Does Technology's Good Outweigh Its Bad?

The Question

My framing: fire and the atomic bomb as archetypal dual-use technologies, with a specific interest in the argument that offensive capability spikes first but defensive capability eventually catches up.

Offense-Defense Balance Theory (The Named Theory)

This is the closest match to "offensive capability spikes first, defense catches up." It originates in international relations theory and has been extended to technology broadly.

Robert Jervis (1978): The Foundational Paper

• Title: "Cooperation Under the Security Dilemma"
• Published: World Politics, Vol. 30, No. 2, January 1978, pp. 167-214
• Key argument: Two variables shape conflict: the offense-defense balance and whether offensive/defensive postures are distinguishable. When defense has the advantage and postures are distinguishable, the security dilemma is ameliorated. When offense dominates and postures are indistinguishable, conflict spirals. Technology shifts this balance over time.
• Why it matters: Jervis's framework is the origin of the idea that the offense-defense ratio determines stability. New technologies initially destabilize by creating offensive advantage. Defensive adaptation restores equilibrium.
• URL: cambridge.org

Jack S. Levy (1984): Historical Analysis

• Title: "The Offensive/Defensive Balance of Military Technology: A Theoretical and Historical Analysis"
• Published: International Studies Quarterly, Vol. 28, No. 2, June 1984, pp. 219-238
• Key argument: Traces the offense-defense balance across eight centuries of military technology. Finds the concept is ambiguous, with multiple definitions that are not interchangeable, particularly between pre-nuclear and nuclear eras. Concludes that while the pattern of offensive spikes followed by defensive adaptation appears repeatedly, measuring the balance precisely is methodologically fraught.
• URL: academic.oup.com

James Fearon (1995): Quantitative Extension

• Title: "The Offense-Defense Balance and War Since 1648"
• Published: Stanford University working paper (presented at APSA)
• Key argument: Tests whether shifts in the offense-defense balance correlate with war initiation since the Peace of Westphalia. Finds the relationship is real but conditional on how "balance" is defined.
• URL: web.stanford.edu (PDF)

Sean Lynn-Jones (1995): "Grasping the Technological Peace"

• Title: "Offense-Defense Theory and Its Critics"
• Published: Security Studies, Vol. 4, No. 4
• Key argument: Defends offense-defense theory against critics by arguing that when defense-dominant technologies prevail, peace is more stable. When offense-dominant technologies emerge (like pre-WWI mobilization advantages), war becomes more likely. The pattern: new offensive technologies create instability, then defensive technologies and doctrines catch up.

Steven Pinker: Technology and the Decline of Violence

The Better Angels of Our Nature (2011)

• Author: Steven Pinker (Harvard)
• Published: Viking, 2011
• Key argument: Violence has declined dramatically across every measurable dimension, from war deaths per capita to homicide rates to treatment of children. Technology plays a dual role: it periodically creates new forms of violence, but in each case humanity succeeds in reducing them. The printing press spread empathy through literacy. Commerce made people more valuable alive than dead. Communication technologies expanded the "circle of empathy."
• Relevant to the question: Pinker explicitly addresses the pattern of technology-as-threat followed by adaptation-to-net-benefit. His thesis: despite nuclear weapons, bioweapons, and other offensive technologies, the long-term trend bends toward reduced violence.
• URL: stevenpinker.com

Enlightenment Now: The Case for Reason, Science, Humanism, and Progress (2018)

• Author: Steven Pinker
• Published: Viking, 2018
• Key argument: Extends the Better Angels thesis to 15 dimensions of human welfare: life expectancy, health, sustenance, wealth, inequality, environment, peace, safety, terrorism, democracy, equal rights, knowledge, quality of life, and happiness. Nearly all have improved, often dramatically, driven by Enlightenment values and technological progress. Pinker's core claim: the world is getting better on almost every measurable dimension, and technology is a primary driver.
• Why it matters for the question: This is the strongest mainstream academic case for "technology is net-positive over time." Published in 2018, it fits the "7-8 years ago" timeframe I was trying to place.
• URL: stevenpinker.com

