Author William Gibson famously said, “The future is already here — it’s just not very evenly distributed.”
We are living in a world where self-driving cars, universal language translators, artificial intelligence, robots and drones are no longer the realm of science fiction. These technologies are in fact an everyday reality for an increasing number of people.
Much of the current “state of the art” in technology advancement has occurred, at least in part because of the relentless advancement in microprocessors. In 1965, Gordon Moore one of the co-founders of Intel made a now famous prediction. Moore observed that the number of transistors on an Integrated Circuit had been doubling every year since their introduction. At that time there had been nearly 60 transistors on an integrated circuit, and Moore predicted that this doubling would continue for at least the coming decade resulting in chips with as many as 60,000 transistors on a single chip. This prediction has become known as Moore’s Law.
Roughly speaking, Moore’s Law has slowed a little since then, but we have still seen a doubling of computing power every two years since 1975. It’s not possible to visualise something that doubles regularly on a straight line, to plot this kind of progress you need to plot an exponential line (which looks like a hockey stick). For comparison my iPhone has 3.3 Billion transistors on its main CPU, this stands in contrast to the 100 Million transistors a high-end laptop had ten years ago and the five dozen on an advanced IC when Moore made his prediction in 1965.
Thanks in part to Moore’s Law, the technologies highlighted above, and most other information-based technologies are advancing at an exponential rate. However, we are ill-equipped to understand exponential progress.
When we attempt to imagine what the world will look like at some point in the future, we make predictions based on what we perceive as historical evidence of constant change. On closer inspection, it becomes clear to see that our rate of technological progress has in fact been improving exponentially. In part, we miss the increased pace of change because of the nature of human adaptability. It turns out that we are only impressed by any new technology for a brief period when we first see it, and once the novelty subsides, it simply becomes a tool to use which quickly fades to the back of our consciousness. Ray Kurzweil refers to this concept as “The Law of Accelerating Returns”.
“When is the last time you marvelled at the capabilities of your internet connected smartphone?” How do you think someone from 30 years ago would react to your phone?”
The inevitable consequence of our adaptability means that we mostly underestimate the future. Without needing to state the obvious, predicting the future technological advancement is near on impossible. However, if most of us were to try we would probably do a base case assessment, start with some history and make some extrapolated prediction that might look like this;
This form of prediction will however significantly undershoot the actual level of progress we are likely to make. Based on our current trajectory, we, as a society will advance technologically more in the next 20 years than we have in the previous 200 years. On an adjusted graph, that looks like this;
Why thinking straight is a problem
What is worth noticing in the two graphs above is that the line to our left (our historical progress) appears as a gradual curve, whereas in figure 2 the line to our right (our future progress) is incredibly steep. The steep incline is in part because of the nature of exponential numbers. Take transistors on a chip as one measure of growth; it took more than eleven years to produce the first integrated circuit that contained more than one thousand transistors. By comparison in the past decade, transistor count has increased roughly from one hundred million to more than three billion. When something has been doubling for some time, there will always come the point when every future doubling makes all past activity feel like a rounding error.
Additionally, the progress we are now making is cumulative in nature. So far we have connected 3.5 billion people to the network and provided reliable and cheap devices for connectivity. However, this progress is essentially just providing the plumbing for the main event, which I believe is yet to begin. We have been here before, in the 1910s and 1920s after the establishment of electric utilities, factory electrification was widespread. The fact that most factories that had replaced their steam engines with electric motors saw little improvement in productivity was apparently at the time a surprise. When we look at this example in retrospect, it turns out that factory owners simply didn’t know how to take advantage of their new power source at the time. In most cases, productivity improved by orienting production machines around workflows rather than the straight lines that were previously required by their steam driven machines. This improvement took a mere twenty years to implement.
The way most traditional businesses prepare for the future is similar to our linear prediction model at Figure 1: above. In this model budgets and expectations rarely deviate by more than single-digit percentage points. This method might have worked in the past, but increasingly I believe that thinking this way is a dangerous strategy. On the one hand, if you operate in a market that is subject to digital disruption, and you set your expectations in a linear fashion you will be consistently caught off guard by downside surprises, and you will significantly diminish your chances of survival due to a lack of realistic long-term plans. On the other hand, if you are in an exponentially growing market and you limit your vision to linear growth you will quickly become irrelevant when your competitor doubles in size every year, compared to your 5 percent growth.
- Understand Exponentials
- Think Bigger”