Silicon Valley Stumbles in World Beyond Software

The first guy to head the project was one Nicholas Roy, an MIT Robotics professor whose professional claim to the position seems to be mentoring a pair of students who succeeded in building a drone that flew around a parking garage - predetermined layout, no weather. Completely out of his depth, he failed.

After he failed, Google hired Dave Vos, another guy who spent some time in an MIT graduate office - this time one in an aerospace specialty - who spent the last couple of decades in tech management positions. At least he had the experience to come in and say, hey, maybe we need some wind tunnels, maybe we need some professional tooling that's taken for granted in most aerospace companies, because trying to build an aircraft without that tooling is like trying to run a web-facing application off developer workstations instead of setting up a proper production architecture. And he, too, failed.

The real problem that Google (and probably Amazon too) faces is that they need to synergize two very different engineering teams to make delivery drones work. You need aerospace folks to build reliable aircraft, you need AI folks to give those aircraft targets to aim at, and you need them to sit on the same floor in the same building. The aerospace folks aren't going to move to the Valley because there's no aerospace hub in SV, and SV isn't a good place to start one because the necessary land - you know, the wide open spaces you need for runways and hangars - is a pipedream in SV. And the AI folks, you know, the ones the Big Four are paying close to a million dollars in salary and begging not to leave for a competitor - are never going to leave SV to work on delivery drones.

The best course of action that Alphabet has is to acquire a proven aerospace company, give it some great AI folks, and come back in a year to check on their progress. But most of these Project X types like their ivory towers. It's how Google convinced them to join Project X in the first place, selling them on the promise that they'd be in the same ivory tower they were in back in MIT, just with guaranteed funding and a request to think about commercial viability. So, not gonna happen.

And we have the example of Peapod and WebVan which didn't just foretell but proved unless it was for specialized markets, it's a proposal for hemorrhaging money. Even if you get automation and you perfect the travelling salesman's problem you're still mostly dealing with low value products. Now, if it was delivery of expensive (Rx) drugs, or maybe even heavy but valuable items like gold or jewelry, legal documents, etc. okay. people will pay for that and you can make a profit but toiletries and pantry goods?

The same is gradually happening with things like batteries, motors, etc. Everything is becoming more standardized and more like software.

The current limitations have to do with both physics and engineering, and also software reliability, but those barriers are all receding. Solar Cell roof shingles are an idea likely had by many kindergarteners, but the genius has to do with making a longer-term bet and creating the financing to make it into a viable business in a world that rewards short-term results.

The important thing is that a Billionaire today can make a bigger impact than a superpower government could a decade ago. The glorification of the bygone era of massive infrastructure investment voiced by Thiel at the GOP convention can be alternately viewed as a wish that we had more Musks taking big gambles and a general cynicism about the tech elite who develop cults of personality after a single, largely luck-driven, often low-tech home run.

The Interstate highway system was a buildout that could happen because of the solving of coordination problems that had prevented escape from a local maximum. As governments lose relevance, it is up to the Musks of the world to usher in both the technological vision and the social coordination solutions.