Chronological feed of everything captured from Upstream with Erik Torenberg.
Orchid Health has commercialized the world's first whole genome sequencing for embryos, allowing for 99% genomic data analysis, a significant improvement over previous methods. This technology enables prospective parents to screen for thousands of monogenic and polygenic diseases, empowering informed reproductive choices and potentially reducing the incidence of severe conditions. The increased data resolution and accessibility to this technology through IVF centers nationwide mark a substantial advancement in reproductive medicine.
Guillaume Verdon, known as Beth, founder of effective accelerationism (e/acc), was recently doxxed, revealing his identity as the founder of X-Tropic, a stealth AI hardware startup. This incident, perceived as a warning shot against e/acc's growing influence, forced Verdon to publicly connect his pseudonymous and professional identities. He leveraged the situation to articulate e/acc's core tenets: freedom of information and compute, bottom-up self-adaptation, and decentralization as a hedge against the over-centralization of AI by incumbents.
The concept of "Live Players" describes individuals who challenge conventions and drive innovation. This includes figures like Elon Musk, Peter Thiel, and Sam Altman, who each exhibit distinct approaches to navigating mainstream thought and pursuing ambitious goals. Thiel identifies flaws in consensus, Musk follows first-principles thinking, and Altman harnesses enthusiasm for viable trends. These individuals are crucial "stem cells" for societal evolution, preventing stagnation by constantly re-evaluating and pushing boundaries, even when their methods or ideologies differ.
The OpenAI crisis illustrates that human trust and talent density supersede formal governance in the AI era. Strategic dominance in AI is shifting from pure research toward distribution and deployment, favoring incumbents like Microsoft and Meta who possess vast proprietary data and existing corporate integration. This suggests that the 'deployment phase' of AI favors the 'slow takeoff' model of augmenting existing productivity over sudden, disruptive institutional replacement.
The stagnation of nuclear energy resulted from a 'witches brew' of over-optimistic safety claims by the AEC, a shift in electricity demand following the 1973 oil embargo, and the imposition of non-engineering safety metrics like ALARA. This environment was exacerbated by an anti-growth environmental movement and regulatory capture, transforming nuclear plants into high-risk, cost-prohibitive 'utility killers' regardless of their actual safety records.
The venture capital landscape is undergoing a fundamental shift away from the "factory line" model of predictable unicorn creation. This era, characterized by an overabundance of capital and a focus on packaging companies for quick exits, is ending. The future of venture will favor artisanal, non-consensus investing in businesses with real value creation, moving away from mass-produced, self-similar ventures.
Burn Hobart argues the conventional framing—that AI displaces junior workers—inverts the actual threat vector: LLMs excel at the senior-layer functions of decomposing ambiguous projects into scoped tasks and providing contextual guidance, not just executing rote work. This creates a structural shift where permissionless fields (finance, software) will see solo operators or micro-teams wielding LLM-as-mentor to bootstrap companies without venture capital or institutional mentorship. On macro topics, Hobart reads US-China trade tensions as likely temporary bluffing with asymmetric cost timing, and frames China's demographic/real-estate crisis as manageable via state-directed bank recapitalization—more "Japan's lost decade with Chinese characteristics" than acute financial collapse.
New York Times columnist Ross Douthat argues that the new atheist movement's cultural decline stems from two compounding failures: its critique of religion turned out to be a critique of human nature broadly (dogmatism, tribalism, moral panic persist in secular societies), and it offered no compelling "cosmic optimism" once secular-liberal progress narratives collapsed. Douthat contends that the positive case for religion is stronger than commonly acknowledged — grounded in the fine-tuning argument, the hard problem of consciousness, quantum indeterminacy, and the persistence of mystical experience — and that multiverse theories invoked to rebut fine-tuning are themselves unfalsifiable metaphysical speculation no more parsimonious than theism. His new book targets the "curious almost-believer": spiritually interested people who assume religion requires abandoning reason, and aims to show that a basic religious worldview is not just socially useful but probably true.
Lynn Alden argues that monetary history is best understood through technological change rather than political decisions — the telegraph centralized financial power by enabling light-speed ledger updates, and Bitcoin/stablecoins are now reversing that centralization by enabling peer-to-peer settlement outside regulated banking systems. The U.S. reserve currency status structurally exports dollars via trade deficits, hollowing out domestic manufacturing and creating a negative net international investment position (~-70% of GDP) that is mathematically unsustainable. Critically, at 130%+ debt-to-GDP, raising interest rates no longer suppresses inflation — it worsens fiscal deficits faster than it reduces bank lending, making real assets and scarce stores of value (Bitcoin, energy, real estate) the rational hedge. Governments that resist crypto adoption are fighting a losing battle, as demonstrated by Nigeria and Argentina's failed bans.
