The moment when you can now take an online class on building self-driving cars from a for-profit education platform is probably the moment where we have officially gone from a world where autonomous vehicles were a trope found in pop cultural futurism to one where they have become an expected soon-to-be feature of our lives. Between Google’s self-driving car project and Tesla’s Autopilot feature, we spend so much time talking about autonomous driving that it can almost seem absurd that we still have to suffer through our morning commutes in our primitive dumb cars. When the first commercial self-driving car is finally in our hands, its release will almost be no more surprising than a new iPhone. It is about time, we will all say. What took you so long?
In that sense, self-driving cars can be seen an obvious innovation built on a multi-decade trend of rising computation power, increasing network connectivity and cheaper hardware manufacturing. Anyone who understood what computers and sensors can achieve together saw it coming. In 1998, kids all across the world were introduced to the LEGO Mindstorms robotics kit and one of the most basic projects they could build was a simple robot that followed a path on the floor using a primitive light sensor. Many of them asked the logical follow-up question: What about cars? The future was clear.
In the 90s, researchers were publishing computer vision papers on road lane detection. Mobileye, the Israeli company that powers the vision hardware of most car manufacturers including Tesla, was founded in 1999. Serious efforts soon started to take shape in the domain. The first DARPA Grand Challenge was held in 2004 in the Mojave Desert and the university students who took part in that and subsequent races are today working at the likes of Google and Tesla to take their work to its logical conclusion.
Two decades of incremental improvements finally brought us to this moment where something that seemed very obvious to a middle school student with a LEGO kit over a decade ago is finally about to become reality. On hindsight, this journey seemed predestined. To be fair, along the way there were and continue to be difficult challenges that require non-linear thinking to overcome, but the ultimate destination has remained unchanged. It is an extrapolation of a trend and not the creation of one. The eventual spread of autonomous cars may give birth to new discontinuous trends in the foundations of our social and economic structures – or just the extrapolation of existing digital trends to the physical world – but in any case the technology itself is the result of evolution and not an isolated act of spontaneous innovation. Sans any prejudice, one might call it an obvious innovation.
In Silicon Valley, we have perfected the art of obvious innovation. We saw in eBay the power of the marketplace in a digital world with negligible transaction and scaling costs. From that realization followed Uber, Airbnb and others. The flexible software infrastructure that made early consumer products like Amazon and Google scalable are today the building blocks of enterprise solutions offered by countless startups. As decade-old lessons on recommendation systems and real-time analytics finally diffused to traditional industries like healthcare and transportation, there are certainly great opportunities for innovation and disruption, and numerous multibillion companies were indeed created on those premises. But these are all examples of obvious innovation, of applying a known solution to a new problem domain. The nature of venture capital creates a bias towards this form of innovation. As much as investors like to believe that they take the long view in supporting technology growth, the well-honed fund raising cycle we have today almost guarantees the perpetuation of this unconscious bias. Witness the rise of enterprise software.
None of this is a bad thing. We need continuous improvements to fully realize the promise of technology in providing us with a better world. It is absolutely incredible that with a device in my pocket today I can instantly communicate with any place on Earth and then make arrangements to transport myself there. Obvious innovation played a key role in getting us to this futuristic present, but non-obvious innovation will be necessary to take us further.
Taking a step back, one can make the claim that the invention of the transistor was the singular non-obvious innovation that played the greatest role in bringing us to where we stand today. While a massive over-simplification, I think it is not indefensible. Every continuous innovation we experience today builds on the increasing density of transistors and the associated computation power. That initial proof of feasibility was essential and it involved the discovery of quantum mechanics – a huge paradigm shift from how we thought about the physical world. To borrow from Peter Thiel’s words, this was when zero became one. Going down the obvious path, we could have eventually built the best computer out of vacuum tubes that physics will allow and not come close to achieving a fraction of the capabilities that we possess today. The limits of the incremental approach is often not the global maximum.
Whenever I am reminded of this topic, it is always at this stage in my thought process that I ask myself: What are the non-obvious innovations that we can think of going forward? Sure, increasingly sophisticated automatons are likely to displace human workers and dramatically reshape the labor market in the near future, but will they still be powered by incrementally better lithium ion batteries? We will definitely fly in planes that are incrementally more fuel efficient and connected to the Internet of Things, but will we also fly in hover cars or instantly travel across the world by simply uploading our minds to a different body? It is harder to anticipate the non-obvious innovations because their realization often requires fundamental breakthrough in basic research that is not guaranteed to ever be forthcoming until it suddenly happens. These are the black swans that irreversibly change the direction we are moving in.
Of course, the line is not as clear as I paint it. Within every obvious long-term trend is a series of small non-obvious discoveries and advancements. In order for the invention of the transistors to ultimately bring us the modern internet, people had to figure out encryption, information theory and countless other subjects. Conversely, every significant advancement often stands on the shoulder of the giants that came before.
Still, I find the categories of obvious and non-obvious innovation useful to keep in mind as we think about where digital technology is taking us. As we await an expected future of self-driving cars, smart personal assistants and augmented reality, be on the lookout for that truly disruptive breakthrough that can completely alter our calculus. Like the Mule in Isaac Asimov’s Foundation series, our informed predictions of the future always come with an unspoken caveat that everything might be different. Expect the unexpected.