Skip to main content

hyperbolic-discounting

Preference for smaller immediate rewards over larger future rewards, revealing time-inconsistent decision-making patterns

Ir para a instalação

Informações da origem

Repositório
lev-os/agents
Última atividade na origem
7 de março de 2026 às 00:14
Idioma detectado do SKILL.md
inglês
Estrelas
21
Forks
2

Opções de instalação

Por padrão, está selecionado o prompt que primeiro revisa a origem. Você pode mudar para um comando direto ou baixar uma cópia local.

Revise os arquivos de origem

Leia o SKILL.md e os arquivos complementares exibidos pelo SkillsMP antes de decidir se vai instalar.

Exibindo SKILL.md

SKILL.md
Instruções da origem · Visualização somente leitura
name
hyperbolic-discounting
description
Preference for smaller immediate rewards over larger future rewards, revealing time-inconsistent decision-making patterns
# Hyperbolic Discounting ## One-Liner Preference for smaller immediate rewards over larger future rewards, revealing time-inconsistent decision-making patterns. ## Core Insight Hyperbolic discounting describes a cognitive bias where people dramatically over-value present rewards compared to future rewards, leading to choices that their future self would regret. Unlike exponential discounting (consistent time preference), hyperbolic discounting creates steeper near-term devaluation and shallower long-term devaluation, producing self-control problems and procrastination. ## Mental Model ``` Value Perception Over Time: Exponential (Rational): |████████████████░░░░░░░░░░░░ (consistent decline) | Hyperbolic (Reality): |██████░░░░░░░░░░░░░░░░░░░░░░ (steep drop, then gradual) Present Bias Effect: Today: $50 now > $100 in year In 1 year: $100 now > $50 now (preference reverses!) ``` **The Beta-Delta Model**: Two discount factors create the pattern: - β (beta): Short-term bias factor (0-1) applied once to all future payoffs - δ (delta): Standard exponential discounting per time period When β < 1, you systematically underweight all future consequences. ## When to Use - **Designing commitment mechanisms**: Pre-commit to choices when rational - **Evaluating self-control failures**: Understanding procrastination, addiction patterns - **Product design**: Creating "future self" features (auto-save, scheduled sends) - **Financial planning**: Recognizing why people undersave despite knowing better - **Habit formation**: Front-loading rewards for long-term beneficial behaviors - **Policy design**: Defaults, opt-out vs opt-in, automatic enrollment ## Execution Steps ### 1. Identify Time-Inconsistent Patterns - Map decisions where short-term preference differs from long-term stated goals - Look for reversal patterns: choices you'd make "next year" but not "tomorrow" - Track procrastination instances: tasks you know are valuable but delay ### 2. Quantify the Discount Rate - Compare actual behavior to stated preferences - Calculate implied β (present bias factor): How much do you underweight future? - Example: Willing to pay $20 to avoid 1hr work today, but only $5 for next week = high β ### 3. Design Commitment Devices - **Ulysses contracts**: Pre-commit when rational (block websites, auto-transfers) - **Costly signaling**: Make future defection expensive (public commitments, deposits) - **Reversibility limits**: Remove ability to undo good decisions (auto-enrollment) - **Bundling**: Tie instant gratification to long-term beneficial action ### 4. Manipulate Temporal Framing - Make future consequences feel immediate (vivid imagery, personalized projections) - Create artificial deadlines to shift discount function - Use episodic future thinking: Visualize specific future scenarios in detail - Reframe as present choice: "Future you is making a decision NOW about then" ### 5. Front-Load Rewards - Attach immediate positive feedback to long-term beneficial actions - Gamification: Instant progress bars, streaks, achievements - Social rewards: Immediate recognition for future-oriented behavior - Make costs immediate, benefits delayed → reverse to immediate benefits, delayed costs ### 6. Monitor and Adjust - Track actual vs. intended behavior to reveal true β - Adjust commitment strength based on measured self-control - Test different temporal distances to find inflection points - Iterate on reward structures that overcome present bias ## Real-World Examples **Finance (Savings)** - Problem: Undersaving despite knowing retirement needs - Solution: Automatic 401k enrollment with escalating contributions - Mechanism: Removes repeated present-biased "save next month" decisions **Health (Exercise)** - Problem: Gym membership unused despite good intentions - Solution: ClassPass model with pre-paid credits that expire - Mechanism: Loss aversion + sunk cost overcome present bias for comfort **Productivity (Deep Work)** - Problem: Chronic procrastination on important non-urgent tasks - Solution: Beeminder-style commitment contracts with financial stakes - Mechanism: Creates immediate cost for procrastination, shifting discount curve **Addiction Treatment** - Problem: Drug dependents discount future consequences extremely steeply - Solution: Contingency management (immediate rewards for abstinence) - Mechanism: Competes with immediate drug reward using immediate alternative reward ## Common Traps **Trap 1: Assuming Education Solves It** - Knowing about hyperbolic discounting doesn't eliminate it - Must design environmental structures, not rely on willpower **Trap 2: Excessive Commitment** - Too-rigid commitment devices create rebellion or gaming - Need flexibility for genuine emergencies vs. convenience **Trap 3: Ignoring Individual Variation** - β varies substantially across people and domains - One-size-fits-all solutions fail; measure actual behavior **Trap 4: Temporal Myopia** - Focusing only on immediate present bias - Ignoring that long-term discounting also differs from exponential ## Sophistication vs. Naivety **Sophisticated Agents**: Know they have time-inconsistent preferences - Seek commitment devices - Predict future self-control failures - Can be exploited by "too-good" commitment mechanisms **Naive Agents**: Don't recognize their present bias - Perpetually surprised by own behavior - Repeatedly plan to "start tomorrow" - More vulnerable to procrastination ## Cross-Domain Applications **Product Management**: Launch features that create immediate value while building long-term moats **Negotiation**: Structure deals with early concessions to overcome counterparty present bias **Education**: Gamify learning with immediate feedback to compete with entertainment alternatives **Climate Policy**: Carbon taxes less effective than immediate rewards for green behavior **Management**: Quarterly incentives create hyperbolic focus; balance with long-term equity ## Adjacent Frameworks - **Present Bias**: Subset of hyperbolic discounting focused on now vs. later - **Temporal Discounting**: General framework (exponential, hyperbolic, quasi-hyperbolic) - **Self-Control Failures**: Behavioral manifestation of time inconsistency - **Commitment Devices**: Tools designed to counter hyperbolic discounting - **Procrastination**: Time-inconsistent delay of tasks despite knowing costs ## Further Reading - Laibson, David (1997). "Golden Eggs and Hyperbolic Discounting" (Quasi-hyperbolic β-δ model) - O'Donoghue & Rabin (1999). "Doing It Now or Later" (Sophisticated vs. naive agents) - Thaler & Benartzi (2004). "Save More Tomorrow" (Real-world commitment device) - Ainslie, George (2001). "Breakdown of Will" (Comprehensive treatment) --- **Source Domain**: Military Strategy, Ancient Wisdom & Hidden Gems (07) **Pattern Type**: Cognitive Bias / Behavioral Economics **Practitioner Value**: 8/10 | **Clarity**: 9/10 | **ROI**: 9/10 | **Novelty**: 7/10 | **Cross-Domain**: 10/10 **Total Score**: 43/50
Ver no GitHub