Dwarkesh PodcastJason Crawford - The Roots of Progress & the History of Technology
CHAPTERS
- 0:00 – 1:58
Jason Crawford’s mission: explaining modern living standards through technology history
Dwarkesh introduces Jason Crawford and his project The Roots of Progress. Jason lays out the core puzzle: why living standards were flat for millennia and then surged over the last few centuries, and what we can learn to sustain and accelerate that trajectory.
- •The Roots of Progress: history of technology plus a philosophy of progress
- •Modern life improvement is historically unprecedented
- •Focus on root causes and key inventions/discoveries
- •Concern about maintaining progress and preventing reversal
- 1:58 – 4:10
Is progress slowing? Why “progress isn’t automatic” creates urgency
Dwarkesh reframes the motivation for progress studies around potential stagnation in the last ~50 years (e.g., travel speeds). Jason agrees slowdown is plausible, but emphasizes the deeper issue: progress can slow or speed depending on controllable levers.
- •Stagnation hypothesis adds urgency to study progress mechanisms
- •Progress isn’t inevitable; it can slow due to friction
- •What matters most: future range of outcomes and controllable levers
- •Past slowdown matters mainly insofar as it informs future action
- 4:10 – 7:31
Progress begets progress—until S-curves saturate
Jason explains how progress can decelerate despite compounding effects. Individual technologies follow S-curves—slow start, rapid takeoff, then diminishing returns—so continued acceleration requires overlapping waves of new domains.
- •Even “slower” modern progress far exceeds pre-industrial rates
- •Technologies saturate as they mature (electricity example)
- •Long-run growth comes from overlapping multiple S-curves
- •Key question: how quickly we discover and ramp new S-curves
- 7:31 – 11:11
Hypotheses for slowdown: talent allocation, cultural ambivalence, and bureaucratic friction
Jason offers speculative explanations for why progress might have felt underwhelming recently. He points to cultural shifts post-WWII (atomic fear, environmentalism, oil shocks), possible changes in talent incentives, and growing regulatory/corporate/university bureaucracy.
- •Postwar events may have increased anti-tech/anti-progress sentiment
- •Potential effect: fewer people motivated to enter frontier-building fields
- •Regulatory and institutional “creeping bureaucracy” adds overhead
- •Safety rules are rational, but cumulative costs may be underweighed
- 11:11 – 13:48
Learning from intellectual movements: how ideas reshape institutions over decades
Dwarkesh asks whether progress studies can copy tactics from successful academic movements like critical theory. Jason stresses that big philosophical ideas diffuse slowly but ultimately shape culture, media, and government—so the project must be a long game.
- •Movements in ideas can have major downstream institutional effects
- •Diffusion from academia to society often takes decades
- •Worth studying how prior movements formed and spread
- •Progress studies should be patient and strategically oriented
- 13:48 – 17:48
Progress Studies as a ‘school of thought,’ not a new academic silo
Jason argues progress studies is best seen as a set of premises and values that guide what questions researchers ask, rather than a replacement discipline. He notes related work already exists across economic history and innovation economics, but lacks a unifying banner.
- •No obvious existing academic paradigm to “piggyback” on
- •Many adjacent contributors already: Mokyr, McCloskey, Azoulay, Arora, etc.
- •Progress studies = premises/values about progress being real, vital, non-automatic
- •Goal: unify and coordinate existing scattered lines of inquiry
- 17:48 – 20:49
Progress Studies for Young Scholars: teaching technology history to high schoolers
Jason describes an online program built with Higher Ground Education/ATI to teach the history of technology. He argues students often take industrial civilization for granted, yet civic debates require basic literacy in how living standards are produced.
- •Summer course expanded into ongoing after-school/homeschool offering
- •Technology and progress history is neglected in standard curricula
- •Civic rationale: policy debates hinge on understanding industrial systems
- •Everyday examples (refrigeration, transport, elevators) to make progress tangible
- 20:49 – 23:47
Why innovation feels understudied: appreciation drives research attention
Dwarkesh presses on why such a foundational phenomenon could be underexplored. Jason replies that even “progress about progress” isn’t automatic; fields form only when people value them, and cultural ambivalence about progress may steer attention away from its benefits.
