Does Software Actually Slow Innovation?

Episode Summary

Executive Summary: The episode explores James Bessen’s thesis that slowing productivity growth is driven less by weak competition or M&A and more by dominant firms’ mastery of proprietary software, data, and organizational complexity. The hosts debate whether antitrust, unbundling, and data sharing can restore diffusion, innovation, and healthier competition.

Main Topics: Software, data, and productivity slowdown (Priority: 5/5): Bessen argues that firms investing heavily in proprietary IT and data management can dominate industries while slowing the growth of smaller productive firms, helping explain broader productivity stagnation. Complexity as a barrier to diffusion (Priority: 5/5): Large, software-intensive systems are difficult to imitate or spin off, limiting the spread of innovations that historically allowed new entrants to challenge incumbents. Policy responses: antitrust, unbundling, and diffusion (Priority: 5/5): The discussion centers on whether government should nudge or force firms to open up technologies and data, with IBM unbundling and AT&T transistor licensing cited as precedents. Data ownership, sharing, and industry specificity (Priority: 4/5): The hosts and guest agree data matters, but note that regulation must be tailored by sector because data is entangled with systems, relationships, and business models. Corporate secrecy, employee mobility, and legal barriers (Priority: 4/5): Non-competes, trade secret law, and aggressive litigation are portrayed as slowing knowledge diffusion and blocking startup formation. Broader societal consequences and regulation (Priority: 4/5): Software-enabled opacity can undermine regulation and contribute to problems like emissions cheating and financial instability, increasing inequality and concentration. Limits of simple market-fundamentalist assumptions (Priority: 3/5): The discussion challenges the idea that firm profit-maximization automatically benefits society, especially when innovation behaves like a public good.

Key Arguments: Dominant firms’ heavy investment in proprietary software and other intangibles is associated with slower growth among more productive firms in the same industry. The productivity slowdown is stark: labor productivity growth fell from 2.7% annually (2000-2007) to 1.4% thereafter, implying much slower doubling times. Software lowers the cost of managing complexity, but complex systems and data create barriers to imitation and diffusion. Historically, innovation spread through licensing, spin-offs, and unbundling; today non-competes and trade-secret restrictions reduce that diffusion. Government antitrust pressure can push firms to unbundle profitable technologies, as with IBM’s hardware/software separation and AT&T’s transistor licensing. Data and software are often intertwined with organization, making one-size-fits-all regulation ineffective; sector-specific antitrust is favored over blunt regulation. Corporate secrecy can distort markets and regulation, as shown by emissions cheating and opaque financial instruments before the 2008 crisis. Innovation resembles a public good more than a standard widget market, so pure profit-maximization does not reliably deliver socially optimal outcomes. Aggressive litigation and IP enforcement can deter startups and allow incumbents to freeze out challengers, even when claims are weak.

Data Points: Labor productivity growth: 2.7% to 1.4% per year - Bessen’s cited comparison of 2000-2007 versus the period since, used to illustrate the productivity slowdown. Productivity doubling time: 25 years vs. 50 years - At 2.7% growth, output doubles in about 25 years; at 1.4%, it takes about 50 years. Total investment in proprietary software: $239 billion - Proprietary software investment grew 74% over the decade ending in 2019. Growth in proprietary software investment: 74% - The decade-ending-2019 increase in total proprietary software investment. Own-developed software stock among top four firms: 8-fold increase - Bessen notes the top four firms’ stock of self-developed software rose much faster than the rest. Relative investment of top four firms vs. firms ranked 5th-8th: About 2x - Top-four firms’ software investment doubled relative to mean investments of second-tier firms.

Pivotal Quotes: "the degree to which the dominant firms in an industry invest in intangibles, and in particular, proprietary information technology, the rate of growth of productive firms in that same industry is slower" — James Bessen: Core thesis connecting dominant-firm software investment to slower challenger growth. "we have socialism for the very rich, rugged individualism for the poor" — Bethany McLean / Lucia Zingales (podcast framing): Opening critique of unequal capitalism and the podcast’s theme. "I don't think you should nudge it. You should force the hell out of them." — Lucia Zingales: Stronger policy response to IBM-style unbundling and antitrust intervention.

Implications: Listeners should expect more scrutiny of software, data, and IP as sources of market power. The episode argues for stronger, more tailored antitrust and freer diffusion of knowledge to protect innovation, competition, and regulation.