DOES culture evolve according to laws just like those that govern biological development?
This has been hotly debated ever since biologist Richard Dawkins posited the idea of memes 鈥� units of information that replicate, mutate and evolve 鈥� in his 1976 book The Selfish Gene. So might it be possible to create the equivalent of a genealogy or genome for cultural 鈥渙rganisms鈥�?
A team lead by Mark Bedau, a philosophy researcher at Reed College in Portland, Oregon, has now investigated this for one particular cultural 鈥渓ife form鈥�: patents. 鈥淭echnology is a window on culture, and patents are a window on technology,鈥� Bedau told delegates at the Artificial Life XI conference in Winchester, UK, earlier this month.
Advertisement
What also makes patents attractive is that they offer a large and readily accessible data set. Patented inventions are uniquely numbered, and have to be novel and useful, so they are distinct 鈥渙rganisms鈥�. Reproduction corresponds to citation by another patent, and since every invention has to cite the patents it borrows from, it is easy to map out its genealogy. Every patent also lists claims for the invention it describes 鈥� a kind of genome that can be reduced to keywords representing distinctive traits.
Bedau鈥檚 team analysed 2.8 million US patents spanning 1976 to 2007, extracting around 300,000 keywords from their abstracts and charting how various traits mutated and reproduced. They found that patents, just like organisms, fall into families that share traits: 鈥渂rowser鈥� and 鈥渋nternet鈥� were common keywords for one family, for example, whereas 鈥渘ozzle鈥� and 鈥渘ucleic鈥� never appeared in the same families (see diagram).
The work has also uncovered some general characteristics of patent evolution. It revealed that patents are interconnected to a far greater extent than biological organisms, so the family tree is more like a 鈥済enealogical thicket鈥�, says Bedau. And the technique also revealed that the range of subjects covered by patents is increasing at a predictable rate.
聯Patents are interconnected to a far greater extent than biological organisms聰
Bedau鈥檚 team also examined the popularity of traits over time, which allowed them to identify hot technologies such as the internet and biotech booms of the 1990s. Bedau suggests this might help investors spot burgeoning or overlooked areas of research.
These results confirm that the geneaological analysis corresponds well with the known dynamics of technological progress. While it is still early days, Bedau notes that it also turned up some surprises. For example, several of the most reproductively successful patents of the past 30 years turned out to be associated with inkjet printing. 鈥淚t鈥檚 a really great technology for precisely positioning tiny bits of matter: skin, cells, DNA, metal and so on,鈥� he says.
Delegates at the conference pointed out that the patent system has quirks that may skew the findings. For example, patent examiners encourage applicants to make copious citations 鈥� possibly with little justification. Bedau acknowledges these limitations, but suggests that they will give patent genealogy its own unique character. Other bodies of work, such as academic papers, will have their own distinctive features. 鈥淭here will not be one form of cultural evolution, but many,鈥� he says.
