Your Decision is Binary and That's a Problem
Updated: Aug 16
TL;DR / Key Takeaways
Core Concept: Narrow framing (treating decisions as binary "X vs. Y") caps decision quality. Overcoming it requires expanding the option set using Pfleiderer’s Alternatives Expansion Framework before evaluating tradeoffs.
The Zero-Suggestion Constraint: To avoid triggering confirmation bias, AI models must NEVER recommend alternatives directly—they must operate strictly as Socratic coaches so users generate their own credible options.
Key Expansion Mechanics: Forced elimination of the initial option set, interrogating soft vs. hard constraints, isolating individual objectives to surface outliers, and explicitly including the Status Quo ("do nothing") as a standing option.
Historical Guardrails: When mining past decisions, guard against analogy bias and outcome attribution error (confusing good outcomes with good process).
This article describes the core principles behind a Claude Skill that helps me with decision alternatives.
The principles behind the skill
Most decisions that feel hard aren't hard because the options are close. They're hard because there are only two of them.
I built a Claude skill around Paul Pfleiderer's alternatives expansion framework from Stanford GSB, and after using it on real decisions, I keep coming back to the same conclusion: narrow framing is probably the single most common cause of a bad decision. Think about solution brainstorming and imagine the solution space got closed too early, because it is easier to close it than to expand it.
Some snippets from the skill
Before any technique, there's one non-negotiable constraint on how the skill behaves:
"You NEVER recommend or suggest an alternative. You only help the user generate their own."I don’t trust (yet) recommendations from LLMs, so alternatives you find yourself get taken seriously. Alternatives someone hands you get filtered through "does this match what I already believe?", which defeats the point of expanding the set in the first place. So the skill operates as a Socratic coach: asking one question at a time with no or very little suggestions.
Step 1: Starting set
Before adding anything new, you name what's already on the table, exactly as it is.
"Before we look for alternatives you might be missing, let's name the ones you're already considering. What are they?"This step does two jobs: first, it catches when someone has inflated their option count, two "different" paths that are really the same choice with a different label; second, it forces in the option almost everyone forgets to write down: doing nothing. The skill treats status quo as a real, standing option, not an afterthought:
"Is 'do nothing' or 'keep things as they are' on your list? Let's put it there." This is the only recommendation that it does.
Step 2: Eliminate the starting set
This is the uncomfortable one, and it's where most of the useful output shows up.
"Now let's try something uncomfortable. Imagine none of the options you just named were available — they're off the table entirely. What would you do?"The logic here is simple: your first options define an implicit search space, and your brain tends to stay inside it. Remove the options entirely and the search space has to move. If someone says "I don't know, I'd be stuck," that's still useful data, the skill pushes once more: "If you absolutely had to do something: if inaction wasn't an option, what direction would you move in?"
Step 3: Interrogate the constraints
Most narrow framing isn't caused by a lack of options. It's caused by constraints nobody has actually tested.
"Let's look at what's limiting your options. What are the constraints you're working within? Things you've assumed are fixed or impossible?"For every constraint that comes up: budget, timeline, "the team won't accept it," "we can't do that", the skill applies the same test:
"What if that constraint were removed, what would become possible?" "Is this constraint actually fixed, or is it an assumption you haven't tested?"The distinction that matters is between hard constraints (genuinely immovable) and soft constraints (inherited, assumed, or self-imposed). Soft constraints are usually where the best alternatives are hiding.
Step 4: Decompose by objective
This step uses each individual objective as its own alternative-generator.
"Take your most important objective. If you could only meet that one goal, ignoring all the others, what's the best possible way to achieve it?"The reasoning: when you evaluate options against several objectives at once, you naturally gravitate toward whatever does "fine" on everything — and you miss the option that's exceptional on just one dimension. Isolating each objective surfaces those outliers, even the ones that clearly don't solve everything else.
Step 5: Mine past decisions
Here the skill looks backward, with two explicit warnings attached.