- •Creating a field of study requires deliberate effort and prioritization
- •There is existing history-of-technology work, but not enough focus
- •Cowen/Collison’s framing: more attention should be paid to existing work
- •Cultural ambivalence can reduce both frontier-building and scholarship
- 23:47 – 25:37
Scientific, technological, and moral progress: intertwined revolutions
Jason addresses how moral/political change relates to scientific and technological advances. He suggests it’s not coincidental that scientific, industrial, and democratic-republican transformations clustered historically, with institutional shifts (e.g., weakening guilds) enabling innovation.
- •Science/technology/moral-political progress are inseparable in practice
- •Three major revolutions occurred in related timeframes
- •Institutional liberalization (e.g., dissolving guild constraints) mattered
- •Causality likely runs in multiple directions among the three domains
- 25:37 – 26:37
Have political institutions progressed? Clear wins, mixed overall trajectory
Dwarkesh asks whether U.S. governance has improved, citing pandemic response. Jason says political/moral progress is the hardest to evaluate: there are major improvements (civil rights), but trajectories are non-monotonic and messier than tech progress.
- •Moral/social/government progress is less measurable and more contested
- •Some clear advances: civil rights compared to the 1950s
- •Not progress on every dimension; backsliding occurs
- •Institutional performance (e.g., crisis handling) remains debatable
- 26:37 – 34:54
The linear model of innovation: useful, but reality is cyclical
Jason critiques the simplistic “basic science → applied engineering → products” story. He argues science is foundational and unpredictable in its applications, yet invention also precedes and drives theory, and measurement technologies often enable scientific breakthroughs.
- •Science underpins key industries (electronics, medicine, metallurgy)
- •Applications can’t be foreseen; curiosity-driven science is justified
- •Invention can outpace explanation (steam engine before thermodynamics)
- •Transistor case: rapid back-and-forth between experiments and theory
- 34:54 – 36:03
Why still separate science and technology? Distinct concepts for focused study
Dwarkesh challenges the value of distinguishing science from technology given their tight coupling. Jason argues they remain meaningfully distinct categories, and separation enables specialization—so long as researchers don’t mistake the relationship as simple or linear.
- •Science and technology are distinct even if inseparable
- •Analytical separation helps specialized study (like physics vs chemistry)
- •The key is acknowledging complex, non-linear connections
- •Avoid reifying the linear model while keeping useful distinctions
- 36:03 – 41:29
The risks of progress: interdependence, black swans, and building resilience
Dwarkesh raises the vulnerability created by specialization and interdependence (e.g., pandemics). Jason agrees risks are real but argues the remedy is more applied intelligence—statistics, systems analysis, slack/buffers—rather than broadly “slowing down.”
- •Humans have always been interdependent; vulnerability isn’t new
- •Modern complex systems need engineered resilience and redundancy
- •Use statistics and systems thinking to anticipate shock distributions
- •Reframe: expand the ‘progress portfolio’ to include safety and robustness
- 41:29 – 48:38
Existential threats, prediction limits, and ‘prescriptive optimism’ as a stance
Dwarkesh cites Tyler Cowen’s worry that destructive capabilities may outpace defenses. Jason rejects fatalism and emphasizes that long-run predictions are mainly useful insofar as they spur action; he distinguishes descriptive optimism from prescriptive optimism—commitment to fight for better outcomes despite uncertainty.
- •Acknowledges asymmetric risks (nukes, engineered pandemics) but urges action
- •Skepticism about meaningful 800-year forecasts; focus on today’s levers
- •Two optimisms: descriptive (what will happen) vs prescriptive (what we’ll do)
- •Prescriptive optimism supports ambition when trends are good and resolve when they’re bad
