However, consumer products aside there are some cases where a complex, but not complicated, solution is better than the simple solution. For example, city planners have this eternal tendency to build flyovers whenever and wherever they can, but the flyovers themselves can't handle the increasing loads after a certain period of time. So, then what?
A more complex, but uncomplicated solution would be to make the traffic lights respond dynamically at a city wide scale with traffic. If you can figure out the volume of cars on a given stretch of road vs. other roads you can then use a routing algorithm to predict the best timing and path to guide the cars. It's more complex than the flyover, but it's cheaper and the leftover money can be then put into making better mass transportation in order to cut down car growth.
Further, you can figure out the volume of traffic by using accelerometers embedded in the road. Any vibration propagating through a solid medium has certain characteristics, which can be accurately predicted by studying solid acoustics. Hence, if you have an array of sensors (they are cheap) you can track the vibrations down to their respective sources. In this scenario accuracy is not an issue, a roundabout number should be good enough.
Powering the sensors isn't much of an issue either and we can put wire them to micro-controllers that crunch some of the data and sends it higher up the chain.
So, at the end we have an entire city that behaves and responds like an organism to the traffic flowing in its veins. It will be just beautiful.
The irony is that its inherent complexity ensures that people won't buy it. After all, who would want to trust solid math?