Skip to content

creative-ideation

Category: audit
Field: economics
License: private (curator-owned)
Updated: 2026-05-20
Stages: referee-simulation

Curator-private skill — copy text from 100xOS/shared/skills/reasoning/creative-ideation.md.

Creative Research Ideation

The Goal

Generate novel, feasible research ideas that a top economist would find worth pursuing. The best ideas start with a real-world phenomenon that demands explanation, not with a gap in the literature.

The Three-Filter Framework

Every research idea must pass three filters before it is worth pursuing:

Filter 1: Is it a real phenomenon?

  • Does something surprising, consequential, or puzzling happen in the world?
  • Can you state it in one sentence without referencing any academic paper?
  • Would a smart non-economist find it interesting or counterintuitive?
  • Is the phenomenon specific enough to study in one paper?

Bad: "No one has studied how DeFi affects price discovery" (literature gap, not phenomenon) Good: "DEX prices lead CEX prices for small-cap tokens, but lag for large-cap — why?" (observable pattern that demands explanation)

Filter 2: Is there a credible identification strategy?

  • Can you identify a causal channel, or is this purely descriptive?
  • Is there a natural experiment, instrument, or sharp design?
  • What is the key identifying assumption, and is it defensible?
  • Does the data exist (or can it be constructed) to implement this design?

Sources of identification in crypto/DeFi: - Protocol governance votes that change parameters (DiD) - Regulatory shocks with clear timing (event study) - Threshold-based rules in smart contracts (RDD) - Geographic variation in regulation (cross-country DiD) - Token airdrops as exogenous wealth shocks - Gas price spikes as exogenous transaction cost variation - Bridge exploits/hacks as exogenous liquidity shocks

Filter 3: Does the answer matter?

  • Who would change their behavior based on the answer?
  • Does it inform policy (regulation, protocol design, risk management)?
  • Does it test or extend economic theory in a meaningful way?
  • Would a finance or economics journal care, or is this only interesting to crypto insiders?

Idea Generation Process

Step 1: Start with phenomena, not methods

List 5-10 things that happen in the domain that are surprising, consequential, or unexplained. Do not think about methods yet.

Step 2: For each phenomenon, ask "why?"

  • What economic mechanism could explain this?
  • What existing theory predicts the opposite? (tension = paper)
  • What would have to be true for this to happen? (testable implications)

Step 3: For each "why," identify the data

  • What variation in the data would let you test the mechanism?
  • Is there a natural experiment or quasi-random variation?
  • Can you construct the key variables from available data?

Step 4: For each viable idea, position it

  • What are the 3 closest existing papers?
  • How does this idea differ from each?
  • Which journal would this target?

Step 5: Rank by feasibility × impact

Score each idea on: - Data availability (1-5): Can you get the data without heroic effort? - Identification credibility (1-5): How clean is the causal design? - Novelty (1-5): How far is this from existing work? - Importance (1-5): How much would the answer matter? - Feasibility (1-5): Can one person do this in 3-6 months?

Total = Data × min(Identification, Novelty) × Importance × Feasibility

Patterns That Generate Good Ideas

Theory meets new data

"Classic theory X predicts Y, but we've never been able to test it because Z data didn't exist. Now blockchain/DeFi/new-data-source provides exactly the variation we need."

Mechanism design meets reality

"Protocol designers chose mechanism X. Theory predicts consequence Y. Does it happen? What are the unintended consequences?"

Traditional finance in a new setting

"Phenomenon X is well-documented in equity markets. Does it also occur in DeFi, where frictions/institutions/information structure differ? What does the comparison reveal about the mechanism?"

Regulatory arbitrage as identification

"Regulation X applies in jurisdiction A but not B. Comparing outcomes across jurisdictions identifies the causal effect of the regulation."

Design discontinuities

"Smart contracts create sharp thresholds (liquidation ratios, fee tiers, governance quorums). These are natural RDD setups."

Anti-Patterns to Avoid

  • "First to study X": Being first is not a contribution. What do we learn?
  • "We use fancy method Y": Methods serve questions, not the other way around.
  • "Crypto is important because market cap": Importance of the setting ≠ importance of the question.
  • "We build a dashboard/index/tool": Engineering is not research unless it generates new economic insight.
  • "Everything is correlated with Bitcoin": Correlation is not identification. What is the mechanism? What is the counterfactual?
  • Literature gap as motivation: "No one has studied X" is an observation, not a reason. "X happens and we don't understand why" is a reason.