Caleb Watney - America's Innovation Engine

Dwarkesh Patel opens by asking Caleb Watney, director of innovation policy at the Progressive Policy Institute, to explain his Atlantic argument that America's innovation engine is slowing. Watney concedes that the engine is a broad, multivariate thing, so he narrows the claim to three trends he finds troubling for America's capacity to innovate: high-skill immigration and international talent flows, the country's cutting-edge university system, and the industrial clusters, or geographic agglomerations, from which much innovation comes.

He says each of these was already showing negative trends before the pandemic, and that the pandemic may act as a breaking point for all three. The conversation starts with the third, clusters and remote work, and later turns to immigration and other topics.

Asked why remote work cannot make up for the loss of physical agglomeration, Watney points to a long economic literature on geographically connected networks of entrepreneurs, innovators, scientists, academics and supply chain managers. When these people share a region they run into each other spontaneously, trade ideas, and hold long hallway or office conversations that cannot simply be replicated online. In his account this produces more creativity, more ideas and more patents, and he says there is substantial real-world evidence for it.

He grants the counterargument that online platforms have multiplied over twenty years and might replicate much of the effect. His reply reframes the comparison. Someone inside a physical cluster is not choosing between physical and digital interaction, because they already have both, or, in his words, "the best of both worlds." The real choice is physical plus digital against digital alone, and he thinks the gap there is still large enough that losing existing high-productivity clusters would be a shame.

He draws on his own experience. Twitter gives him many ideas, but an extended two-hour conversation with a friend in their office, with eye contact and spontaneous ideas, is a different thing. He says Zoom struggles to reproduce this partly because Zoom meetings tend to be scheduled, leaving little flexible time. The best real-world conversations, he says, often start as something else and grow, whereas people deep in remote work suffer Zoom fatigue and do not want to stay on a call to chat with someone extra.

Watney does not frame the two modes as contradictory. He argues that people need a better taxonomy of the interactions that lead to innovation: spontaneity and in-depth conversation seem to be available mainly in physical spaces, while digital spaces offer access to far more people. He would prefer that the discussion break down the specific attributes of each kind of interaction rather than ask which is better.

Patel argues the opposite side. Remote work could connect the whole world into one hyperproductive city that cannot exist physically, both because of limits on how large and open cities can be and because of zoning laws, such as San Francisco's, that artificially restrict who can join the innovation scene. Talent in developing countries and in rural America could then share in the benefits, and he asks why that would not outweigh the loss of physical agglomeration.

Watney agrees that digital networks can have huge benefits and that they allow arbitrage of bad regulation such as zoning and of physical distance. He repeats, however, that the relevant alternative is not physical or digital alone. People in Washington, D.C., Silicon Valley or New York can already talk with people around the world and even hire remotely, and they additionally get a cultural context that he says only comes from being in person and sharing conversations.

Patel suggests one reason Zoom falls short: in a coffee shop or office, a person can hang around in case someone might help them leave the company and start another, or even find a mate, without ever stating that motive, whereas Zoom requires explicit communication of purpose. Watney agrees and notes that some workplaces experimented with leaving cameras on, as if working next to someone in an office, but people seemed uncomfortable and disliked it. He is unsure why the easy, spontaneous conversation does not carry over.

He does not rule out a future, perhaps thirty years away or sooner, in which virtual reality has blossomed and everyone works in virtual workspaces regardless of location, which could diminish the importance of physical agglomeration. On the present he is direct: "But I certainly don't think we're there yet." He sees the shift of Facebook, Twitter and other large tech firms to all-remote work for the foreseeable future as a large experiment whose effects he will watch.