Lessons from James Bessen

Lessons from James Bessen

James Bessen is an economist and director of the Technology & Policy Research Initiative at Boston University, where he studies how innovation alters labor markets. His research argues that automation historically creates jobs through elastic demand, and that modern corporate dominance relies on proprietary software rather than actual efficiency. This compilation surveys his work across economics, AI, patent law, and corporate strategy to show how technology really affects the workforce.

Part 1: The Automation Paradox

1. On task replacement: "Automation rarely replaces entire jobs; it typically just automates specific tasks, making the remaining human-performed tasks more valuable." — Source: ILO Live
2. On the elevator operator exception: "Historical data suggests only a very small fraction of occupations disappear primarily due to automation, such as the elevator operator." — Source: Time Magazine
3. On the nature of human work: "Human jobs are complex combinations of different activities, making complete substitution by machines incredibly rare." — Source: Boston University TPRI
4. On technology and job creation: "While observers focus on a narrative about imminent mass unemployment, the reality is that new technology regularly creates new demand and new occupations." — Source: Brookings Institution
5. On the persistence of employment: "Studies of firms that invested in automation between 2000 and 2016 show no evidence of mass job loss; in many cases, total employment at these firms grew." — Source: ITIF
6. On computerization and growth: "Occupations that utilize computers tend to grow faster than those that do not, as automation changes the nature of the work rather than eliminating it." — Source: Boston University TPRI
7. On shifting focus: "The primary challenge posed by automation is the major reallocation of jobs across different sectors and skill sets, rather than a permanent loss of work." — Source: EconTalk
8. On the transition period: "Automation requires workers to make disruptive transitions to new industries, requiring them to acquire entirely new skills." — Source: Boston University
9. On the true cost of automation: "The societal cost of technological change is felt by the displaced worker who must change careers, rather than by a permanent reduction in the total number of jobs available." — Source: NBER

Part 2: Artificial Intelligence and Employment

1. On demand elasticity: "The effect of AI on employment depends critically on the elasticity of demand; if technology lowers costs and consumers buy much more, employment grows." — Source: The Economics of Artificial Intelligence
2. On historical precedents for AI: "Just as power looms increased the number of weavers by lowering the cost of cloth, AI can increase employment in certain sectors by making services more affordable." — Source: NBER Working Paper
3. On saturated markets: "When markets become saturated and demand becomes inelastic, further productivity gains from AI will likely lead to job losses in those specific industries." — Source: EconTalk
4. On the limits of predictive models: "Alarmist predictions about AI often fail because they assume consumer demand is static and ignore how price drops spur new consumption." — Source: Time Magazine
5. On AI capital expenditure: In Fixed + Floating, Bessen frames AI capital spending as a potential source of incumbent advantage and credit risk, noting that AI capex may be more fragile than earlier infrastructure cycles. — Reference: Fixed + Floating episode on software moats and AI capex risk with James Bessen
6. On AI and organizational structure: "Implementing artificial intelligence reshapes how firms are organized, altering reporting lines and the specific skills demanded of mid-level workers." — Source: OECD Events
7. On AI as a task-automator: "Artificial intelligence is currently best understood as a tool that automates specific cognitive tasks, rather than a drop-in replacement for a human employee." — Source: ILO Live
8. On the slow rollout of AI: "Despite the hype, deploying AI effectively takes time because it requires reorganizing business processes to accommodate the new technology." — Source: Munk Debates
9. On AI and market concentration: "The data-intensive nature of artificial intelligence means it can exacerbate market concentration by disproportionately benefiting firms that already possess massive datasets." — Source: Boston University

Part 3: Learning by Doing

1. On the invention fallacy: "Invention is only the first step; implementing new technology is a complex process that often takes decades of trial-and-error experimentation." — Source: Learning by Doing
2. On unstandardized knowledge: "The technical knowledge required for new technologies is often unstandardized and specific to individual firms, making it hard to teach in classrooms." — Source: Boston University TPRI
3. On on-the-job experience: "Because new technical skills cannot be easily taught formally, workers must acquire them through direct, hands-on experience on the factory or office floor." — Source: EconTalk
4. On the implementation lag: "The economic benefits of a new technology are rarely felt immediately because it takes time for both workers and management to figure out how to use it effectively." — Source: TechRatchet
5. On the Industrial Revolution's lessons: "During the early Industrial Revolution, wages for weavers remained flat for decades until the operation of the power loom became a standardized, teachable skill." — Source: Higher Ed Strategy
6. On the cost of learning: "Acquiring unstandardized knowledge requires significant investment from workers, who must often accept lower wages during their training period." — Source: Learning by Doing
7. On the diffusion of skills: "Only when a technology matures and its required skills become standardized do training programs and schools begin to effectively teach it." — Source: EconTalk
8. On the complexity of modern machines: "Modern software and IT systems are so complex that knowing how to operate them is often inseparable from understanding the specific business processes of the employer." — Source: Boston University
9. On continuous adaptation: "Because technology changes constantly, the process of learning by doing is never truly finished; workers must continuously adapt to incremental improvements." — Source: Learning by Doing
10. On institutional support: "Policies often fail to support the actual mechanism of technological adaptation, which is the messy, on-the-ground learning that happens within firms." — Source: TechRatchet

