Being a child born just a few years before the new millennium, my lifetime has coincided with what seems to be the fastest period of growth in technology our world has witnessed so far. (I claim no responsibility for this, yet.) Since 1996, we have grown the capacity of a single hard-drive nearly 4000-fold from 16 gigabytes to 60 terabytes. Electric cars have become a legitimate and viable transport option. Stage 1 rockets can now land on their own and be reused. The race to become an interplanetary species is in full swing. It seems innovation is on an exponentially growing curve. However, Vittorio Loreto disagrees.
On the evening of October 24th, Loreto presented on the topic of “Novelties, Innovation, and the Adjacent Possible”. In his presentation, he referenced studies which his team had performed which suggested that innovation and the production of novelties are, in fact, on a downwards slope. The studies surveyed a handful of internet-based communities such as Last.FM, GitHub, Wikipedia, and Project Gutenberg. What Loreto’s team saw was that over time, the rate of innovation (contribution of novelties) steadily declines. Loreto argues that innovation, much like any resource, can be depleted. As we mine innovation, it becomes harder to create novelties organically.
One might argue that Loreto’s observations of novelty rates on sites like GitHub or Wikipedia may be somewhat biased or flawed. The rate of innovation in a medium, whether it be a traditional written language or a programming language/paradigm, must be subject to the capacity of the medium. While such a number would still be astronomical, only so many sensical phrases can be written in the English language. Similarly, only so many valid lines of code can be written within a language or paradigm. Once we begin approaching the bounds of a medium, standalone novelties do become more difficult to produce. However, novelty should not be mistaken for the only form of innovation. Context adds a whole new dimension to the bounds of the medium, one I suspect would be too complex to observe and model on a chart (e.g. frequency of A occurring immediately between B and C, followed by X VS. just the frequency of A).
Moore’s law presents an excellent example of continual innovation without much novelty as of recent. “Moore’s law is the observation that the number of transistors in a dense integrated circuit doubles approximately every two years” (Wikipedia) The rapid evolution of computer processors was to due to novel technological innovations in its early stages. However, recent growth has been dominated by innovative reconfigurations of existing concepts. Moore’s law applies to a number of similar observations including hard disk drive areal density, fiber-optic capacity, pixels-per-dollar, and the quality adjusted price of IT equipment.
Innovation comes in many forms. Some innovation is found true discovery and happenstance. Other innovation is laboriously constructed by drawing on and building up from previous achievements, no less impressive. Vertical innovation should not be discredited in favor of its horizontal cousin.