a16z Podcast | Startups as Science Experiments -- Can VC Disrupt Academia?

Bill Janeway's Hypothesis on Venture Capital and R&D:

• Venture capital success in IT was preceded by four decades of federal R&D funding.
• Biotech success followed two decades of federal R&D investment.
• Clean tech failed due to a lack of federal R&D "seed corn," as government funding bypassed research for direct industry subsidies.
• Forward-looking principle: Future VC investment should target sectors currently receiving federal R&D money, anticipating commercial viability 20 to 40 years later.

Emerging Models for Basic Research:

• In the absence of robust federal funding, research may shift toward an "open, networked, collaborative, peer-to-peer" model.
• Key drivers for this shift include the internet as a coordination mechanism, the rise of open source data/design, and globalization of R&D talent.
• A trend toward "bottoms-up" research is anticipated, characterized by more interdisciplinary mixing and less formal organization compared to the 1950s NIH/NSF model.

Venture Capital Approaches to Science:

• VC firms are experimenting with "science experiments" involving $1M seed investments at $1M valuations (1:1 or even 1:3 splits).
• Counterargument: Research may be compromised by short-term financial incentives if distinguished from commercial development.
• Primary lever identified: Philanthropy from high-tech founders is expected to be the major upside surprise in the next 20–30 years, potentially flowing 10 to 100 times current amounts into universities.
• Enlightened universities are increasingly prioritizing philanthropy and flexibility for professor spinoffs over traditional patent licensing or university venture arms.

Student Trends and Educational Priorities:

• Undergraduate Interest: Strong dominance of Computer Science (CS), with a specific interest in the intersection of CS and biomedicine for personalized medicine and human augmentation.
• Cryptocurrency/Distributed Systems: Significant student attraction to decentralized systems, viewed as a potential robust redesign of the internet's core economics and monetization layers.
• Liberal Arts Tension: Efforts are underway to apply the "entrepreneurial spirit" of CS to social sciences and arts (e.g., computational social sciences), countering the view that practical skill acquisition usurps the "life of the mind."
• Philosophical Debate: A conflict exists between the "life of the mind" (Alan Bloom) and "engineering/making things," with the speaker arguing for the integration of technology into all university fields.

Regulation, Innovation, and Regulatory Arbitrage:

• Eroom's Law: The cost to develop drugs doubles every four years, and development timelines elongate, contrasting with Moore's Law.
• Risk Aversion Trend: As societies mature, they collectively choose lower risk, resulting in reduced innovation until external "barbarians" disrupt the status quo.
• Strategic Response: Instead of fighting the system, startups should pursue "regulatory arbitrage" by establishing operations in jurisdictions with friendly environments (e.g., stem cell research in Korea).
• Local Zones Proposal: Cities should consider legalizing specific high-risk innovations (e.g., self-driving cars, drones) in designated zones to enable innovation without requiring immediate federal legalization.

Geographic and Cultural Shifts in Innovation:

• The "East Coast vs. West Coast" dichotomy is descriptive rather than prescriptive; innovation is now a "state of mind" rather than a geographic location.
• Global Explosion: Entrepreneurship is surging globally, with specific mentions of underground startup ecosystems in the Middle East (Jordan, Egypt, Iran) featuring high female participation (40–50%).
• The low cost of computing and internet access has decentralized the ability to build startups, creating a global movement unconfined to Silicon Valley.

Machine Learning and Sensor Deployment:

• Application Scope: Machine learning is most immediately impactful in classification tasks across finance, healthcare, and resource allocation.
• Catalyst: The convergence of massive sensor deployment (3 to 6 billion smartphones with cameras) and big data allows ML to revolutionize sectors like transportation (emissions reduction) and financial services (Bitcoin).
• Near-term Innovation: Significant new products can be built by combining existing ML techniques with sensor data without waiting for fundamental breakthroughs in AI core science.