine learning.

That said, some people consider only scenarios where external data points were provided, rather than a set of rules, as machine learning. They may be right and I may be wrong - I'm open to debate on that.