A long time ago (a couple of hundreds of years), people could make a refrigerator by applying a certain procedure with pieces of ice, harvested in winter, and other stuff. The ice hadn´t molten at the end of the summer. The knowledge the people had in those old days was empirical, but nonetheless, it was knowledge (only the how of the process wasn't known, but that doesn´t mean it wasn't real knowledge).
Attempts to reproduce this procedure in our times all failed. The ice was molten after a few weeks. Why can those ice houses not be reconstructed in our time (the article in Mother Earth News mentioned below just repeats the empirical procedure used in the old days, but attempts to make them "out of the blue" all failed))? Don't we know in these modern days what processes take place exactly to keep a piece of ice frozen for the whole summer without looking at how it was done in the old days, apart from the fact that it's not necessary anymore because of refrigerators?
In other words, why can't we use non-empirical (as opposed to the empirical science long ago) science to construct (behind the design table) these old fashioned ice-houses? Is the process too complex?