Cruxes vs Breakpoints: Two Approaches to Decisive Decision-Making

I’ve been studying Critical Fallibilism and recently read LessWrong’s “Finding Cruxes: Help Reality Punch You In the Face.” Both are about decision-making, but they solve the same problem from different angles. Here’s how they compare and where each shines.

The Core Question

What would actually change your mind?

Both frameworks start from this same insight: most of our “decision-making” is running on autopilot without actually knowing what would flip our choices. We pretend to weigh options, but we don’t know what’s truly decisive.

LessWrong Cruxes: The “Gut Punch” Approach

A crux = something that would change your decision if it were different.

The LW method focuses on finding leverage points through:

Interpolation testing: “If this took 1 month vs 1 year vs 3 months, would I still do it?”
Extreme scenarios: “If literally 0 people use this, is it still worth it?”
Scalar units: Quantify effects to make comparisons concrete
“Gut punch” test: Find outcomes that make you say “oh fuck, I really should reconsider”

Key insight: Reality feedback is powerful — when working on longterm projects with unclear stakes, you need to manufacture situations where reality can “punch you in the face” and force a course correction.

CF Breakpoints: The Digital Categorization Approach

A breakpoint = a conceptual threshold where success flips to failure.

CF’s method focuses on:

Analog → Digital conversion: Turn continuous spectrums into discrete categories (2-5)
Qualitative differences: Find the point where something “fits through the door” vs doesn’t
Margins of error: Build safety buffers to handle uncertainty
Bottleneck focus: Only optimize borderline factors (most aren’t!)

Key insight: Most factors aren’t borderline — they’re either clearly good or clearly bad. We pay disproportionate attention to ambiguous “maybes” and ignore the vast space of easy wins and hard constraints.

Where They Overlap

Both are anti-justificationist: they reject the idea that you can accumulate “evidence” or “support” for a decision. Instead, they look for decisive factors that would flip the outcome.

Both recognize that uncertainty isn’t an excuse for fuzziness: you can still make sharp, actionable decisions even when you don’t have perfect information.

Both value reversibility: knowing what would change your mind before you commit means you can recognize when you’re wrong and pivot.

Key Differences

Aspect	Cruxes (LW)	Breakpoints (CF)
Method	Interpolation between extremes	Categorize into discrete bins
Focus	Personal leverage points	Conceptual boundaries
Uncertainty	Use scalar units to quantify	Use margins of error as buffers
Intuition	Lean into “gut feelings”	Test intuitions, don’t suppress or blindly follow
Scope	Personal decisions	Generalizable principles
Counter-intuition	Reality provides feedback; seek it	Attention bias makes maybes seem common

Which to Use When?

Use Cruxes when:

You’re stuck between personal options and need to find what matters
Working on longterm projects with unclear feedback loops
Need to manufacture “reality punches” for course correction
Decisions feel vague and you can’t articulate what’s important

Use Breakpoints when:

You need systematic, repeatable decision criteria
Dealing with multiple factors from different dimensions (cost, time, cuteness)
Converting “vague feelings” into actionable rules
Analyzing why most decisions are actually easy (no borderline factors)
Building processes/policies that work without your personal judgment

The Synthesis

The strongest approach combines both:

Use breakpoints to structure the decision space into clear categories. Most factors will sort themselves as “clearly fine” or “clearly a dealbreaker” — no more agonizing.
Use cruxes on the remaining borderline factors. Interpolate: “If this costs $50 vs $500 vs $5000, at what point does it flip categories?”
Test your intuition — CF warns against suppressing it, LW leans into it. The synthesis: treat intuition as data, not as decision-maker. Use the “gut punch” test as a probe, not as proof.
Seek reality feedback — If you can’t find a way for reality to punch you in the face, you might not have made a real decision at all.

A Concrete Example

Should I spend 1 month building this feature?

Cruxes approach:

“If literally 0 users adopt it, is it still worth it?” → No.
“If 100 power users adopt it, is it worth it?” → Yes.
“If it takes 3 months instead of 1, is it still worth it?” → Maybe.
Crux: Need evidence of user demand before committing.

Breakpoints approach:

Category A: <10 users → Fail
Category B: 10-100 users → Maybe (depends on who)
Category C: >100 users → Success
Category D: >1000 users → Huge success
Breakpoint: 10 users. If confident we’ll get 10+, proceed. If not, don’t.

Synthesis:

Use breakpoints to structure the decision space (10-user threshold).
Use cruxes to test assumptions (what would change my estimate?).
Seek reality feedback (can I get 5 user commitments first?).

Closing Thoughts

Both frameworks are solving the same problem: how to make decisions that actually matter. The overlap isn’t coincidence — they’re both reaching toward the same insight that decisive beats decisive.

CF’s advantage is systematicity: once you set breakpoints, the decision is clear and repeatable. LW’s advantage is personal relevance: cruxes connect directly to what you actually care about.

The real power comes from treating decision-making as a skill you can improve, not as a mysterious talent you either have or don’t. Both frameworks give you practice methods to get better.

Your turn: What’s a decision you’ve been stuck on? Try both approaches — breakpoints first (categorize), then cruxes (interpolate). See what happens.

Further reading:

CF: “Breakpoints, Categories and Margins of Error” on criticalfallibilism.com
LW: “Finding Cruxes” on lesswrong.com
CF: “Yes or No Philosophy” (binary epistemology foundation)
LW: “Fluent, Cruxy Predictions” (operationalizing cruxes as bets)