“Well, it was kind of an accident, because plastic is not what I meant to invent. I had just sold photograph paper to Eastman Kodak for 1 million dollars” – that’s what Leo Baekeland, the inventor of photographic paper and early ‘light’ plastic called Bakelite said once he had patented Bakelite. Imagine if he hadn’t made that accident in 1907, or better still imagine if he had actually discovered all the possible uses of plastic way back in 1907 without the need of an ‘accident’ and spared all those years through which plastic ‘evolved’? Well now that’s possible – or almost at least.
An emerging technique for new product development and technology optimization called ‘evolutionary design’ makes sure that while solving the problem of developing the ‘optimum’ product; a computer possibly runs through more than tens of millions of possible solutions.
To be entirely honest, evolutionary design is not a completely new concept. It has been used for quite a while in the development of aerodynamics for cars (supercars mainly), the modification and optimization of plane wings, and other aerodynamic related fields. The problem with evolutionary design was that it was restricted only to the larger firms who could afford the super computing power needed to generate more than 10 million ‘product gene’ modifications. The interesting development has been that primarily, the cost of super computers has fallen and this technology now comes within the reach of firms, which may not be mammoth in terms of turnover and market capitalization.
As the name suggests, evolutionary design uses some mechanisms inspired by biological evolution: reproduction, mutation, recombination, and selection. This is done so with the help of an evolutionary algorithm, which takes into account all the possible solutions to altering a problem. It does so through the mutation of a basic blue print.
Anyone who knows the basics of evolutionary biology will tell you that most mutations are worse than the original. Few are progressively better. With respect to product and technology development, an evolutionary algorithm ‘identifies’ these better mutations and thus makes it possible for existing products to be more efficient.
The Evolutionary Algorithm for Products works like this:
1) It takes the design parameters such as length, width, height, current, voltage, material etc. into account.
2) It treats each aspect of the product as a different component.
3) It then tries to put together all possible combinations of the different aspects of a product and come up with the best possible solution.
Here are a few examples on product innovations that have come forth through evolutionary design
‘At the University of Sydney, in Australia, Steve Manos used an evolutionary algorithm to come up with novel patterns in a type of optical fibre that has air holes shot through its length. Normally, these holes are arranged in a hexagonal pattern, but the algorithm generated a bizarre flower-like pattern of holes that no human would have thought of trying. It doubled the fibre's bandwidth.’ (generated from The Economist)
There are also other uses, which have been developed with respect to cochlear implants. The cochlear implant is often referred to as a bionic ear. Unlike hearing aids, the cochlear implant does not amplify sound, but works by directly stimulating any functioning auditory nerves inside the cochlea with electrical impulses. External components of the cochlear implant include a microphone, speech processor and transmitter, which also provides an individual to adjust the sound for quality and amplification. So in essence, about 20 or more combinations of electrodes need to be precisely adjusted for EVERY individual in order to adequately stimulate the auditory nerve! Evolutionary algorithms are now allowing doctors to solve the problems of patients within weeks (a task that previously could often take years).
Evolutionary design could affect ‘super specialized’ businesses as well. A research team at Stanford University prepared a Wi – Fi antenna for a company that did not want to pay any patent fees to Cisco. The team basically fed in all the possible product dimensions and specifications into an algorithm and developed a product that worked around Cisco’s patent and was actually a better product!
So in a sense this is the death of new product development right? Well not really – as new individual needs evolve and people look for new benefits from technology, new products will automatically evolve. Also, evolutionary design develops the most optimum design for products with a given set of ‘raw materials’ so to speak. So in a sense it only develops the ‘best possible solution’ from what is fed in to it. There may be inputs for development that product developers and researchers may still not have discovered the use of. One must keep in mind that an algorithm is a list of well-defined instructions.
Also what Evolutionary design does is that it makes the product perform better. Now with respect to commercial and business products, this means that there will be fewer product defaults and a benefit in overall productivity hopefully. What evolutionary design still cannot do is this – it cannot create a better ‘use experience’, which may arise out of software that is culturally ingrained (eg. Linux is a better programming platform and technically the ‘superior software’ but Windows still sells more as it is psychologically and culturally accessible).
The aesthetic design element will thus play an even more important element in the purchase decision of consumers while deciding which product to buy because most of them will be ‘superior performers’.
It will all boil down to who understands the consumer better, creates a better brand and use experience keeping in mind culture, values, traditions, personal goals etc. (or will we have an optimizer for that as well?)