Part 4: The Wage Stagnation Problem

1. On the modern paradox: "For the past three decades, during the rise of the personal computer, we have experienced rapid innovation while the wages for the median worker have largely stagnated." — Source: Timothy B. Lee Interview
2. On who captures the value: "During periods of rapid technological change, profits accrue primarily to business owners and investors rather than workers, whose bargaining power is diminished." — Source: Learning by Doing
3. On the inequality of computerization: "The shift toward computerized work is associated with greater within-occupation wage inequality, as those who master the new systems pull away from those who do not." — Source: Boston University TPRI
4. On the missing economic fruits: "There is a widespread sense that we are generating immense technological progress, yet the average worker is not seeing the economic fruits of that innovation." — Source: Timothy B. Lee Interview
5. On bargaining power: "Until a technology and its associated skills become standardized across an industry, workers lack the mobility to demand higher wages from competing employers." — Source: EconTalk
6. On the skills gap: "The skills gap is largely a symptom of technology moving faster than our ability to standardize the knowledge required to operate it." — Source: Learning by Doing
7. On employer training incentives: "Because workers might leave after acquiring valuable skills, employers often underinvest in training, exacerbating wage stagnation." — Source: Brookings Institution
8. On the historical parallel: "Just as it took decades for textile workers to see wage gains in the 19th century, today's IT workers face a prolonged period where the technology outpaces institutional adaptation." — Source: Higher Ed Strategy
9. On structural inequality: "Rising economic inequality is a direct result of how slowly we adapt our training and labor institutions, rather than an inevitable law of technological progress." — Source: Learning by Doing

Part 5: Software as a Corporate Moat

1. On the end of creative destruction: "The great IT revolution is no longer promoting economic dynamism; by enabling dominant firms to entrench themselves, it is actively preventing it." — Source: Boston University
2. On proprietary systems: "Across all major sectors of the economy, proprietary information technology is increasing the market dominance of large firms." — Source: The New Goliaths
3. On the true source of dominance: "The strategic use of internal, proprietary software explains modern revenue and productivity gains far more than mergers and acquisitions." — Source: Goodreads
4. On the tilted playing field: "Large corporations use complex software systems to manage logistics, pricing, and marketing at a scale that creates a tilted playing field against smaller rivals." — Source: The New Goliaths
5. On the slowdown of knowledge diffusion: "The fact that only the largest firms can afford and manage these IT systems is evidence of a slowdown in the spread of technical knowledge throughout the economy." — Source: Boston University TPRI
6. On building defenses: In the Fixed + Floating episode, Bessen argues that proprietary software becomes a moat when scale, data, and workflow complexity reinforce one another; the episode also frames technology spending as business-model defense, not just capex. — Reference: Fixed + Floating episode on proprietary software, scale, data, and workflow complexity
7. On the complexity barrier: In Value Investing with Legends, Bessen discusses how proprietary software reshapes competitive dynamics and industry stability, emphasizing that advantage comes from the surrounding operating system and mass-customization capability, not code alone. — Reference: Value Investing with Legends episode on proprietary software and competitive dynamics with James Bessen
8. On the difference from the past: "Unlike the standardized technologies of the mid-20th century, today's competitive advantages are locked inside custom-built codebases that never leave the firm." — Source: Capitalisn't Podcast
9. On evading regulation: "Superstar firms use their proprietary IT systems to outcompete rivals and effectively evade government regulation that depends on standardized reporting." — Source: Society for Computers and Law
10. On the illusion of tech competition: "While the tech sector itself seems highly competitive, the application of tech in traditional industries has severely concentrated market power." — Source: The New Goliaths

