Tinkering versus engineering

When Massimo Banzi was introducing the philosophy behind Arduino at the recent beginners' workshop in London, he described his approach as 'tinkering'. Massimo referred to himself as a "renegade engineer", suggesting that there are differences between conventional engineering and tinkering. He came across this term for the first time when he was at the Exploratorium in San Francisco, which is putting on another tinkering exhibition in June 2008: "Tinkering, focused activity with the right materials in the right environment, can lead to great new inventions, but more importantly it builds self confidence and critical thinking skills in technology". That definition emphasises the advantages of tinkering but doesn't clearly distinguish it from engineering (which clearly has the same benefits that they describe).

Massimo gave a more psychological definition: "what happens when you try something you don't quite know how to do, guided by whim, imagination and curiosity". Conventional engineering is goal driven and so this is one way it differs from tinkering. Massimo talked about the value of building things without having any particular goals as this means there is no right or wrong, or concept of failure.

Massimo suggested several other ways in which tinkering differs from conventional engineering. Firstly, he believes people should be at the centre of any design process, rather than technology (which is typically what engineers do and perhaps explains why interfaces can seem so alien sometimes).

Secondly, Massimo's approach is to "describe by showing": quickly build prototypes in order to explore ideas. This approach is often used in conventional engineering too, but there tends to be more emphasis on formal methods rather than hands-on exploration.

Thirdly, he talked about how tinkering involves working out how existing systems work and then re-working and re-using them. As examples, he talked about scavenging the Olivetti junk yards in Ivrea for electronic parts, circuit bending and toy hacking.

I've been interested in the engineering/tinkering distinction for a while. In a 2002 paper with Paul Layzell we came up with the following distinctions that echo Massimo's:

ENGINEERING	TINKERING
Clear goal/plan	Often no goal/plan
Not necessarily dependent on previous designs	Uses whatever is to hand
Aims for best solution given constraints	Makes some kind of workable object
Insulates subsystems and minimises unforeseen side effects	Combines systems or transforms them for new uses

Arduino workshops demonstrate that the tinkering approach is pedagogically very valuable, but it can be a potent design methodology as well. Evolutionary electronics research, which could be viewed as automated tinkering, can produce circuits that would not have been produced by conventional design, for example, Adrian Thompson's tone discriminator and Paul Layzell's evolved radio.

Adrian's research page is a great resource for finding out more about evolutionary electronics. It also has two novel, possibly unique, examples of applying haiku rules to academic paper titles (both were presented at a conference in Japan and have 17 syllables organised in a 5, 7, 5 pattern):

An Evolved Circuit,
Intrinsic in Silicon,
Entwined with Physics.

Through the Labyrinth
Evolution Finds a Way:
A Silicon Ridge.

Darwin viewed natural evolution as a tinkering process: “throughout nature almost every part of each living being has probably served, in a slightly modified condition, for diverse purposes” (C. Darwin (1886) The Various Contrivances by which Orchids are Fertilised by Insects. D. Appleton.). The biologist Francois Jacob was even more explicit: “Evolution proceeds like a tinkerer who, during millions of years, has slowly modified his products, retouching, cutting, lengthening, using all opportunities to transform and create” (F. Jacob (1989) The Possible and the Actual. Penguin Books.).

We are evidence that evolution can generate complex, creative systems. But as a blind process, it can take an incredibly long time. Surely it's always better to use a conventional, top-down approach to design if you want to get anything done in a reasonable time? Adrian Thompson argues in his paper Notes on Design through Artificial Evolution: Algorithms and Opportunities that for some design challenges conventional engineering cannot help us and the only way we can proceed is to adopt a tinkering or evolutionary approach. Keep hacking away at those Arduino boards...