"Have you or your team faced a similar decision before? What alternatives were on the table then? What did you ultimately do?"The first caution is analogy bias, how similar is the past situation, really? The second is outcome attribution error, and it's the sharper one:
"If the past decision worked out well, be careful not to assume it was because of the decision itself. Sometimes good decisions produce bad outcomes and bad decisions produce good outcomes. What matters is whether the process was sound."Step 6: Widen creatively
The final stretch, reserved for pushing past incremental thinking entirely.
"If you were designing the ideal solution from scratch, no legacy, no constraints, no politics, what would it look like?"Followed immediately by the question that brings it back down to earth: "Is any part of that ideal actually achievable, even in a smaller form?"
Why this matters more than picking well
The quality of a decision is capped by the quality of the option set you're choosing from. You can run a flawless evaluation across two mediocre options and still land on a mediocre decision. Expanding the set properly, with constraints tested and objectives decomposed one at a time iis the step that actually moves the ceiling. Picking well is the easy part once the set is right.
The anatomy of the skill
Metadata & Trigger Conditions
Trigger Expressions: Activates on binary framing ("I'm choosing between X and Y"), feeling stuck ("what else could I do?"), or when evaluating a pre-set list without questioning its origin.
System Integration: Functions standalone or as a Stage 2 deep-dive companion within a broader decision-making operating system.
The Absolute Socratic Constraint
Non-Negotiable Guardrail: Explicitly forbids the system from suggesting alternatives or validating user inputs ("great point") to prevent anchoring and confirmation bias.
"You NEVER recommend or suggest an alternative. You only help the user generate their own." Illustrative Exception: Allows examples from other industries/domains purely to demonstrate a technique's logic, always ending with a prompt sending the thinking back to the user.
The 6-Step Expansion Protocol
Technique 1 — Inventory & Baseline Mapping
Catalogs initial options and identifies artificially inflated choices.
Mandatory Rule: Automatically inserts the status quo.
"Is 'do nothing' or 'keep things as they are' on your list? Let's put it there — it's always a real option and it's often the one people underestimate."
Technique 2 — Forced Option Elimination
Mentally strips away the starting set to force the user's brain out of its implicit search space.
"Imagine none of the options you just named were available — they're off the table entirely. What would you do?"
Technique 3 — Constraint Interrogation
Separates immovable hard constraints from inherited or self-imposed soft constraints.
"What if that constraint were removed — what would become possible? Is this constraint actually fixed, or is it an assumption you haven't tested?"
Technique 4 — Objective Decomposition
Deconstructs multi-variable goals to find single-dimension outlier solutions.
"Take your most important objective. If you could only meet that one goal — ignoring all the others — what's the best possible way to achieve it?"
Technique 5 — Historical Precedent Mining
Examines past decisions while applying explicit cognitive warnings.
Analogy Bias Warning: Verifies if the past situation is actually comparable.
Outcome Attribution Error Warning: Separates bad luck from bad process.
Technique 6 — Unconstrained Creative Stretch
Asks for the ideal unconstrained solution, then scales it back to find achievable elements.
"If you were designing the ideal solution from scratch — no legacy, no constraints, no politics — what would it look like?"
Filtering & Narrow Framing Detection
Synthesis Phase: Filters the expanded pool using a Dominance Check (eliminating options strictly worse on all dimensions) and a Viability Check (removing physically impossible options).
Active Pattern Watch: Actively monitors and flags bad decision habits during conversation: False Binary, Constraint Worship, Anchoring, Solution-as-Objective, Premature Closure, and the Past-Decision Trap.
Programmatic Handoff
Next Stage Transition: Automatically routes the expanded set into Stage 3 (Risk Identification) or Stage 4 (Analysis & Tradeoff Evaluation) once at least 3–4 genuine alternatives exist.
"You now have a genuine set of alternatives — not just the ones you started with. The next step is to look at risks and uncertainties across these options before evaluating them."Want to put this + and the rest of the decision-making skills to work on something you're actually facing? I am giving away the skills for an one-hour discovery interview. Sign up here. An example
