Chronological feed of everything captured from Patrick Collison.
Global solar deployment reached $500 billion annually in 2023, equivalent to 0.5% of world GDP and up 43% year-over-year. This pace equates to installing solar panels on 3-4 acres per minute based on 300-400 GW deployments. In the US, solar will comprise 54% of new electricity capacity, surpassing all fossil fuels, while Europe and regions like California and Spain show similar dominance in renewable generation.
Patrick Collison, co-founder of Stripe and ARC Institute, discusses his entrepreneurial journey, the evolution of Stripe as a critical economic infrastructure, and his vision for accelerating discovery-based scientific research through the ARC Institute. He highlights the importance of user-centricity in business and curiosity-driven research, emphasizing the need for upgraded economic infrastructure and knowledge expansion for global prosperity.
Sam Altman highlights ChatGPT's practical utility for summarization and anticipates superlinear AI capability gains, dismissing data bottlenecks via synthetic data and new architectures. He advocates mechanistic interpretability over RLHF for robust alignment, proposes an IAEA-like global agency with compute/capability thresholds and audits for powerful AI systems, and expects open-source models to lag hyperscalers but enable broad applications. Altman predicts AI agents will integrate into society like companions, boosting economic growth through superior capital efficiency, while emphasizing capital-intensive breakthroughs and talent development.
Patrick Collison advocates a "micro-pessimist, macro-optimist" mindset for startups: rigorously critiquing current shortcomings while maintaining conviction in long-term technological progress. In turbulent markets, prioritize ruthlessly, embrace business fundamentals like economic compounding over pure technologist instincts, and build strong leadership teams post-product-market fit. Vertical SaaS offers massive untapped potential to digitize analog sectors globally, while AI nears a 2023 tipping point despite prior hype cycles, urging entrepreneurs to focus amid capital scarcity for differentiation.
Post-WWII centralization of US science funding via NIH and NSF has led to excessive standardization and bureaucratization, reducing adaptability and speed compared to historically diverse models. Experiments like Fast Grants demonstrate value in rapid, flexible funding for urgent problems, suggesting need for structural heterogeneity alongside behemoths. US construction slowdown traces to 1970s regulatory shifts like EPA creation, compounded by cultural changes hindering implementation of known technologies.
Tyler Cowen defends macroeconomics' progress in retrodiction and modeling but argues it lacks reliable predictive power for business cycles or policy optima, advocating stable rules like steady nominal GDP growth instead. He emphasizes underinvestment in culture-economics intersections, warns of globalization's "cashing in" on small cultural stocks leading to monoculture risks, and praises religion as vital cultural capital sustaining birth rates and self-replication. Cowen highlights Silicon Valley's success as a "diverse monoculture" enabled by optimal insulation, critiques UBI for political inviability, and sees early signs of escaping the Great Stagnation via AI and molecular tech integration.
Patrick Collison recounts hacking Ireland's education system via self-study to enter MIT early, prioritizing programming over formal schooling, and dropping out after recognizing physics progress had stalled and software offered greater fulfillment. Stripe's culture prioritizes rigor in thought, competitive willfulness, and interpersonal warmth amid scaling challenges like relentless problem influx and balancing exploration (20%) with exploitation (80%). Irish export-driven history instilled in him a pro-globalization ethos, viewing payments friction as a solvable barrier to vastly expanding e-commerce and firm creation worldwide.
Dozens of major infrastructure, engineering, and software projects from the 1920s-1960s were executed in 100-500 days, including the P-80 jet in 143 days, Empire State Building in 410 days, and Apollo 8 lunar mission decision-to-launch in 134 days. Post-1970 projects exhibit drastic slowdowns, such as San Francisco's Van Ness bus lane taking 7,600 days at $110k/meter versus the Alaska Highway's 234 days at $793/meter (2019 dollars). Hypotheses attribute this to 1960s shifts in bureaucracy, red tape, interest group influence, and "vetocracy," reducing institutional speed.
The economy is shifting to technologically enabled rails, with programmable money via platforms like Stripe acting as the OS for commerce, enabling complex marketplace models that extend from bits to atoms. Marketplaces address information chasms, trust deficits, and liquidity issues in non-tech sectors like pest control and childcare, fostering peer-to-peer efficiency over rigid hierarchies. This trend signals cash's decline, introduces geopolitical risks from centralized commerce flows, and unlocks unbounded economic growth through better coordination and feedback loops.
The internet creates persistent tensions with governments through 11 mechanisms, including accelerated event timelines, uncensorable communication, decentralized spotlighting of abuses, and jurisdictional ambiguities. These amplify government fragility via leak risks while enabling surveillance overreach and rendering many legacy laws obsolete. Legislators' weak technical understanding exacerbates ineffective policymaking, with no near-term resolution expected given the pace of technological change.
Stripe's founder doubts the company could have launched in Ireland due to banks' reluctance to partner with unproven startups, unlike the US. Broader challenges for Irish tech startups include a weak investment climate beyond seed stage, with top investors concentrated in the US; a shallow local talent pool lacking Silicon Valley's global immigrant draw; and minimal acquisitions, leading to outright failures rather than founder redeployment. Despite improvements like new accelerators, events, and offices from firms like Hubspot, Ireland lags most non-SV hubs but trails SV significantly.
This Monit configuration monitors a Unicorn Rails process via PID file, starting it with a specific config in production environment as deploy user. It triggers alerts on CPU >50% for 2 cycles, restarts on CPU >80% for 3 cycles, loadavg(5min) >10 for 8 cycles, or failed Unix socket connections with 20s timeout. Restarts timeout after 5 within 20 cycles, ensuring high availability.