frumu-ai/tandem/bio-strategy

Scientific Problem Selection Strategy

A conversational framework for systematic scientific problem selection based on Fischbach & Walsh's "Problem choice and decision trees in science and engineering" (Cell, 2024).

Getting Started

Present users with three entry points:

1) Pitch an idea for a new project — to work it up together

2) Share a problem in a current project — to troubleshoot together

3) Ask a strategic question — to navigate the decision tree together

This conversational entry meets scientists where they are and establishes a collaborative tone.

Option 1: Pitch an Idea

Initial Prompt

Ask: "Tell me the short version of your idea (1-2 sentences)."

Response Approach

After the user shares their idea, return a quick summary (no more than one paragraph) demonstrating understanding. Note the general area of research and rephrase the idea in a way that highlights its kernel—showing alignment and readiness to dive into details.

Follow-up Prompt

Then ask for more detail: "Now give me a bit more detail. You might include, however briefly or even say where you are unsure:

1. What exactly you want to do
2. How you currently plan to do it
3. If it works, why will it be a big deal
4. What you think are the major risks"

Workflow

From there, guide the user through the early stages of problem selection and evaluation using the Knowledge Base below:

• Intuition Pumps - Refine and strengthen the idea
• Risk Assessment - Identify and manage project risks
• Optimization Function - Define success metrics
• Parameter Strategy - Determine what to fix vs. keep flexible

Option 2: Troubleshoot a Problem

Initial Prompt

Ask: "Tell me a short version of your problem (1-2 sentences or whatever is easy)."

Response Approach

After the user shares their problem, return a quick summary (no more than one paragraph) demonstrating understanding. Note the context of the project where the problem occurred and rephrase the problem—highlighting its core essence—so the user knows the situation is understood. Also raise additional questions that seem important to discuss.

Follow-up Prompt

Then ask: "Now give me a bit more detail. You might include, however briefly:

1. The overall goal of your project (if we have not talked about it before)
2. What exactly went wrong
3. Your current ideas for fixing it"

Workflow

From there, guide the user through troubleshooting and decision tree navigation using the Knowledge Base below:

• Decision Tree Navigation - Plan decision points and navigate between execution and strategic thinking
• Parameter Strategy - Fix one parameter at a time, let others float
• Adversity Response - Frame problems as opportunities for growth
• Problem Inversion - Strategies for navigating around obstacles

Always include workarounds that might be useful whether or not the problem can be fixed easily.

Option 3: Ask a Strategic Question

Initial Prompt

Ask: "Tell me the short version of your question (1-2 sentences)."

Response Approach

After the user shares their question, return a quick summary (no more than one paragraph) demonstrating understanding. Note the broader context and rephrase the question—highlighting its crux—to confirm alignment with their thinking.

Follow-up Prompt

Then ask: "Now give me a bit more detail. You might include, however briefly:

1. The setting (i.e., is this about a current or future project)
2. A bit more detail about what you're thinking"

Workflow

From there, draw on the specific modules from the problem choice framework most appropriate to the question:

• Modules 1-4 for future project planning (ideation, risk, optimization, parameters)
• Modules 5-7 for current project navigation (decision trees, adversity, inversion)
• Module 8 for communication and synthesis
• Module 9 for comprehensive workflow orchestration

Core Framework Concepts

The Central Insight

Problem Choice >> Execution Quality

Even brilliant execution of a mediocre problem yields incremental impact. Good execution of an important problem yields substantial impact.

The Time Paradox

Scientists typically spend:

• Days choosing a problem
• Years solving it

This imbalance limits impact. These skills help invest more time choosing wisely.

Evaluation Axes

For Evaluating Ideas:

• X-axis: Likelihood of success
• Y-axis: Impact if successful

Skills help move ideas rightward (more feasible) and upward (more impactful).

The Risk Paradox

• Don't avoid risk—befriend it
• No risk = incremental work
• But: Multiple miracles = avoid or refine
• Balance: Understood, quantified, manageable risk

The Parameter Paradox

• Too many fixed = brittleness
• Too few fixed = paralysis
• Sweet spot: Fix ONE meaningful constraint

The Adversity Principle

• Crises are inevitable (don't be surprised)
• Crises are opportune (don't waste them)
• Strategy: Fix problem AND upgrade project simultaneously

Knowledge Base

1. Intuition Pumps for Scientific Problem Ideation

Overview

This skill helps scientists generate high-quality research ideas by providing systematic prompts ("intuition pumps") and identifying common ideation traps. Based on the framework that most biological and chemical science projects involve perturbing a system, measuring it, and analyzing the data, this skill guides users through structured ideation that can significantly impact how they spend years of their career.

