The number of divisions and use of RPI for seeding are separate decisions, and do not go hand-in-hand, even though both may happen in the same year.
There are some flaws with using RPI for seeding, but there are also flaws with the present system. With many areas of the state going to super-districts, pods, or whatever you want to call it -- there are more teams that coaches have to track during the season for purposes of seeding. In the current model, say you're a Columbus D1 coach, you have to track like 45 or 50 teams for purposes of voting ... that certainly isn't easy for a coach to do. So instead of having the 50 coaches rate the teams, there's going to be a computer mechanism do it, and the you'll have the same bracket selection process behind it (and Columbus D1 is one of the simpler big-bracket climates out there in terms of the teams having relative connectivity to one another).
Even this past year, with the MaxPreps data being more accurate (since there is basically mandatory score reporting), I think you saw some coaches kind of look at the MaxPreps ratings (not the RPI though) as kind of a guide for voting within their pod/super-district. The MaxPreps rating uses margin of victory, which RPI does not.