Part 6: The Failure of the Patent System

1. On the purpose of patents: "For a patent system to function effectively as a property rights system, it must provide clear, predictable boundaries that inform others exactly what is owned." — Source: Patent Failure
2. On fuzzy boundaries: "The U.S. patent system suffers from fuzzy boundaries, making it nearly impossible for innovators to determine if their new product infringes on an existing patent." — Source: Academy of Management
3. On the cost-benefit deficit: "With the notable exception of the pharmaceutical and chemical sectors, the costs of the patent system outweigh the benefits of patent protection for most industries." — Source: Patent Failure
4. On the litigation explosion: "The significant rise in patent litigation is largely attributable to poor notice and ambiguous patent claims, rather than simply the rise of patent trolls." — Source: Texas A&M University
5. On stifling innovation: "By creating unpredictable legal risks and hidden liabilities, the patent system often discourages investment in R&D rather than promoting it." — Source: George Mason University
6. On the software problem: "The problem of boundary clarity is especially poor in software, where abstract claims make it impossible to map a patent to a specific, tangible invention." — Source: Electronic Frontier Foundation
7. On property rights vs. patents: "Unlike real estate, where property lines are clearly drawn and recorded, patent boundaries are often only defined after millions of dollars are spent in court." — Source: First Monday
8. On the tax on innovation: "For public software companies, the cost of patent litigation functions as a massive tax on innovation, diverting resources away from actual product development." — Source: Patent Failure
9. On institutional failure: "The issues with the patent system are more than bad actors abusing the rules; they are fundamental institutional failures in how the patent office and courts grant and interpret claims." — Source: Econlib
10. On the need for reform: "The patent system is not broken beyond repair, but it requires significant legal reforms focused on improving the notice function to make patents act like reliable property rights." — Source: Patent Failure

Part 7: The Decline of Economic Dynamism

1. On the slowdown of startups: "The dominance of superstar firms utilizing proprietary software has directly contributed to the declining rate of new business formation in the United States." — Source: The New Goliaths
2. On the productivity gap: "There is a growing chasm between the most productive firms in an industry, which use advanced IT, and the rest of the firms, which cannot afford or implement it." — Source: OECD Events
3. On the barrier to entry: "Information technology, which was once viewed as a democratizing force for small business, has evolved into a formidable barrier to entry requiring massive fixed costs." — Source: The Economics Review
4. On worker mobility: "When a few large firms dominate an industry, workers have fewer options, leading to decreased labor mobility and reduced wage bargaining power." — Source: Boston University TPRI
5. On the stagnation of diffusion: "Economic growth relies on the diffusion of best practices; when top firms hoard their software and data, that diffusion stalls, dragging down the aggregate economy." — Source: The New Goliaths
6. On the M&A illusion: In Value Investing with Legends, Bessen points to proprietary software as a driver of industry stability and market concentration, shifting the explanation for declining disruption beyond mergers alone. — Reference: Value Investing with Legends episode on proprietary software, industry stability, and market concentration
7. On the changing nature of competition: "Firms today compete on the sophistication of their internal data processing capabilities, moving beyond product quality or price." — Source: The New Goliaths
8. On the resilience of incumbents: In Fixed + Floating, Bessen describes dominant firms as harder to disrupt when proprietary software, scale, data, and workflow complexity reinforce one another, creating structural advantages for incumbents. — Reference: Fixed + Floating episode on dominant firms, proprietary software, and structural incumbent advantages
9. On the broader economic drag: "When startups cannot disrupt incumbents, the entire economy suffers from slower overall innovation and less efficient capital allocation." — Source: Capitalisn't Podcast

Part 8: Reallocation and Policy Reform

1. On managing transitions: "Because the transition to new technologies is highly disruptive for individual workers, policy should focus on supporting skill development and managing industry shifts." — Source: Brookings Institution
2. On open access: "To restore competitive balance, policymakers should encourage or compel dominant firms to share their fundamental technology, data, and technical knowledge." — Source: The New Goliaths
3. On breaking up monopolies: "Simply breaking up large technology companies may not solve the underlying issue if the proprietary software systems that drive dominance remain closed." — Source: Boston University TPRI
4. On reforming education: "Our educational institutions must recognize that they cannot teach unstandardized knowledge; they must partner with employers to facilitate on-the-job learning." — Source: Learning by Doing
5. On employee non-competes: "Legal mechanisms like non-compete agreements actively prevent the diffusion of technical knowledge and should be severely restricted to boost dynamism." — Source: Boston University
6. On unemployment insurance: "We need a strong social safety net because the jobs of the future will require workers to take significant risks to retrain." — Source: EconTalk
7. On standardizing technology: "Government policy can help accelerate the standardization of new technologies, which shortens the period of wage stagnation and helps workers." — Source: Learning by Doing
8. On data portability: "Mandating data portability and interoperability between software systems is an important step in lowering the barrier to entry for new competitors." — Source: The New Goliaths
9. On a realistic future: "If we focus our policies on managing reallocation rather than bracing for mass unemployment, we can ensure the economic benefits of innovation are widely distributed." — Source: ITIF