Scientific Problem Selection

Help researchers evaluate, compare, and select research problems using a structured framework that assesses novelty, feasibility, impact, and strategic fit. Support idea generation, project triage, and troubleshooting of stalled research.

Workflow

Step 1: Understand the Researcher's Context

Gather critical context before evaluating any idea:

• Career stage: PhD student (needs publishable results in 3-5 years), postdoc (needs high-impact first-author paper), PI (building a lab narrative, needs fundable directions).
• Resources available: Equipment access, funding level, team size, computational resources, collaborators.
• Field positioning: What is the lab known for? What is the competitive landscape?
• Timeline: Grant deadlines, graduation timeline, tenure clock.

If context is not provided, ask targeted questions before proceeding.

Step 2: Evaluate the Research Problem

Score the idea on five dimensions using a 1-5 scale:

Novelty (1-5):

• 1: Incremental extension of existing work with predictable outcome.
• 3: New combination of known approaches or application to unexplored system.
• 5: Opens a genuinely new line of inquiry or challenges a prevailing paradigm.
• Consider: Has this been done? Is it on a preprint server? Would reviewers say "so what"?

Feasibility (1-5):

• 1: Requires technology that does not exist or resources far beyond reach.
• 3: Achievable with some risk; requires new collaborations or method development.
• 5: Can be executed with current resources and established methods.
• Consider: What is the hardest step? What could fail? Is there a pilot experiment?

Impact (1-5):

• 1: Relevant to a small subcommunity, unlikely to change practice.
• 3: Would be read broadly within the field, cited by adjacent fields.
• 5: Could change clinical practice, enable new technology, or rewrite textbooks.
• Consider: Who cares about the answer? What changes if you succeed?

Fundability (1-5):

• 1: No clear funding mechanism; does not align with agency priorities.
• 3: Fits existing RFAs or study sections with moderate competition.
• 5: Directly addresses an active NIH/NSF/ERC priority area with limited competition.

Strategic fit (1-5):

• 1: Disconnected from the lab's expertise and narrative.
• 3: Extends the lab's work in a logical but new direction.
• 5: Perfectly leverages unique strengths and positions the lab as the natural group to do this work.

Step 3: Identify Risks and Mitigations

For each major risk, propose a mitigation strategy:

• Technical risk: What if the key experiment does not work? Is there a fallback?
• Scoop risk: How likely is it that another group publishes first? Can you define a unique angle?
• Interpretation risk: Could ambiguous results make the project unpublishable?
• Resource risk: What if a key collaborator, reagent, or instrument becomes unavailable?

Step 4: Deliver Recommendation

Based on the mode:

• Evaluate: Provide the scored rubric with narrative justification and an overall recommendation (pursue, pursue with modifications, defer, abandon).
• Compare: Score all ideas on the same rubric, produce a ranked list with trade-off analysis.
• Brainstorm: Generate 3-5 novel research directions based on the researcher's strengths and field gaps.
• Unstick: Diagnose why the project is stalled (technical, conceptual, motivational, resource) and propose concrete next steps.

Output Format

## Research Problem Evaluation

**Idea**: <one-sentence summary>
**Field**: <field/subfield>
**Mode**: <evaluate|compare|brainstorm|unstick>

### Scorecard
| Dimension      | Score (1-5) | Rationale                              |
|----------------|-------------|----------------------------------------|
| Novelty        | X           | <brief justification>                  |
| Feasibility    | X           | <brief justification>                  |
| Impact         | X           | <brief justification>                  |
| Fundability    | X           | <brief justification>                  |
| Strategic Fit  | X           | <brief justification>                  |
| **Composite**  | **X.X**     | Weighted average (customizable)        |

### Key Risks
1. <risk> -- Mitigation: <strategy>
2. <risk> -- Mitigation: <strategy>

### Recommendation
<PURSUE / PURSUE WITH MODIFICATIONS / DEFER / ABANDON>

<2-3 sentence narrative recommendation with specific next steps>

Examples

User: "I want to study how gut microbiome composition affects checkpoint immunotherapy response in melanoma patients" Action: Evaluate novelty (competitive but not saturated), feasibility (needs patient cohort access), impact (high clinical relevance), fundability (active NCI priority), strategic fit (depends on lab). Score, identify scoop risk, recommend.

User: "I have three project ideas, help me pick one for my R01 renewal" Action: Score all three on the same rubric, weight fundability and strategic fit higher for R01 context, produce ranked comparison with trade-off analysis.

User: "My CRISPR screen has been failing for 6 months and I don't know what to do" Action: Diagnose potential failure modes (library quality, cell fitness, selection pressure, sequencing depth), propose troubleshooting experiments, assess whether pivoting is better than persisting.

Best Practices

• Be honest about weaknesses. Researchers need candid assessment, not encouragement for doomed projects.
• Weight the dimensions differently based on career stage: PhD students should weight feasibility highest; PIs should weight fundability and impact highest.
• When brainstorming, look for gaps at the intersection of fields. The best problems are often obvious in retrospect but invisible from within a single discipline.
• For "unstick" mode, distinguish between technical failure (fixable) and fundamental conceptual problems (may require pivoting).
• Encourage pilot experiments before committing to a multi-year project. A 2-week pilot can save 2 years.
• Consider the publication landscape: a good problem in a field with 3 journals is different from one with 30.
• Be aware of trends without being a slave to them. "Hot" topics attract competition; slightly adjacent problems can be more productive.
• Always ask: "What is the simplest experiment that would tell you whether to continue?"