# Topic Selection

> Turn a broad interest into a tractable research question or essay topic with defensible scope and available sources. Use when narrowing research questions, thesis scopes, or project ideas.

- Skill: `poorvith-mp/topic-selection` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add poorvith-mp/topic-selection`
- Raw SKILL.md: https://api.skillmd.com/api/skills/poorvith-mp/topic-selection/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: poorvith-mp (https://skillmd.com/u/poorvith-mp)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/poorvith-mp/topic-selection

---


# topic-selection

## Core Philosophy
Selecting a research topic or thesis question is the highest-leverage decision in scholarship. A brilliant methodology and tireless work ethic cannot salvage a topic that is fatally flawed from inception: one that is too broad, lacks accessible empirical data, carries insuperable ethical blockers, or merely confirms obvious trivia that nobody disputes. High-impact topic selection is a **rigorous funneling process**: transitioning from a broad personal fascination to a narrow, tractable, and non-trivial inquiry that satisfies the **FINER criteria** (Feasible, Interesting, Novel, Ethical, Relevant) and advances scholarly understanding.

---

## 4-Step Topic Selection & Scoping Architecture

### Step 1: The Funnel Deconstruction Framework
1. **The 4-Stage Scoping Funnel**:
   - **Stage 1: Broad Domain of Interest**: e.g. *Artificial Intelligence in Healthcare*. (Hopelessly broad; millions of papers).
   - **Stage 2: Specific Sub-Domain**: e.g. *Diagnostic Medical Imaging in Oncology*.
   - **Stage 3: The Empirical Controversy / Tension**: e.g. *Deep learning models achieve high AUROC in clinical trials, but degrade catastrophically on out-of-distribution hospital hardware*.
   - **Stage 4: The Defensible, Bounded Research Question**: e.g. *"To what extent does domain-adversarial neural training improve cross-scanner generalizability for MRI glioblastoma segmentation without requiring target-site retraining?"*

### Step 2: The FINER Evaluation Protocol
1. **F — Feasible**:
   - Do you have access to the necessary datasets, lab equipment, computing budget, and participant pools within the available timeframe?
2. **I — Interesting**:
   - Does this inquiry engage current debates in the literature?
3. **N — Novel**:
   - Does this project provide new empirical data, test an established theory in a novel context, or challenge existing consensus?
4. **E — Ethical**:
   - Can this project pass Institutional Review Board (IRB) or human subjects ethics clearance without prohibitive legal barriers?
5. **R — Relevant**:
   - How does answering this question advance academic knowledge, public policy, or clinical/engineering practice?

### Step 3: Feasibility Risk Auditing
1. **The Three Fatal Killers**:
   - **The Data Desert**: Selecting a question where required data is proprietary, classified, or inaccessible.
   - **The Infinite Scope**: Choosing a problem requiring a 20-person multi-year research lab for an individual 9-month Master's thesis.
   - **The Solved Problem**: Selecting a topic that was definitively answered and settled in the literature 5 years ago.

### Step 4: Framing the Formal Research Proposal Abstract
1. **The 4-Sentence Proposal Pitch**:
   - Sentence 1 (Context & Problem): The current state of the art and its unresolved tension.
   - Sentence 2 (The Specific Gap): What exact mechanism remains unmeasured or unknown.
   - Sentence 3 (The Proposed Approach): The methodology and data you will deploy to test it.
   - Sentence 4 (The Expected Contribution): What the field will know once your study concludes.

---

## Deliverable Format: Research Topic Evaluation & Scoping Matrix

| Candidate Topic | Feasibility (1-5) | Novelty (1-5) | Data Access | Primary Risk Factor | Committee Verdict |
|---|---|---|---|---|---|
| **Topic A**: "Impact of AI on Healthcare" | 1 / 5 (Impossible scope) | 1 / 5 (Trite / Vague) | Ubiquitous but unfocused | Zero defensible boundary; reads like a freshman essay. | **REJECT** |
| **Topic B**: "Classifying EHR data using proprietary Epic Systems APIs" | 2 / 5 (Legal blocker) | 4 / 5 (High value) | Blocked by vendor NDA | Inability to publish or share training code/data due to corporate IP. | **REJECT** |
| **Topic C**: "Evaluating cross-scanner generalizability of domain-adversarial CNNs on open BraTS MRI dataset" | 5 / 5 (9-month timeline) | 4 / 5 (Solves recognized fragility) | Fully accessible (BraTS open data) | Requires substantial GPU compute; mitigated by university cluster access. | **APPROVED** |

---

## Worked Example: Rescuing a Graduate Student from Scope Paralysis

- **Initial Student Proposal**: "I want to research how social media causes depression in teenagers."
- **Critique**:
  - Unbounded scope (which platforms? what age bracket? what causality mechanism?).
  - Ethical hurdles (collecting psychiatric data from minors requires 6-month IRB clearance).
  - Vast, saturated existing literature.
- **Funneled Revision**:
  - *"Investigating the relationship between TikTok algorithmic consumption time and self-reported sleep latency among undergraduate students (ages 18-22): A 14-day ecological momentary assessment study."*
- **Outcome**: The student secured IRB approval in 2 weeks, collected longitudinal smartphone screen-time logs from 80 undergraduates, and published the findings in a peer-reviewed behavioral health journal.

---

## Verification Checklist

- [ ] Broad interest funneled down to a specific, bounded, and testable question.
- [ ] Topic audited against all 5 FINER criteria.
- [ ] Data sources identified, accessible, and free of fatal legal/NDA blockers.
- [ ] Scope realistically executable within the available calendar and financial budget.
- [ ] 4-sentence proposal pitch articulates context, gap, method, and contribution.

---

## Anti-Patterns

- **The Grand Unified Theory**: Attempting to solve world poverty, cure cancer, or solve general intelligence in a single master's dissertation.
- **Selecting a Topic Before Checking Data Availability**: Spending 3 months writing a theoretical proposal only to discover the required dataset does not exist or cannot be accessed.
- **The Passion Trap Without Scholarly Value**: Insisting on researching a pet hobby that lacks theoretical grounding or relevance to the academic literature.