The problem with starting from scratch and doing it completely differently is that inevitably you'll end up researching only to get to the same point (for all intents and purposes). Some things just converge towards the same solution given the set of conditions.

So between spending all that (partially duplicated) effort to rethink everything and the fact that the end result would have little buy in due to the high threshold of effort involved with adopting and using it, this is rarely a tempting proposition for many. Either to develop, or to use.