The Limits to Thinking Like a Computer Scientist

Neil Chilson

Acting Chief Technologist

Federal Trade Commission

nchilson@ftc.gov

@TechFTC

@telecomlawyer

Disclaimer

​

My views are my own and do not necessarily reflect the views of the FTC, its Commissioners and staff, or anyone else for that matter.

Overview

• Computational Thinking Is Important and Useful
• Computational Thinking Has Limits, Say Hayek and the Lego Movie
• Policymakers Should Use Regulatory Humility and Analytical Egalitarianism

Computational Thinking - What Is It?

Computational thinking is the thought processes involved in formulating a problem and expressing its solutions in such a way that a computer—human or machine—can effectively carry out. - Jeanette M. Wing

Often decomposed into three parts

• Abstraction - my focus
• Automation
• Analysis

WARNING - I’m building a bit of a strawman

Abstraction is Key

• Recognizing patterns in a class of problems and creating a generic solution
• Recognize what changes - the data
• Recognize what doesn’t change - the algorithm

Legal analogy: Going from common law to codified law

• “Black Box Thinking”
• Big problems can be decomposed into a set of smaller problems
• Once smaller problem abstracted, we treat the solution as a black box

Legal analogy (imperfect): scope of review on appeal

Abstraction is model building

Computational Thinking is Useful and Valuable

A huge swath of human endeavors benefit from computational thinking

• Does computational thinking = good thinking?
• Helps understand computing - critical as “software eats the world”

Benefits for the legal world

• Computation is changing how lawyers practice
• Automating the rote parts of legal practice
• Computation is changing what lawyers practice
• As software eats the world, raises many legal questions

​

​

​

But….Computational Thinking Has Limits

(like all thinking)

Hayek’s Knowledge Problem

F.A. Hayek - Use of Knowledge In Society (1945)

• Regulator cannot collect enough information to accurately predict future state
• Information is widely distributed and often contradictory
• Collection is inefficient
• Much knowledge is tacit - it cannot be explained even by the owner of the knowledge
• Much knowledge is irreducible
• Much of complexity theory builds on Hayek’s insights
• Order can emerge from decentralized decisionmaking
• Final results sensitive to initial conditions
• Thus: more information can be taken into account through decentralized processes -> Better solutions

Note: Hayek describes the limits of all thinking, not just computational thinking

​

​

The Lego Movie

Regulatory Humility

• Recognize limits of what regulation can accomplish
• Good intentions aren’t sufficient
• Focus on actual or likely harms
• Reduces knowledge problem
• Use the right tools
• Prefer those that facilitate emergence

Analytical Egalitarianism

• When we try to model policy outcomes, (a dynamic system) the model of the problem should include the solution - and possibly the modeler!
• We often model the societal problem, but treat the policy prescription as outside the model.
• But regulators are humans, too, mostly. They have incentives and flaws. The regulatory “solution” will reflect this.
• And the system reacts and adapts to policy interventions. Problems if we can’t anticipate those adaptations.

Conclusion

• Computational thinking is important, but in law and policy, it has its limits.
• We should prefer local decision making over global, and systems that support emergent order to those that attempt to constrain or replace it.

​

​

​

​