Derek Thompson has a piece in the latest issue of The Atlantic where he argues that what he calls the Eureka Theory of History is wrong, and what conclusions policy makers should draw from that.
In case you’re not familiar, The Eureka Theory is a narrative that structures the story of human progress around genius inventors (and as such, it bears a lot of similarity to what historians call the Great Man Theory).
What’s wrong with the Eureka Theory, Thompson argues, is that brilliant inventions are less important than how society chose to deploy them.
Example: Edward Jenner’s discovery that led to the smallpox vaccine was a landmark scientific achievement. But that accomplishment was dwarfed compared to the collective effort of rolling out the vaccine globally, to eventually eradicate a virus that has killed some 300 million people.
Building on the distinction between impressive discoveries and the dull but ultimately impactful process of implementing them at scale, Thompson maintains that the doctrine which underpins American innovation policy since the Second World War is wrong. And not just a little wrong, but wrong to its very core.
It all seems to be Vannevar Bush’s fault. He was head of R&D for the American war effort, and as peace ensued he published the influential article “Science: The Endless Frontier”, wherein he counselled the government to focus its spending on basic research and leave deployment to private companies.
That’s where things started going south according to Thompson. He sees how American progress is stalling and he thinks it’s because its science and innovation policy is set up in such a way that it fails to reap the benefits of the discoveries it helped foster.
Examples: The microprocessor came together out of a number of inventions funded by American taxpayers, but Taiwanese companies now have the market cornered. Same thing with solar panels: US inventions that are making dents overseas. Same thing with mRNA vaccines, and with a long list of other technologies too.
In startup parlance, one could say that America is failing to capture enough of the value it creates.
I came across Thompson’s firebrand article while reading Kai-Fu Lee‘s book AI Super-Powers : China, Silicon Valley and the New World Order.
Kai-Fu Lee is a man of many faces. Born in Taiwan, he came early to the US. At Carnegie Mellon he studied under Hans Moravec and went on to do trail-blazing research in the fields of machine learning and speech recognition. During the nineties and early aughts, he was a high ranking lieutenant in both Apple, Microsoft and Google; the Chinese branch of which he headed. Then he switched games and became an investor. For the last 25 years he’s been based in China, where he’s said to be one of the most popular social media influencers.
It was interesting to overlay Lee’s views with those of Tompson’s. Both see the split between invention and deployment as crucial, but where Tompson’s glass is half-full, Lee’s is brimming with optimism. From his vantage point, it seems that China is quickly catching up with the US in the race towards AI supremacy. The reason for that, Lee argues, is that AI entered the age of deployment.
Which means that much of the value created in the field of AI is now generated by implementing the key findings that came out of primarily US, British and Canadian labs over the last decades.
In previous industrial revolutions, it used to be that geography mattered. The regions where key inventions were made, became global leaders and remained so for decades, if not centuries.
That dynamic has now changed in a number of ways, most of which tips the scales in favour of China.
First of all: Most important breakthroughs are now coming out of academia, as opposed to corporate labs. That means patenting is frowned upon and that the system has built in incentives driving researchers to quickly publish their findings.
Second of all: The nature of those findings are such that they travel fast. No need to ship engine parts across oceans, just download the latest algorithms and get started.
Third of all: Where scientific breakthroughs are often the crowning achievements of life long pursuits of knowledge, building and quickly scaling up a startup on someone else’s insights typically requires just enough scientific expertise, and well honed tinkering skills*. And training tinkerers is extremely fast and cost effective, compared to training scientists.
*I’ve previously written on the distinction between tinkerer’s and scientists and how modern day engineers have both strands of DNA in their blood streams.
Fourth of all: While game changing scientific discoveries are few and far between policy is now what can move fast and create unfair advantages, and it does so by making it more or less easy for corporate players to access data; the key to remaining competitive in AI’s age of deployment. Example: The complete lack of privacy protection granted to Chinese citizens*, as contrasted by the European legal framework which is clearly on the side of the consumer, makes for fantastic opportunities if you’re an unscrupulous AI entrepreneur.
* One of my beefs with Lee is how blatantly unapologetic he is about the Chinese police-state. Here’s how he’s describing the tradeoff between ‘convenience and security’: “There’s no right answer to questions about what level of social surveillance is a worthwhile price for greater convenience and safety, or what level of anonymity we should be guaranteed at airports and subway stations.” Oh really, no right or no answer? I wonder what the millions of persecuted Uigurs would make of that…
All in all: Chinese entrepreneurs take the not-invented-here meme and flips it on its head: where among western innovators nothing is more stigmatised than copy-cats, the Chinese take pride in being good at capitalising on breakthroughs coming out of other nation’s universities.
At least if we’re to take Kai-Fu Lee’s word for it.