At the heart of this idea is that we can avoid the compromises imposed by the systems we use. But all system designs involve making compromises. The best you can do is design a new system that makes different compromises, but it's not clear that necessarily makes it better. The existing systems we have were often not the first of their kind, often there were many competing systems that could have won. These ones we have did win for reasons, many of them really solid ones and often because they made the right compromises.

Take JPEG for example, Six different companies helped develop it, there are six different patented ideas in the arithmetic encoding system and a slew of papers were influential in it's development over decades. You're really going to come up with better ways to do all of that over again from scratch?

OK maybe you come up with a better file system or kernel architecture, but why does that mean you need a new graphics compression format? You're also piling up all the risks together. Coupling together a huge swath of unrelated concerns arbitrarily means if some of them sink they'll bring all the other stuff down with them.