You can't see into someone else's model of the world. You only have your model of their model — and it's almost always wrong. This is the first thing to get honest about.
People defend their belief systems by default. Not as a preference, but as architecture: brains are built to minimize surprise, and a coherent self-story is load-bearing. Hit it head-on and you'll lose. The belief system isn't the obstacle to the conversation — it is the conversation's substrate.
But defense isn't the whole picture. Beliefs are rigid at the core and malleable at the edges. Which one you're touching determines everything.
Three things govern whether a signal lands:
Framing. A challenge to a core belief gets rejected on contact. The same idea, reached through the person's own reasoning or arriving through a reframed context, can pass through cleanly. You're not changing the content — you're changing the angle of approach.
Stakes. The signal has to matter to them, in their terms, or it's noise. This requires modeling them — which you'll do imperfectly — but imperfect modeling beats none.
Role. People improvise from their position to keep their story intact. If you occupy a role with conviction, others tend to bend around it. Not always — entrenched power, hard evidence, and pre-assigned roles all push back — but more often than you'd expect in ambiguous or status-fluid situations.
The deeper point: social reality is co-constructed. Not invented from nothing, but propped up by the players together. The "hard lines" people defend are usually narrower than they appear, and most apparent blockers yield to a shift in framing rather than a stronger push.
So: model humbly, frame carefully, and play the role you actually want to occupy. The room has more give than it looks.