Last year Google socialised the idea of ARA, a flexible platform to create any kind of devices by using specific modules that can be assembled as plug and play. Like turning software "app" portability into hardware. This Fall ARA will be shipping to developers.
In UK, researchers at Bristol university have created CubiMorph, conceptually similar to Google's ARA: here again the idea is to create a framework that can support a seamless plug in of many different components to create fairly different devices.
You can look at the clips showing ARA (a more sophisticated video) and CubiMorph (you'll notice the engineering imprint).
The question is if this approach to allow a modular construction of devices can be successful.
99.99% of the history of the Human Faber has seen the creation of very specialised tools, each one fit to perform a very specific function. Till the invention of the computer the idea of a generic tool that could be customised to serve different purposes was not common at all.
Since the computer, however, we have seen a drive towards creating high volume standard "chassis" where embedded software provided the differentiation needed.
The Moore's law killed the modularity that was made possible by the "bus" architecture. It became more effective to wait for a new release of a PC than to upgrade it by added some plug in hardware or changing existing hardware. Actually, the miniaturisation (driven mostly by economic factors) led to place everything onto a single board, if not on a single chip, making replacement basically impossible. Nowadays only memory expansion is doable (and not always) on a PC.
Now that the Moore's law has come to an end (in economic terms) some are forecasting a revamping of modularisation since this would allow the upgrade of some parts as one would have to wait much longer to see a significant improvement of the whole.
If this is indeed to become true, the whole area of modularisation and flexible assembly could indeed pick up steam, and with it ideas like ARA and CubiMorph.
Personally I have some doubts. Volumes are still very important to decrease cost and ease of use in the mass market, such as the one that can be provided by a monolithic device are crucial for success.
The price of terminals has decreased significantly (it is being kept to certain levels for marketing reasons, not for Bill of Material constraints) and adding one more sensor, if needed, is not a big deal. Getting rid of 40% of a cellphone, because it is not needed to be used in a given application is not going to decrease the cost significantly. It is like saying that you would not need a digital camera on your smart phone and you can get a smartphone that is significantly cheaper just because it does not have the camera. The difference in price will be negligible, and actually it would cost more to produce two lines of product one with the camera and the other without it.
The idea behind ARA like approaches is that you can mass produce some bare bone product that can be sold at very very low price and then let customers to put all the bells and whistles they like paying for those a premium price (and possibly ending up paying more that what they would have payed for a full optional device). We have some examples of this strategy on the market, like in freemium products, apps that are provided for free and then make money by selling you adds on as you use them.
Reflex cameras have always been example of this approach. You get the camera body, with usually one lens, and as you get more and more addicted you buy more and more lenses ending up in spending much more than the cost of the camera.
Will see. I remain skeptical though.