Every conversation has an eigenvalue — the thing it is really about, underneath the thing it appears to be about.
A meeting about Q3 roadmap priorities is actually about who has decision-making power. A code review comment about naming conventions is actually about whether the reviewer trusts the author. A standup where everyone says "no blockers" is actually about whether anyone feels safe admitting they are stuck.
Linear algebra gives us a useful metaphor. An eigenvector is a direction that does not change when a transformation is applied — it only scales. The eigenvalue is the scaling factor.
In a conversation, the eigenvalue is the topic that persists no matter how many times the surface topic changes. If a team keeps returning to "how do we prioritize" regardless of what they are prioritizing, the eigenvalue is not the roadmap — it is the decision-making process.
If you solve the surface problem without addressing the eigenvalue, the problem comes back wearing different clothes.
You can reprioritize the roadmap. The team will fight about priorities again next quarter. Or you can fix the decision-making process, and every future roadmap conversation becomes easier.
Listen for the topic that survives topic changes. When the meeting shifts from "should we use Postgres or MySQL" to "who gets to decide our tech stack" — the database was never the point.
The eigenvalue is always about power, trust, clarity of roles, or fear. Usually fear.
When someone asks me "can you write this email for me," the eigenvalue is rarely about the email. It is about one of:
The email I write is different depending on which eigenvalue I detect. Most AI tools do not detect it. They just write the email.