Cross-Subgraph Resonance Testing (Metaphor Forging)
Test framings live across unconnected audiences; the ones that land in distant subgraphs will travel.
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 90%
A method for developing ideas and framings that actually spread. Think out loud in as many conversations as possible, watch for the 'aha' reaction, and treat a hit across two people from very different backgrounds as strong evidence of a large ceiling — the social-graph intuition being that an idea validated only inside one dense cluster has a small implied maximum audience. Then invest incrementally in each hit, tightening the framing every time you use it, and build the winners into metaphors that deliberately leave the last dot unconnected.
Origin
Komoroske's own process — he describes himself as an external thinker who literally cannot think without talking, so he maximises interesting meetings. The viral-ceiling intuition comes from social network graph theory; the metaphor-as-parable technique he links to his preference for open-ended parables (and to why The Green Knight stuck with him).
Core principles
- 01You discover what you think by talking, so engineer maximum interesting conversation surface.
- 02The 'aha' — an unprompted 'oh, right' — is the unit of signal.
- 03One hit is a data point; a hit across two distant subgraphs implies a large ceiling.
- 04Invest incrementally on each hit, tightening the framing every time.
- 05Metaphors work because they connect nine of ten dots and invite the listener to connect the tenth.
- 06Because the listener supplies the last dot, a metaphor can carry a controversial claim safely.
How to run it
- 1
Maximise interesting conversation surface
Deliberately put yourself in the most interesting information streams with people who have taste and a coherent perspective — including ones you disagree with. Take as many generative meetings as you can. This is your idea-collection substrate.
Pro tip Play the collector: allow yourself to hear a good idea and immediately build on it, rather than sorting it as 'someone else's'.
Watch out Attribute honestly. Komoroske notes that when he cannot recall a metaphor's origin, he assumes with 95% confidence somebody else said it first.
- 2
Float the framing and mark the ahas
Say the framing out loud and watch the reaction. A genuine 'aha' — them going 'oh, right' — is the marker. Write down the exact wording that produced it: 'that was the best formulation of that one so far.'
Pro tip Capture the winning formulation mid-conversation. This is why fast note-taking is load-bearing for the whole method.
Watch out Polite interest is not an aha. Only the surprise reaction counts.
- 3
Test across maximally distant subgraphs
Deliberately float the same idea to people from unconnected worlds — a salesperson and an engineer, not two PMs. If people with very little overlap in the social/professional graph both find it interesting, the implied maximum audience is much larger than a hit inside one dense cluster.
Pro tip The measure is not 'how many people liked it' but 'how diverse were the people who liked it'.
Watch out If it only lands inside your own densely-interconnected cluster, the implied ceiling is small no matter how enthusiastic they are.
- 4
Reinvest a little on each signal
Every time a framing gets a signal, invest slightly more thought in the framing for next time. Keep coming back to it, keep tightening it, keep watching how it responds with different people. This is gardening applied to ideas.
Pro tip Over thousands of iterations this is where the library of metaphors comes from — not from a metaphor-generation session.
- 5
Build it into an evocative nine-of-ten-dots metaphor
Package the winning idea as a metaphor that connects nine of the ten dots and requires the listener to connect the last one. Because they complete it themselves, they own the conclusion. Make the metaphor vivid and a little subversive so it survives the extra interpretive work you are asking of them.
Pro tip This is how you say the unsayable: state the argument fully and it reads as 'against the official strategy' and you are dangerous. Leave one dot open and the listener connects it and says 'oh my God, that applies to us.'
Watch out Because the argument is less explicit, the metaphor must carry more load — a dull metaphor with a missing dot just reads as vague.
In the wild
Only after the fact did Komoroske work out why that deck, of everything he had written, got such enormous traction: slime is gross and bad, and the deck argues slime is good — an instantly subversive hook. It was also built almost entirely from emojis, which is not how anyone reads a deck.
→ The most-shared thing he has produced; perplexity's co-founders cited it publicly when explaining how they organize their team, which is how Lenny found him.
Describing LLMs, Komoroske landed on 'it's like an electric bike for idea spaces — you can cover so much more ground so much more quickly'. It lands because it connects nine dots to Steve Jobs's 'computers are a bicycle for the mind' and leaves the listener to close the tenth.
→ The listener supplies the last connection themselves and feels the insight as their own — the exact mechanism that makes a framing spread.
Common mistakes
Validating only inside your own cluster
If the only people who love the idea are densely interconnected with each other, all you have learned is that it works for that audience. The implied ceiling is small, and you will be blindsided when it fails to travel.
Connecting all ten dots
A fully spelled-out argument gives the listener nothing to do and, if it contradicts the official line, marks you as a danger. Leaving the last dot open both engages them and gives the idea political cover.
Trying to generate metaphors on demand
The library comes from thousands of live conversational tests plus incremental reinvestment on the hits. Sitting down to invent an evocative metaphor for today's deck skips the entire selection mechanism.
Is it for you?
Best for
Strategists, founders and writers who need their ideas to actually spread inside an organization or a market, and who think best out loud.
Not ideal for
Confidential or pre-announcement work that cannot be floated in casual conversation, and people who genuinely think best alone on paper.
From the transcript
“if you have if a person in sales and a person in engineering both find the same idea interesting that's a very good sign that…”
“if you had people that are in very different subgraphs a very little overlap they both find it interesting that implies a much larger”
“each time you get a signal like that you invest a little more time in it and you think a little bit more about framing…”
“one of the benefits of metaphorical thinking is you connect nine of the 10 dots and you allow you invite the The Listener to engage…”
“you have to make the metaphor evocative and interesting”
From the episode
Thinking like a gardener not a builder, organizing teams like slime mold, the adjacent possible, and other unconventional product advice
Alex Komoroske (Stripe, Google)