The test case building at the start of the article had just about as much go wrong with it as anyone should expect. Designers wanted to show off, sought the ma

The test case building at the start of the article had just about as much go wrong with it as anyone should expect. Designers wanted to show off, sought the maximum possible savings under the most optimistic cases, and just couldn't stick to fundamentals, nor a good study of the conditions at the site. Something designed with smarts, instead of 'to be smart' in the know it all sense might have met expectations. Though just like machine learning can produce solutions that tear through the test cases it's given, there are probably factors like acoustic isolation that get overlooked by the quest for maximums outside of a wholistic set of goals.