What's wrong with the usual open source way?

Nothing at all, in theory. It's just that, currently, no one has been able to produce an open source game on par with commercial titles using a purely collaborative approach. Games are one of the most complicated kinds of software to develop: They need focus and dedicated effort, and these are hard things to build out of a large, diverse, and distributed group of people.
 
Furthermore, we believe that asking members for a monetary contribution, rather than soliciting volunteer development efforts, is a more efficient way to marshal community support: the number of people with the talent, time, and inclination to develop a project like this for free is small, whereas almost everyone has the capacity to donate money. Making participation exclusive to those with programming or game development skill neglects the possible contribution of the vast majority of the community. I elaborated on this subject a bit more in this article, but basically I'm of the opinion that creating a next-generation city sim with an all-volunteer approach would require an amount of effort and luck out of proportion with the task.
 
Nevertheless, there are more optimistic people than me out there attempting to do that very thing; Off the top of my head, there are several open source city sim projects currently active: LinCity, OpenCity, WorldSim, and CityMania, to name a few.