commit

Author: tecunningham

_freeze/posts/2025-XX-XX-tech-offense-defense/execute-results/html.json

@@ -580,44 +580,40 @@ Apply the framework to real cases:
 
 # Summary
 
-Technology doesn't uniformly help offense or defense—the structure of the conflict determines which side benefits.
+Technology doesn't uniformly help offense or defense—**structure determines outcomes**.
 
-Four key mechanisms:
+| Mechanism                   | Key Feature           | Favors      | Examples                     |
+| --------------------------- | --------------------- | ----------- | ---------------------------- |
+| Finite search & fixing      | Exhaustible holes     | **Defense** | Code fuzzing                 |
+| Finite search & monitoring  | Limited attention     | **Offense** | Network intrusion            |
+| Verification (ground truth) | External validation   | **Defense** | Document authentication      |
+| Examination (appearance)    | No ground truth       | **Offense** | Essay detection, deepfakes   |
 
-| Mechanism                   | Key Feature              | Technology Favors | Examples                              |
-| --------------------------- | ------------------------ | ----------------- | ------------------------------------- |
-| Finite search & fixing      | Exhaustible vulnerabilities | **Defense**       | Code fuzzing, vulnerability scanning  |
-| Finite search & monitoring  | Limited attention        | **Offense**       | Network intrusion, perimeter security |
-| Verification (ground truth) | External validation      | **Defense**       | Document authentication, watermarking |
-| Examination (appearance)    | No ground truth          | **Offense**       | Essay detection, deepfakes            |
+**Speed matters:** Rapid technological change creates offensive advantage—each capability arrives before the previous one is defended.
 
-The speed of technological change creates an additional offensive advantage: rapid advancement prevents defense from catching up, even in structurally defensive domains.
+**Implications:**
+- Defensive domains → Accelerate discovery and patching
+- Offensive domains → Invest in provenance systems (create ground truth)
+- All domains → Consider slowing capability advancement to allow defense time
 
-**For policy and strategy:**
-- In defensive domains (finite search, ground truth), invest in accelerating discovery and patching
-- In offensive domains (appearance-based), invest in provenance systems to create ground truth
-- In all domains, consider slowing the pace of capability advancement to allow defense time to adapt
-
-The offense-defense balance is not fixed—it's a design choice embedded in our institutions and technologies.
+The offense-defense balance is not fixed—it's a design choice.
 
 # References
 
-Key papers on offense-defense balance and conflict:
-
-Garfinkel & Skaperdas (2007) "Economics of Conflict: An Overview"
-: Comprehensive review of contest models and conflict economics. Discusses how returns to scale affect equilibrium efforts.
+**Garfinkel & Skaperdas (2007)** "Economics of Conflict: An Overview"
+: Contest models and conflict economics. Returns to scale affect equilibrium efforts.
 
-Garfinkel & Dafoe (2019) "How does the offense-defense balance scale?"
-: Argues that growth in investments tends to favor defense when investment levels are high.
+**Garfinkel & Dafoe (2019)** "How does the offense-defense balance scale?"
+: Investment growth tends to favor defense at high levels.
 
-Glaser & Kaufmann (1998) "What Is the Offense-Defense Balance and How Can We Measure It?"
-: Defines the balance as the ratio of forces attacker needs relative to defender's deployed forces.
+**Glaser & Kaufmann (1998)** "What Is the Offense-Defense Balance and How Can We Measure It?"
+: Defines balance as ratio of attacker forces needed to defender forces deployed.
 
-Schneier (2018) "Artificial Intelligence and the Attack/Defense Balance"
-: Argues AI will favor defense because current attacks exploit human weaknesses that AI can shore up.
+**Schneier (2018)** "Artificial Intelligence and the Attack/Defense Balance"
+: AI favors defense by shoring up human weaknesses that attacks exploit.
 
-Heather Adkins (2024) "The Offense-Defense Balance"
-: Discusses how vulnerability discovery is accelerating with LLMs, with the key question being who runs faster at finding and fixing vulnerabilities.
+**Adkins (2024)** "The Offense-Defense Balance"
+: Vulnerability discovery accelerating with LLMs. Key: who runs faster at finding and fixing?
 
-Bressler, Trager, Dafoe (2021) "The Offense-Defense Balance and the Costs of Anarchy"
-: Shows that welfare can be better under high offensive advantage than intermediate levels, due to reduced conflict intensity.
+**Bressler, Trager, Dafoe (2021)** "The Offense-Defense Balance and the Costs of Anarchy"
+: Welfare can be better under high offensive advantage than intermediate (reduced conflict intensity).