The Counterarguments: Existential Risk Literature

Nick Bostrom (2019): The Vulnerable World Hypothesis

• Title: "The Vulnerable World Hypothesis"
• Published: Global Policy, Vol. 10, No. 4, 2019
• Key argument: Imagines technology development as drawing balls from an urn. Most balls are white (beneficial) or gray (mixed). But the urn may contain "black balls," technologies that by default destroy the civilization that invents them. The question is not whether benefits outweigh harms on average but whether a single catastrophic technology could end everything. Bostrom proposes the Principle of Differential Technological Development: ideally, protective technologies should arrive before the destructive technologies they protect against.
• Why it matters: This is the strongest formal argument against naive techno-optimism. Bostrom does not argue that technology is net-negative. He argues that the distribution has fat tails, and one bad draw could be terminal.
• DOI: doi.org/10.1111/1758-5899.12718
• Direct PDF: nickbostrom.com/papers/vulnerable.pdf

Toby Ord (2020): The Precipice

• Title: The Precipice: Existential Risk and the Future of Humanity
• Author: Toby Ord (Oxford, Future of Humanity Institute)
• Published: Bloomsbury, 2020
• Key argument: Humanity faces a one-in-six chance of existential catastrophe in the next 100 years. The primary risks are anthropogenic: engineered pandemics, unaligned AI, and nuclear war. Ord's key insight: "We're capable of growing our power exponentially on a lot of metrics. But there aren't many metrics where we'd say humanity's wisdom, or institutional ability has been growing exponentially." The gap between capability and wisdom is the danger.
• Why it matters: Directly addresses the "does defense catch up?" question. Ord's answer: not automatically. It requires deliberate institutional investment.

Max Tegmark (2017): Life 3.0

• Title: Life 3.0: Being Human in the Age of Artificial Intelligence
• Author: Max Tegmark (MIT)
• Published: Knopf, 2017
• Key argument: AI risk comes not from malevolence but from misalignment, "competent, indifferent systems optimizing for the wrong thing." Tegmark frames AI development as humanity's most consequential decision. Technology is neither inherently good nor bad. The outcome depends entirely on governance and alignment.
• Why it matters: Provides the nuanced middle ground. Technology is net-positive only if we solve the alignment problem.

AI and Cybersecurity: Offense vs. Defense

Andrew Lohn (2025): Georgetown CSET

• Title: "Anticipating AI's Impact on the Cyber Offense-Defense Balance"
• Published: Georgetown Center for Security and Emerging Technology (CSET), Policy Brief, May 2025
• Key findings: AI will help both attackers and defenders. There is no single answer to whether AI favors offense or defense. The cyber domain is too multifaceted. Critical insight: "Small increases in the number or quality of defenses can compensate for exponential increases in the number of independent attacks and for exponential speedups." This directly supports the "defense catches up" thesis in cyber.
• URL: cset.georgetown.edu

Jason Healey, Divyam Nandrajog, and Robert Jervis (2025): CSET

• Title: "The Impact of AI on the Cyber Offense-Defense Balance and the Character of Cyber Conflict"
• Published: Georgetown CSET / arXiv, April 2025
• Key findings: Cataloged 44 propositions for how AI impacts cyber offense-defense balance plus 48 additional propositions about the character of cyber conflict. No simple answer. AI changes the landscape in predictable ways that can be steered. Jervis (who initiated offense-defense theory in 1978) co-authored this before his death.
• URL: cset.georgetown.edu
• arXiv: arxiv.org/abs/2504.13371

UC Berkeley (2025): Frontier AI Cybersecurity Impact

• Title: "Frontier AI's Impact on the Cybersecurity Landscape"
• Published: UC Berkeley Center for Research on Digital Innovation
• Key finding: Used a marginal risk assessment framework to catalog frontier-AI-leveraged attacks and defenses across the entire attack and defense lifecycle. Drew from hundreds of academic papers and real-world incident reports.
• URL: rdi.berkeley.edu

Brandon Valeriano (2022): Cyber Offense-Defense Failure

• Title: "The Failure of Offense/Defense Balance in Cyber Security"
• Published: Cyber Defense Review, Vol. 7, No. 3, Summer 2022
• Key argument: The traditional offense-defense balance framework fails to translate cleanly to cyber because cyber operations lack the physicality that makes the theory work in conventional warfare. The offense-defense distinction is less useful in cyberspace than in kinetic domains.
• URL: cyberdefensereview.army.mil (PDF)

