Caleb Watney, director of innovation policy at the Progressive Policy Institute, argues that America’s innovation engine is slowing due to three interconnected trends: declining high-skill immigration, strain on the university system, and weakening of industrial clusters. The pandemic may serve as a breaking point that accelerates these negative dynamics.
Agglomeration effects and remote work
Physical clustering of entrepreneurs, scientists, and innovators in regions like Silicon Valley, DC, and New York generates spontaneous idea-sharing, more patents, and greater creativity that digital tools cannot fully replicate.
Remote work can connect global talent and arbitrage bad zoning laws, but the relevant comparison is not physical versus digital agglomeration—it is physical + digital versus digital alone. Currently, people in innovation clusters get both.
In-person interactions allow for unplanned, extended conversations with eye contact and flexible time that scheduled Zoom calls struggle to mimic. People also experience Zoom fatigue and are reluctant to linger after formal meetings.
Some workplaces have experimented with always-on cameras to simulate co-working, but people find it uncomfortable and it fails to reproduce spontaneous interaction.
Virtual reality could eventually reduce the importance of physical proximity, but the technology is not there yet. The long persistence of Silicon Valley despite extremely high rents suggests the productivity benefits of clustering remain large.
Remote work may benefit individual firms but reduce public spillovers: fewer new firms are formed, and idea generation across the ecosystem slows. Companies like Facebook do not internalize the cost of reduced entrepreneurship in the broader economy.
Why it matters if America loses its innovation lead
Technology platforms embed cultural values—free speech norms, intellectual property frameworks, private-public sector boundaries—that become global infrastructure. American-built platforms have historically reflected broadly liberal democratic values.
If Chinese firms instead built the dominant global platforms, especially in transformative fields like AI, very different values would be baked into the infrastructure shaping the future.
China’s authoritarian model may be good at catch-up growth but faces structural barriers to frontier innovation, including difficulty retaining top talent. Over 90% of Chinese AI PhD students who study in the US choose to stay.
However, the US cannot be complacent. American governance institutions have become more sclerotic, making it harder to mobilize quickly even if a competitor emerges as a clear threat.
Historically, the US has “coasted” in its innovation role and then kicked into high gear when challenged (e.g., after Sputnik). But there is concern that institutional rigidity may prevent such a response today.
Immigration policy
Tom Cotton’s proposal to bar Chinese students from STEM fields is counterproductive. China’s CCP explicitly identifies talent retention as its biggest barrier to technological leadership. Broad restrictions on Chinese STEM students would keep that talent in China—exactly what the CCP wants.
Chinese industrial espionage is real but poorly measured. More targeted approaches—background checks on sensitive projects, clearer paths to permanent residency to reduce CCP leverage—are preferable to blanket bans.
Eric Weinstein’s argument that STEM labor shortages are a myth designed to suppress wages ignores the positive-sum nature of high-skill immigration. Immigrants patent at higher rates, start businesses at higher rates, and make their American counterparts more productive through agglomeration effects.
Cutting programs like H-1B increases offshoring: when companies cannot bring talent to the US, they move operations abroad. This dynamic will worsen with the rise of remote work.
H-1B reform priorities:
Replace the random lottery with a salary-ranked (including equity-adjusted) system to prioritize the most highly skilled candidates.
Make it easier for H-1B workers to switch employers or start their own businesses, rather than being locked to a single sponsor for years.
The O-1 visa (for immigrants of extraordinary ability) is underutilized, has no congressional cap, and is not tied to a specific employer. A new USCIS director could broaden its criteria through regulatory or guidance changes without needing new legislation.
Pausing H-1B during the recession is misguided. The hardest-hit jobs are in-person service sectors, not the high-skill roles H-1B fills. Regions with historically higher H-1B usage have shown higher job growth and wages for native workers.
Immigration is the highest-leverage policy change because the US is so far from the global optimum. Moving 5 percentage points toward freer movement of skilled people matters enormously, whereas equivalent marginal improvements in trade policy (already near optimal) would have little effect.
Big tech, antitrust, and competition
Breaking up big tech companies is a high-risk, low-expected-value strategy. If the premise is wrong, the US destroys its most productive firms and hands an advantage to China.
A better approach is to tear down entry barriers and increase competitive pressure: make it easier for startups to access high-skill talent, reduce regulatory burdens that disproportionately favor incumbents, and open government-held datasets.
Big tech’s immigration advantage is structural: only large firms have the HR infrastructure to navigate the visa bureaucracy. Many talented immigrants who prefer startups end up at big companies solely because that is the only viable immigration path.
Data is not a homogeneous moat. It is context-dependent, requires massive internal infrastructure to utilize, and has value primarily through dynamic feedback loops (seeing how platform changes affect user behavior), not as a static asset that could simply be handed to competitors.
EU regulation like GDPR has increased the market share of Google and Facebook because large firms can afford compliance teams while startups cannot. Complexity functions as a subsidy to incumbents. The US, not Europe, is where the most innovative competing startups are actually being built.
Federal R&D
Federal R&D spending as a percentage of GDP has been declining, and there is too little experimentation in how grants are structured.
New Zealand experimented with randomly distributing research funds to qualified projects and found it successful. The US could run randomized trials on variations of its funding apparatus to identify improvements.
The meta-problem is that the people who would design such experiments have vested interests in the current structure—illustrated by the Thomas Sowell story of being rejected by the Labor Department for wanting to study minimum wage effects.
Government R&D is most needed where private firms will not invest: research that is unpredictable in outcome, unlikely to pay off on a profitable timeline, or not sufficiently profitable. Private R&D focuses on predictable, timely, and profitable questions. Breakthroughs are inherently unpredictable, so companies underinvest in them.
Many scientific advances are inexcludable and inexhaustible—worth the investment for the economy as a whole even if no single firm would fund them.
Undervalued issues and trends
Most undervalued policy issue: How industrial clusters form and scale. It is poorly understood why some regions become self-reinforcing innovation hubs while deliberate attempts to seed new clusters usually fail.
Most undervalued future technology: Climate megaprojects—such as spreading carbon-absorbing olivine on beaches or tapping Yellowstone’s geothermal energy (which also reduces supervolcano risk). These are under-hyped relative to their potential impact.
Most undervalued social trend: Falling fertility rates and demographic aging. Older societies become more conservative and less risk-taking, with fewer people willing to work extreme hours or leave stable jobs to start companies. Even young people in aging societies become less entrepreneurial. Sub-Saharan Africa and parts of Asia, with rising populations, will become increasingly important.
Advice to young people
Focus on problems where the profit motive alone will not deliver solutions. If there is money to be made, someone will eventually solve it; the highest-impact work is in areas the market neglects.
Use an effective altruism framework: combine what matters most with where your marginal effort has the highest impact and where you have comparative advantage.
For Watney, bad policy is the critical bottleneck in America’s growth engine—the missing piece in an “O-ring theory of growth”—and fixing it, while not personally profitable, can have enormous societal impact.
