There is a thing that happens in cities – that we think happens in cities – when people with lots of different ideas bump into each other on the pavement, or at the bar or the grocery store or the gym.

Together, they think up things that would never come out of a conference room or the kind of coffee meeting that has a calendar invite. Weird new ideas take root. Innovation follows.

The urbanist icon Jane Jacobs identified these collisions as central to what makes cities dynamic. Her followers think of them as the product of serendipity. Economists have their own name for the almost-magical benefit these connections create: knowledge spillovers.

“The chance encounters facilitated by cities,” economist Edward Glaeser has written, “are the stuff of human progress”.

Remote work has, well, blurred this picture. Can you have serendipity two days a week? Where do people bump into one another when the downtown coffee shops are closed? How do workers spill their knowledge when they’ve moved to Montana, or the exurbs? Does that even matter anymore?

“It’s a trying time, certainly, for my view of the world,” said Enrico Moretti, a Berkeley economist who has written extensively about why it’s good for workers, companies and the economy when people cluster in particular cities.

GET BT IN YOUR INBOX DAILY

Start and end each day with the latest news stories and analyses delivered straight to your inbox.

Sign UpVIEW ALL

The clustering itself is a thing that matters, economists have argued, because it helps people trade ideas, land better jobs and find others doing highly specialised things. And those benefits of what economists call “agglomeration” have theoretically grown more important as America has shifted over decades to an economy built on ideas.

The question today isn’t just whether white-collar workers can be more productive sitting alongside their colleagues in the office. It’s whether it matters for them to be alongside their colleagues – and near workers at other companies, and other industries, and other people who may be distant acquaintances or simply familiar faces who think about totally different things.

During the pandemic, plenty of companies got by just fine with their workers scattered at home. Many of those workers concluded, “I’m just as productive, too!” And people generally did the opposite of clustering: They moved farther out into the suburbs, and in growing numbers, college-educated workers left the very places economists say are the most productive, including the Bay Area and New York.

Moretti concedes that it may be years before we understand the effects of remote work. But he suspects it will become increasingly clear that we’ve lost something valuable. Yes, maybe you’re just as capable of doing your daily work tasks at home. But what about the idea you don’t realise you don’t have because you never ran into a friend-of-a-colleague at the bus stop?

“I still strongly believe in the economic forces that were at play before Covid, the forces that have created agglomeration in cities and communities for the past 2,000 years,” Moretti said. “I don’t think they’re gone.”

Maybe, though, we should think about them differently?

“Agglomerations in space still matter – but agglomeration in time maybe we got wrong,” said Karen Chapple, who directs the School of Cities at the University of Toronto. “Maybe you don’t need to agglomerate every day.”

She has used location data from cellphones to track the return of people downtown across North American cities. Many places didn’t shut down for long and have relatively little remote work today.

But downtown activity levels – the stuff of serendipity – have particularly lagged in cities central to the innovation economy, including Seattle and San Francisco.

New AI firms, seemingly all in one area

San Francisco, for all its vacant offices and heavily remote tech companies, also offers some of the best evidence that agglomeration still matters. Nascent artificial intelligence (AI) companies have been signing new leases in remarkable proximity to one another. Real estate brokers have already branded the cluster “Area AI”.

AI is a perfect example of an industry that should benefit from agglomeration. It’s young and rapidly evolving. Its funding is concentrated in the Bay Area, too. The people working on it are highly specialised. Their ideas are the definition of cutting-edge.

“For all of us who have been in tech for a little while, this is the fastest feeling of change and advancement and real-life magic that I think we’ve ever felt,” said Barry McCardel, the co-founder and CEO of a company, Hex, building AI-powered tools for analysing data.

That’s a feeling tied to a specific place (where Hex has just doubled its office space). The industry certainly has remote workers, too. But for many, it’s no substitute to be a plane ride or Zoom connection away.

“These ideas are so illegible,” said Kanjun Qiu, who co-founded the AI company Imbue. They’re big and complex ideas – it’s hard to Slack them, or email them, or put them in a white paper.

Exactly how these in-person collisions work – how they turn into ideas, then innovation, then human progress – is still a bit mysterious. Tom Wolfe observed 40 years ago that workers in the Silicon Valley semiconductor industry met after-hours at the same bars to trade stories of their progress (Qiu and several others in AI today lived and socialised before the pandemic in the same San Francisco group house).

The economists David Atkin, M Keith Chen and Anton Popov have more recently tried to identify these effects by using geolocated cellphone pings to track where workers from different firms bump into one another in the Valley. When two firms are in places where workers tend to cross paths more often – including workers with seemingly little business connection to each other – more patent citations between those firms follow.

“The mechanism we have in mind is not: ‘I’m in the line at Starbucks, I strike up a conversation with a stranger, and I’m spilling the beans on my firm’s latest technology three minutes later,’” Atkin said.

Rather, he said, think of Silicon Valley as having an underlying social network of friends-of-friends, former college classmates, onetime co-workers and the like. People running into one another at the bar or supermarket activate links on that network and begin to chat.

Popping some Internet bubbles

The question, then, is how often? If white-collar workers choose two, three days a week, that affects everyone else potentially colliding downtown, too: restaurant servers, bartenders, dental hygienists, small business owners. Coffee shops that were afloat five days a week can’t survive on two. Then other collisions that might have happened in the past disappear, too.

Glenn Kelman, the CEO of Seattle-based Redfin, worries not just about innovation, but civil society. Remote work makes it easy to inhabit Internet bubbles, he said; the physical world forces people who are different together.

It is hard, though, to ask workers to give up flexibility with childcare, or more affordable housing in another city, in the name of agglomeration and innovation. Maybe that trade-off isn’t worth it.

Some cities may eventually evolve with an answer. Downtown could have fewer offices, more residents and attractions, and other kinds of collisions not so dependent on daily office culture. But for now, what kind of place is a city that creates serendipity just a few days a week? NYTIMES