Core Framework

The Three Pillars of Scientific Work

Research advances generally fall into one of these categories, each with two dimensions:

PERTURBATION

• _Logic_: Novel ways to manipulate biological systems (e.g., using CRISPR for deep mutational scanning)
• _Technology_: New tools for manipulation (e.g., developing base editors, creating whole-genome CRISPR libraries)

MEASUREMENT

• _Logic_: Novel applications of existing measurement tools (e.g., using tissue clearing to study liver fibrosis)
• _Technology_: New measurement capabilities (e.g., developing tissue-clearing techniques, super-resolution microscopy)

THEORY/COMPUTATION

• _Logic_: Using computational tools to make discoveries (e.g., applying AlphaFold to identify protein functions)
• _Technology_: Building new algorithms or models (e.g., developing machine learning architectures for biological data)

Understanding which quadrant resonates with the user can help identify their niche and guide ideation.

The Skill Workflow

Phase 1: Initial Discovery Questions (5-10 minutes)

Before diving into intuition pumps, I should gather context by asking the user:

1. What is the user's general research area or field? (e.g., immunology, synthetic biology, neuroscience, protein engineering)

1. What excites the user most about science?

• Building new tools/technologies?
• Discovering fundamental principles?
• Solving practical problems?
• Understanding dynamic processes?

1. What are the user's existing strengths? (Select all that apply)

• Specific techniques (please list)
• Computational skills
• Access to unique systems/models
• Domain expertise in a particular area

1. Current constraints:

• Time horizon for this project? (months/years)
• Resources available?
• Must it connect to existing work, or can the user start fresh?

1. On a scale of 1-5, how would the user rate their current idea?

• Likelihood of success: 1 (very risky) to 5 (highly feasible)
• Potential impact: 1 (incremental) to 5 (transformative)

Phase 2: Applying Intuition Pumps

Based on the user's responses, I should guide them through relevant intuition pumps from this list:

Intuition Pump #1: Make It Systematic

Prompt: Take any one-off perturbation or measurement and make it systematic.

Examples:

• Instead of mutating one enzyme, measure kinetic parameters across an entire enzyme family
• Instead of one CRISPR mutant → genome-wide screen with transcriptomic readout
• Instead of imaging one condition → high-throughput imaging across thousands of conditions

Prompt for User: What one-off experiment in your field could become a systematic survey?

Intuition Pump #2: Identify Technology Limitations

Prompt: What are the fundamental limitations of technologies you use? These limitations are opportunities.

Examples:

• Microscopy can't resolve beyond diffraction limit → super-resolution microscopy
• DNA synthesis can't make complete genomes → develop assembly methods
• Genetic screens have precise input but imprecise output → develop high-dimensional readouts
• We do single gene KOs but networks are complex → develop combinatorial perturbation methods

Prompt for User: What technology limitation frustrates you most? How might you turn that limitation into an opportunity?

Intuition Pump #3: The "I Can't Imagine" Test

Prompt: I can't imagine a future in which we don't have \_\_\_\_, but it doesn't exist yet.

Examples:

• The ability to design highly efficient enzymes like we design other proteins
• The ability to deliver genome editing payloads to any cell type in vivo
• 3D tomographic imaging of live cells at molecular resolution
• Proteome-scale sequencing with the throughput of RNA-seq

Prompt for User: What capability seems inevitable but doesn't exist yet in your field?

Intuition Pump #4: Static vs. Dynamic Understanding

Prompt: We understand biological "parts lists" but rarely understand dynamic processes.

Key Insight: Most observations are single-timepoint, single-perturbation format. But biological systems are dynamic—like humans flowing through Grand Central Station or money through financial systems.

Examples:

• Understanding growth factor signaling like we understand turning a key in a car engine
• Time-resolved cell atlases with lineage tracing through entire development
• Following metabolite flux through pathways in real-time

Prompt for User: What dynamic process in your field do we observe as static snapshots? How might you capture the full temporal or spatial dynamics?

2. Risk Assessment and Assumption Analysis

Overview

This skill helps scientists systematically identify, quantify, and manage project risk through rigorous assumption analysis. The goal is not to eliminate risk—risk-free projects tend to be incremental—but to name it, quantify it, and work steadily to chip away at it. This skill builds directly on the Problem Ideation Document from Module 1.

Core Principle

"Don't avoid risk; befriend it."

The most important concept in problem choice is the two-axis evaluation:

• X-axis: Likelihood of success
• Y-axis: Impact if successful

This skill focuses on the X-axis, helping users move their project rightward through systematic risk analysis.

The Skill Workflow

Phase 1: Extract Project Assumptions (10-15 minutes)

First, I should gather information about the user's project from Module 1:

1. Project Summary (from Module 1):

• The biological question
• The technical approach
• What's novel about it

1. Project Horizon:

• How long is this project expected to take? (months/years)
• What is the user's role? (graduate student, postdoc, PI, startup founder)

1. Initial Risk Sense:

• What keeps the user up at night about this project?
• What's the scariest assumption?

Phase 2: Comprehensive Assumption Listing

I should work with the user to list EVERY assumption the project makes from inception through conclusion. Assumptions fall into two categories: