The Discount Rate
The discount rate is the economic term for the weight an agent places on future consequences relative to present ones. A low discount rate means the future matters nearly as much as the present — the agent saves, invests, builds slowly, accepts current costs for future benefits. A high discount rate means the future matters less — the agent consumes now, invests less, and makes choices that prioritize immediate satisfaction over long-term flourishing. In the Becker-Murphy rational addiction model, the discount rate is the parameter that separates the addict from the moderate consumer. The addict is not irrational. The addict is optimizing with a high discount rate, which produces systematically different choices than the same optimization with a moderate rate.
In the AI Story
In productive addiction, the discount rate is pushed higher by two forces. The first is social validation: the building session produces output the culture celebrates, lowering the perceived cost of continuation. The second is the visibility asymmetry between present returns and future costs. The present return on a building session is immediately observable: working code, visible features, measurable progress. The future cost — erosion of judgment, degradation of relationships, atrophy of attentional capacity — is not observable. It accumulates invisibly. The builder cannot see these costs until they have compounded beyond easy reversal.
The discount rate is not stable. It is affected by the consumption itself. Each session of high-intensity, high-return AI-augmented building trains the nervous system to weight immediate returns more heavily and future costs more lightly. The discount rate drifts upward with use. The agent becomes progressively less capable of the long-horizon thinking that would reveal the accumulating costs — not because she is less intelligent, but because the very cognitive apparatus that performs long-horizon evaluation has been reshaped by a pattern of use that systematically favors the short horizon.
This is the deep mechanism that makes productive addiction so difficult to manage. The AI tool does not merely provide a pleasant experience. It reshapes the agent's temporal orientation. Segal's self-report — catching himself at three in the morning, unable to stop, recognizing the pattern as the same compulsive loop he had experienced with earlier technologies — is a rational agent observing his own discount rate in real time and finding it dangerously high. The observation is itself a form of intervention: the agent who can see the discount rate is, at least momentarily, operating at a lower discount rate than the agent who cannot.
Origin
The discount rate is a foundational concept in economics, appearing in the earliest work on intertemporal choice by Irving Fisher and Paul Samuelson. Becker and Murphy's contribution was to make it the central parameter distinguishing addictive from non-addictive consumption, and to show that the rate is itself endogenous — shaped by the consumption patterns it governs.
Key Ideas
The discount rate is not a preference parameter. It is a variable that consumption patterns reshape, making it a moving target that cannot be held constant during analysis.
The visibility asymmetry drives the rate upward. Present returns are immediate and visible; future costs are delayed and invisible.
Social context matters. A culture that celebrates intense productive output lowers the perceived cost of continuation, effectively pushing up the discount rate at which continuation becomes rational.
Meta-cognition as intervention. Observing one's own discount rate is itself a form of intervention — momentarily accessing a lower rate than the consumption-driven rate the agent would otherwise apply.
Appears in the Orange Pill Cycle
Further reading
- Paul Samuelson, A Note on Measurement of Utility (Review of Economic Studies, 1937).
- Gary Becker and Kevin Murphy, A Theory of Rational Addiction (Journal of Political Economy, 1988).
- David Laibson, Golden Eggs and Hyperbolic Discounting (Quarterly Journal of Economics, 1997).