Burn Hobart argues that Trump's tariff regime represents the high-water mark and simultaneous collapse of "state capacity Trumpism" — methodically planned on immigration and higher ed, but chaotic and improvisational on trade. Treasury bond behavior is now a proxy bet on the future of U.S. separation of powers: yields rise if the executive wins, fall if courts or Congress force a reversal via recession. Simultaneously, AI is poised to inflict on white-collar workers the same structural employment hollowing that manufacturing automation caused in the 20th century, with companies locking in AI capital expenditure now and permanently reducing future headcount needs. The U.S. dollar's reserve status has a self-healing mechanism — crises that threaten it tend to trigger dollar-denominated debt scrambles that reinforce it — making meaningful de-dollarization within a single presidential term structurally implausible.
Helen Toner, former OpenAI board member and CSET director, argues that AI policy is mis-framed around nonproliferation — a strategy that collapses as capability costs fall rapidly — and should instead prioritize building societal "adaptation buffers": resilience infrastructure like outbreak detection, wastewater monitoring, and formal software verification that buys time regardless of who holds the frontier. She draws a sharp contrast between compute-threshold-based restrictions (which require increasingly invasive enforcement as models commoditize) and conditional, capability-conditional slowdowns tied to measurable safety milestones. On the regulatory side, she finds no comprehensive framework she endorses, instead favoring targeted building blocks: transparency requirements, whistleblower protections tied to mandatory disclosure regimes, and expanded technical capacity inside government.
Palantir CEO Alex Karp argues that internal and external cultural honesty — refusing to adopt performative ideological positions — is the core mechanism behind Palantir's ability to attract elite engineering talent and build disruptive products. Karp frames elite university dysfunction and campus antisemitism not as a societal rot but as an institutional one, driven by what he calls a "thin religion" that produces cognitive contradictions and repels serious talent. He extends this logic to the US-China tech competition, asserting that America's structural advantages — meritocracy, top-tier immigration, and tolerance for unconventional builders — explain why no Chinese or Russian firm competes meaningfully with Palantir, and that squandering those advantages through ideological capture is the primary existential risk.
At the 2025 Hill and Valley Forum, Applied Intuition CEO Cassér and Lux Capital's Josh Wolfe argue that America's primary edge in AI and autonomous systems is immigrant talent — and that overly broad immigration and espionage restrictions risk eroding that edge more than China's technology theft does. China's lack of IP culture means its companies share openly, forcing the conclusion that the real competitive imperative is relentless product superiority, not information hoarding. On the battlefield, autonomy — where AI meets the warfighter — is under-adopted across DoD services, and the fix isn't better procurement processes but senior defense officials who have actually built AI systems themselves.
Historian Samo Burja argues that both the Roman Empire and Song Dynasty China underwent genuine industrial revolutions — characterized by water-powered mechanization, mass production, standardization, and large-scale commerce — but these revolutions plateaued as S-curves rather than compounding into sustained exponential growth. The key limiting factor was not technological ceiling but demographic and geographic saturation: once population growth stalled and territorial expansion ceased, there was no economic pressure to push to the next level of energy or mechanical innovation. This reframes the standard narrative of a single, exceptional Industrial Revolution in 18th-century England as the tallest instance of a recurring historical pattern, not a unique civilizational breakthrough.
Samo Burja argues that the Bronze Age collapse (~1200 BC) is the clearest historical proof that technological and civilizational regression is not only possible but can be total and permanent within a generation — entire writing systems, metallurgical knowledge, and state structures vanished without trace. Crucially, he extends this lesson beyond centralized empires: even decentralized, trade-networked systems (like the Bronze Age's tin-copper trade web) are fragile if their interdependencies break. Applied to the present, Burja maps a geopolitical trajectory where US hegemony is eroding, Russia is grinding toward marginal territorial gains in Ukraine, and Europe's best path is a Swiss-style federal model — flexible, locally autonomous, and less dependent on Brussels' veto-cratic structure. He further contends that politically inconvenient archaeological findings — from pre-Columbian North American civilizations to the true scale of Aztec human sacrifice — are systematically suppressed or underreported, distorting both public history and policy intuitions.