Interestingly enough, the adoption of a new technology standard is very similar problem to that of the adoption of a new product. There is a learning/switching cost involved, and for someone to make the move the pain has to be absolutely unbearable, the new solution has to be 10x better, and the cost has to be as low as possible. We often end up in local minimums and paradoxically the only way to overcome those is to run competing searches (aka, multiple products/standards) or simply to restart the search. Then there is also the problem of existing solutions being already "good enough" (think python27, IE8, IPV4, dpkg, cars that run on oil, and electricity that comes from coal).