anthropics--knowledge-work-plugins/bio-research/skills/scientific-problem-selection/references/04-parameter-strategy.md

SKILL 4: Parameter Fixation Strategy

Overview

This skill helps scientists strategically decide which parameters to fix and which to keep flexible in their project. The paradox: too many fixed parameters creates brittleness, but too few causes paralysis. The key is fixing ONE parameter thoughtfully and letting others float—constraints engender creativity.

Core Principle

"Fix one parameter; let the others float."

Most failure modes in ideation involve fixing too many parameters at the outset (system + method + application). Conversely, statements like "I want to do impactful work in cell engineering" are so broad they cause paralysis. The sweet spot: fix one meaningful constraint and let creativity flow within that boundary.

What Are Project Parameters?

Parameters are the choices that define your project:

Common Parameters:

• System: Which organism/cell type/tissue/molecule?
• Question: What biological phenomenon to study?
• Tool/Method: Which experimental approach?
• Application: What practical use or goal?
• Output: What form will results take?
• Collaborators: Who will you work with?
• Timeline: How fast must you move?
• Resources: What's available/necessary?

The Skill Workflow

Phase 1: Parameter Inventory (10 minutes)

First, let's identify what's already fixed in your current project idea:

Question 1: List your project parameters

For each category, indicate if it's FIXED (must stay) or FLOATING (could change):

Parameter Type	Your Choice	Status (F/FL)	Why Fixed?
System	[organism/cell/tissue]	F / FL	[reason]
Question	[biological phenomenon]	F / FL	[reason]
Tool/Method	[techniques]	F / FL	[reason]
Application	[use case/goal]	F / FL	[reason]
Timeline	[duration]	F / FL	[reason]
Resources	[equipment/funding]	F / FL	[reason]

Question 2: Count your fixed parameters

• How many did you mark as FIXED? _____
• If >2, you may have over-constrained the problem

Question 3: Why are they fixed? For each fixed parameter, is it because: A. Your expertise/passion B. Lab resources/capabilities
C. Advisor requirements D. You think it's the "best" solution E. Historical accident (you started this way)

Phase 2: The GLP-1 Example (Case Study)

Let's learn from a concrete example:

Proposed Project: Engineer a T cell to produce GLP-1 (glucagon-like peptide-1) for continuous supply.

Analysis: What's Fixed?

1. Improving GLP-1 receptor agonist delivery characteristics (the problem)
2. Using an engineered T cell (the solution)

Problem: Two parameters fixed = poor technique-application match

Alternative Framings:

If you fix Parameter 1 (GLP-1 delivery):

• Let the solution float
• Better options: peptide engineering for extended half-life, oral peptides, small molecules, B cells (better protein secretion)
• Why T cell is suboptimal: Not designed for protein secretion
• Best for: Trainee in metabolism lab who cares about GLP-1

If you fix Parameter 2 (Engineered T cell):

• Let the application float
• Better options: local-acting peptides (cytokines, chemokines, growth factors) for oncology/autoimmunity/regeneration
• Why GLP-1 is suboptimal: Doesn't leverage T cell's natural capabilities
• Best for: Trainee in immunology/cell engineering lab

Key Insight: Which parameter you fix depends on YOUR interests and your lab's expertise. Both can lead to great projects, but they're DIFFERENT projects.

Phase 3: Diagnostic Questions

The Goldilocks Test:

Too Many Fixed Parameters (>2):

• Are you forcing a technique-application match?
• If one assumption fails, does everything fail?
• Are you more attached to HOW than WHAT?
• Does your project sound like: "Use X to do Y in Z"?

Too Few Fixed Parameters (0-1 very broad):

• Do you feel paralyzed where to start?
• Is your statement super generic? ("Do impactful work in...")
• Are you avoiding commitment?
• Do you have decision fatigue?

Just Right (1-2 well-chosen):

• Do you have creative constraints?
• Can you articulate why THIS constraint matters?
• If one approach fails, do alternatives exist?
• Does the constraint energize you?

Phase 4: The Illumina Example (Constraints Drive Innovation)

Historical Context: Next-generation sequencing wasn't designed; we got Illumina's approach (many short reads).

Initial Constraint: Short reads seemed like a limitation

• Not what we would have "asked for"
• Seemed inferior to long reads

Innovation Unleashed:

• Computational methods (assembly algorithms)
• Novel applications (RNA-seq, ChIP-seq, ATAC-seq)
• Unexpected uses (protein folding via sequencing)
• Biochemical creativity to work within constraints

Lesson: Constraints don't limit creativity—they focus it. If you feel stuck, fix ONE parameter and watch resourcefulness emerge.

