Dreams of the Concorde
The aviation grand dream before 2001 was supersonic transport across the world, bringing it even closer than ever before. The technology would perfected, "futurists" thought, and the commodification of supersonic technology would lead to cheaper flights, faster transport of cargo, even notions of flying cars. Yet, a combination of economics issues and low demand, topped with tragic accidents grounded the Concorde in 2003. While I usually hesitate to draw parallels between fields because it is often low-hanging fruit for making yourself feel clever, I do wonder if we're reaching that point for the tech industry.
The first 60 years of aviation brought us from wooden Wright planes to helicopters, strike fighters, and the Concorde. The last 60 years of aviation has brought us advances in military aviation, but almost nothing in commercial aviation. The easy problems had been solved, and the commercial aviation industry withered and consolidated. Sheer forces of economics brought down the rate of change, and the only real innovation in the aviation industry in the past two decades has been the rise of low cost carriers, which are more of an economical innovation than a technological one.
The tech industry has taken a similar path to this. The first few technological innovations have dramatically shifted the industry, we went from UNIX on a terminal and the SNES to Kubernetes and the Switch. We're filled with dreams of self-driving cars and fears of an all powerful AI. But what if this is just us expecting the same trajectory of innovation of technology to continue? If the tech industry continues to innovate at its current rate, we'll have self-driving cars in a decade.
Yet, there are some warning signs that we're beginning to hit the end of "the easy problems" in tech. Intel has very publicly struggled in manufacturing sub-10nm chips. Moore's law is widely predicted to deviate from reality in the near future. Discussions on AI revolve around "deepfakes" and notions of an oppressive all-knowing AI, but the discourse is completely unhinged from the reality of "AI", where we can barely get a good hot dog/not hot dog classifier. The majority of programmers do not work on industry changing technology, instead they're shackled to creating websites that talk to a single DB over and over, or some variation on RPC services. Getting from a single transistor to a smartphone was incredibly quick (80 years!) but I suspect the road ahead, starting from smartphones to self-driving cars will be far, far more difficult. A few tragic accidents, coupled with sheer economics, may sink our dreams of self-driving cars just like the Concorde.