As I understand, it's not that these languages are forever on a downward trend with respect to case markings, but rather that these things are cyclical. In general, languages seem to cycle between analytical (few morphological cases, as the question is concerned) and synthetic (many cases) over long spans of time. This happens as children analyze speech in slightly different ways than the adults they are exposed to. A quick example of this happening in earlier English is the word napron. Adults referred to it as a napron, and now we know it as an apron.
You can find more about this by searching on terms linguistic cycle, grammaticalization and reanalysis.
It should be noted that languages with robust case systems are not "better" at expressing ideas than those with sparse case systems. In English, for example, we can express the same ideas for which other languages require case markings by adding other words (generally prepositions).