The Collingridge Dilemma: Why Regulation Lags

David Collingridge (1980)

• Title: The Social Control of Technology
• Author: David Collingridge (University of Aston)
• Published: Frances Pinter / St. Martin's Press, 1980
• Key argument: Technology governance faces a double bind. When a technology is young, we can control it but cannot foresee its impacts. When impacts become clear, the technology is entrenched and control becomes expensive and difficult. Collingridge's own formulation: "When change is easy, the need for it cannot be foreseen; when the need for change is apparent, change has become expensive, difficult, and time-consuming."
• Why it matters: This is the formal version of "offense spikes first, defense is slow to catch up," framed not as a military concept but as a governance problem applicable to all technologies.
• His solution: Not the precautionary principle, but "Intelligent Trial and Error" with decentralized decision-making, manageable changes, flexible infrastructure, and rapid learning.

The "Exponential Gap": Modern Framing

Azeem Azhar (2021): The Exponential Age

• Title: The Exponential Age: How Accelerating Technology is Transforming Business, Politics and Society
• Published: Diversion Books, 2021 (Financial Times Best Book of the Year)
• Key argument: The "exponential gap" is the rift between the pace of technological development and the pace of institutional adaptation. Technology grows exponentially. Institutions, regulations, and social norms grow linearly. This gap explains many contemporary problems: market concentration, gig economy exploitation, misinformation. But Azhar is not a pessimist. He proposes policy solutions to close the gap.
• Why it matters: Provides the modern, accessible version of the offense-defense argument. Technology outpaces institutions, creating temporary harms. Institutions can catch up, but only with deliberate effort.

Historical Pattern: Fear, Adaptation, Net Benefit

The pattern of initial fear followed by societal adaptation and net benefit is documented across several works:

• Spencer Weart (2012): The Rise of Nuclear Fear (Harvard University Press). Traces how public fear of nuclear energy evolved from the 1940s through Fukushima. Documents the gap between perceived risk and actual risk, and how institutional safeguards gradually reduced both.
• Elizabeth Eisenstein (1979): The Printing Press as an Agent of Change (Cambridge University Press). The canonical work on how the printing press disrupted existing knowledge structures, created initial chaos (propaganda, heresy, misinformation), and ultimately enabled the Scientific Revolution, Reformation, and Enlightenment.
• Vaughn Bell (2010): "Don't Touch That Dial! A history of media technology scares, from the printing press to Facebook." Slate, February 2010. Traces the recurring moral panics accompanying new communication technologies: writing (Plato), printing press (Gessner, 1565), newspapers, telegraph, telephone, radio, television, video games, internet, social media.
• Cultural and Historical Insights into Technological Fears (2024): "Anxiety Over the Future: Cultural and Historical Insights into Technological Fears." (ResearchGate, 2024). Examines the recurring pattern of technology anxiety across history.

The Synthesis: What the Literature Actually Says

The literature does not offer a simple "yes, benefits always outweigh harms." It offers something more nuanced and more useful:

The optimistic case (Pinker, Enlightenment tradition): On virtually every measurable dimension of human welfare, the long-term trend is improvement. Technology is a primary driver. Each new technology creates disruption and fear. Each time, humanity adapts. The trend line bends positive.

The risk case (Bostrom, Ord, Tegmark): Past performance does not guarantee future results. The trend line bends positive until it does not. Existential risks from sufficiently powerful technologies (engineered pandemics, unaligned AI, nuclear war) could be terminal. The "defense catches up" pattern has held so far, but a single failure at sufficient scale could end the game.

The structural case (Collingridge, Azhar, Jervis): The offense-defense dynamic is real. New technologies create offensive advantage. Defensive capability, institutional adaptation, and regulatory frameworks lag. But the lag is not permanent. With deliberate effort, the gap closes. The question is whether the gap closes fast enough, and whether we invest enough in closing it.

The AI-specific case (Lohn, Healey/Jervis/Nandrajog): In cybersecurity specifically, AI helps both offense and defense. The balance is not predetermined. Small defensive improvements can offset exponential offensive scaling. But this requires investment, and the outcome is not automatic.

Quick Reference Table