Main rule of thumb to avoid being fooled, I think, is to look for existing prototypes. No prototype - it's all just empty promises. Then you can more easily evaluate potential of the project.