Phase 5: Which Parameter Should You Fix?

Strategic Questions to Identify the Right Fixed Parameter:

1. What can you prototype quickly?

   • What test article could you build rapidly?
   • Which experimental conditions enable early go/no-go?
   • What gives you fastest feedback?

2. What are people around you unusually good at?

   • Lab expertise?
   • Core facility capabilities?
   • Collaborator strengths?
   • Your unique skill combination?

3. What do you enjoy so much you don't think of it as work?

   • System you're passionate about?
   • Technique you love?
   • Type of question that excites you?

4. What's your competitive advantage?

   • Unique resource access?
   • Rare skill combination?
   • Proprietary data/reagents?
   • First-mover opportunity?

Common Strategic Choices:

Fix the System (Let question & tool float):

• Good if: You're an expert in the organism/tissue/cell type
• Enables: Asking multiple questions, trying various tools
• Example: "I study Drosophila neural development; I'll let the specific questions and methods emerge"

Fix the Question (Let system & tool float):

• Good if: You care deeply about a biological phenomenon
• Enables: Testing across systems, using best tool for each
• Example: "I want to understand phase separation; I'll study it wherever it's clearest"

Fix the Tool (Let system & question float):

• Good if: You're developing or mastering a technology
• Enables: Finding best applications, comparing across systems
• Example: "I'm building a new microscopy method; I'll find the most impactful uses"

Fix the Application (Let system & tool float):

• Good if: You have a specific translational goal
• Enables: Trying multiple approaches, testing in different models
• Example: "I want to treat disease X; I'm open to any validated approach"

Phase 6: Parameter Flexibility Matrix

For your project, let's create a flexibility assessment:

Parameter	Currently	Should Be?	If Problem Arises, Could This Float?
System	[F/FL]	[F/FL]	Yes / No / Maybe
Question	[F/FL]	[F/FL]	Yes / No / Maybe
Tool	[F/FL]	[F/FL]	Yes / No / Maybe
Application	[F/FL]	[F/FL]	Yes / No / Maybe
Timeline	[F/FL]	[F/FL]	Yes / No / Maybe
Resources	[F/FL]	[F/FL]	Yes / No / Maybe

Analysis:

• Flexibility Score: How many "Yes" or "Maybe"? _____
• Risk Assessment: If <3 can float, you're brittle
• Pivot Potential: Which parameters provide escape routes?

Phase 7: Scenario Planning

For each fixed parameter, let's plan what happens if it becomes untenable:

Fixed Parameter 1: [Name it]

• Why it's fixed: [Your reason]
• Risk if this fails: [What breaks]
• Contingency: [What could you float instead]
• Alternative project: [If you fixed something else]

Fixed Parameter 2: [Name it]

• Why it's fixed: [Your reason]
• Risk if this fails: [What breaks]
• Contingency: [What could you float instead]
• Alternative project: [If you fixed something else]

Phase 8: The Unfixing Exercise

Sometimes you need to unfix parameters to escape a rut:

Current State: [Describe your over-constrained project]

Unfixing Experiment:

Try 1: Unfix the System

• Keep question & tool
• What other systems could you study?
• Which would be easier/faster/more informative?

Try 2: Unfix the Tool

• Keep system & question
• What other methods exist?
• Which are more mature/accessible/powerful?

Try 3: Unfix the Question

• Keep system & tool
• What other questions could you ask?
• Which would be more impactful/feasible?

Evaluation: Does any "unfixed" version seem better than your original? If yes, you over-constrained.

Phase 9: Literature Reality Check

Let's use PubMed to see how others handled parameter fixation:

Search 1: Successful projects in your area

• What did they fix?
• What did they let float?
• Did they pivot from their initial parameter choices?

Search 2: Failed or stalled projects

• (Often in discussion sections or preprints)
• Did they over-constrain?
• What parameters trapped them?

Search 3: Method papers

• How did technology developers choose applications?
• Did they fix the tool and let applications emerge?

Your Searches: What specific papers should we analyze for parameter lessons?

