This is never going to be true of hardware.
Having said that, the "shield" approach is clearly winning among the introductory and hobby market. You can buy loads of them for low cost: http://www.dx.com/p/arduno-37-in-1-sensor-module-kit-black-1...
Productisation is limited by regulatory effects, FCC/CE and so on.
Yes, but to make the analogy more realistic, it is an assembly language for a relatively simple and very slow and inefficient, power-consuming processor.
Nowadays, with discrete electronics components, one unfortunately cannot make fast and lean electronics like any of the big companies are making.
However they still have a very real application in the analog domain(along with a strong understanding of discrete ICs like Opamps and Transistors).
Xmos is offering low cost microcontroller family where you can program your digital peripherals in c. So are others.
Cypress is offering programmable analog chips. Very versatile.
There are extremely low cost fpgas.
To is offering an mcu with programmable wireless protocols, currently only by them but maybe there's hope.
For users interfaces, the phone offers a great users interface, if made accessible like simblee promises.
So basically there are lots of option available but what is missing is ease of use and maybe options to combine them together.
There have been attempts to create higher level abstractions (like Arduino "shields", etc) but the cost (time and money) to produce reusable modules is huge relative to the cost of the components, whereas CPUs are so fast the overhead of higher level languages and function calls is usually insignificant.
I'm not sure what the solution is. Maybe better software, and methods of manufacturing small runs of boards?