It fails to do that because it's actually not a trivial thing to do at all when you're not familiar with the field: it is very difficult for a non-IT professional to take a look at Kruskal's pseudocode ( http://en.wikipedia.org/wiki/Kruskal%27s_algorithm#Descripti... ) and to state if it solves an existing pain or it's just describing trivial and intuitive, common-knowledge steps.
A solution would be to have patent reviewers splited in categories based on their professional accreditation and limit their ability to accept or deny patents to their field of expertise.