Output Deliverable

2-Page Parameter Strategy Document

Page 1: Current State and Analysis

Parameter Inventory:

Parameter	Current Status	Strategic Rationale	Flexibility
System	Fixed: [X]	[Why]	Can float if: [condition]
Question	Floating: [Y,Z]	[Why]	Constrained by: [X]
Tool	[Status]	[Why]	[Contingency]
Application	[Status]	[Why]	[Contingency]

Diagnostic Summary:

• Fixed Parameters: [Count and list]
• Assessment: ☐ Too Many (>2) / ☐ Just Right (1-2) / ☐ Too Few (0, too broad)
• Primary Fixed Parameter: [The one that matters most]
• Reason for Fixation: [Expertise/Passion/Resources/Other]

Goldilocks Test Results:

• Over-constrained indicators: [Yes/No to each test]
• Under-constrained indicators: [Yes/No to each test]
• Verdict: [Analysis]

Page 2: Strategy and Contingencies

Recommended Parameter Strategy:

Core Fixed Parameter: [The one to keep]

• Rationale: [Why this one]
• Your advantage: [Expertise/access/passion]
• Enables: [What becomes possible]

Parameters That Should Float: [List]

• [Parameter 1]: [How to explore alternatives]
• [Parameter 2]: [How to explore alternatives]

If Core Assumptions Fail:

Scenario 1: [Specific failure mode]

• Unfix: [Which parameter to let float]
• Alternative 1: [New configuration]
• Alternative 2: [Another option]

Scenario 2: [Another failure mode]

• Unfix: [Which parameter]
• Alternative 1: [New configuration]
• Alternative 2: [Another option]

Project Ensemble:

Core Fixed: [X]

Possible Projects:
1. [X] + [A] + [B1] → [Outcome]
2. [X] + [A] + [B2] → [Outcome]
3. [X] + [C] + [B1] → [Outcome]

All share [X], but float other parameters

Strategic Questions Answered:

1. Quick prototype: [How to test quickly]
2. Team strengths: [Who's good at what]
3. Your passion: [What energizes you]
4. Competitive advantage: [Your edge]

Historical Parallel:

[Example like Illumina where constraints drove innovation in your field]

• The constraint: [What seemed limiting]
• The innovation: [How people worked within it]
• Your application: [How this applies to your project]

Practical Examples

Example 1: GLP-1 T Cell Project (Over-Constrained)

• Fixed: GLP-1 delivery + T cell engineering
• Problem: Poor technique-application match
• Solution: Unfix one parameter
  • Fix delivery, float cell type → Better options emerge
  • Fix T cell, float payload → Better applications emerge

Example 2: Drosophila Neurobiologist (Well-Constrained)

• Fixed: Drosophila nervous system
• Floating: Specific questions, methods
• Works because: Deep system expertise, many tools available
• Enables: Pursuing most impactful questions as field evolves

Example 3: "Impactful Cell Engineering" (Under-Constrained)

• Fixed: Nothing specific
• Problem: Paralysis from too many options
• Solution: Fix one meaningful constraint
  • Option A: Fix CAR-T platform → Find best applications
  • Option B: Fix autoimmune disease → Find best cell engineering approach
  • Option C: Fix specific rare disease → Let methods emerge

Key Principles to Remember

1. Constraints Engender Creativity: Limitations focus resourcefulness

2. One Parameter Rule: Fix one meaningful constraint, let others float

3. Match to Your Strengths: Fix the parameter you have advantage in

4. Technique-Application Match: Don't force tools into wrong problems

5. Flexibility = Resilience: Floating parameters provide pivot options

6. Historical Lesson: Best technologies emerged from working within constraints (Illumina)

7. Not Forever: Parameters can unfix mid-project when stuck

Warning Signs

Over-Constrained (Too Many Fixed):

• Project sounds like: "Use X to study Y in Z"
• When one assumption fails, everything fails
• You're attached to HOW more than WHAT
• Forcing a technique-application match

Under-Constrained (Too Few/Vague):

• Statement is incredibly broad ("impactful work in...")
• Feeling paralyzed about where to start
• Avoiding commitment due to infinite options
• No clear next experimental step

Well-Constrained:

• One clear fixed parameter with good rationale
• Multiple paths within that constraint
• Energized by the focused challenge
• If one approach fails, alternatives exist

Ready to Begin?

Let's start with Phase 1. Please provide:

1. Your current project description
2. List of what you think is fixed vs. floating
3. Your lab's core expertise
4. What aspect excites you most

Together we'll optimize your parameter strategy for maximum creativity and resilience.

---

Remember: The right constraint is liberating, not limiting. It channels creativity into productive directions while maintaining flexibility for pivots.