It really is terribly simple,
1) models with predictive capability are of utility.
2) Science is in a class of its own, a process without competitors, when it comes to its ability to form such models.
So where is the problem? These are observations that not only essentially everyone agrees with, but practically EVERYONE eagerly embraces, most notably in tasting the sweet fruits of science, such as modern civilization.
However religion, a model with no demonstrable predictive utility, remains prevalent within mankind.
The bizarre thing is that all these people who engage in religion must realize what a good model looks like as they use them everyday of their life. Convergent models with predictive capability are the good one... the ones they chose to use. Yet still, peculiarly, these folks still embrace religion which is evidently divergent with no demonstrable predictive utility.