I can talk from personal experience. I was one of the first people to start adding coordinates to  articles of Indian places, sometime around 2007/8 when wikiproject geographical coordinates started picking up steam. There were lot of villages in Tamil Nadu, where I blindly picked the location shown on google maps (as there really was no other source at that time). After an initial mapping spree, I gave up on the task because it was too tedious at that time. This is roughly the same time I discovered the openstreetmap project and realized my mistake with copying blindly from existing maps.

The coordinates I had added may have possibly been incorrect (I really dont know and there was no way for me to verify them). But by now the error has already gone back into google maps and the position stands validated because its the same coords on wikipedia.

Imagine the irony if map services with the correct position is now seen as being inaccurate because wikipedia states otherwise. Just like article citations, even coordinates must have some indication of the source. The current system is just faulty.