Two senior engineers disagree strongly about an architectural direction and the team is stalled. How should an architect resolve it?

The first move

Establish what would change each person's mind. If neither can name evidence that would change their position, the disagreement is about values or preference rather than about facts, and no amount of further discussion resolves it — a different mechanism is needed.

If both can name evidence, the disagreement is empirical and the answer is to get the evidence: a prototype, a benchmark, a load test, a small experiment. That is usually days of work and it ends the argument definitively, which weeks of discussion will not.

When it is genuinely a values disagreement

Common and legitimate: one engineer weights operational simplicity, the other weights flexibility; one weights speed to market, the other weights long-term maintainability. Both positions are defensible and the disagreement cannot be resolved by evidence.

The moves:

• Name it explicitly. Making it clear that this is a values trade-off rather than a factual dispute frequently defuses it, because both parties stop trying to prove the other wrong.
• Identify who decides, by accountability rather than by seniority. Whoever owns the consequence should own the choice.
• Decide, record the decision and the dissent, and move. The dissent in the record is valuable: it is what a future reader needs when the conditions change.
• Set a review trigger — a condition under which the decision is revisited — which makes the decision reversible and lowers its stakes.

What not to do

• Prolonging discussion for consensus. Some disagreements do not converge, and the cost of the stall usually exceeds the cost of being wrong on a reversible decision.
• Deciding by seniority alone, which produces compliance rather than commitment and leaves the reasoning unrecorded.
• Splitting the difference, which frequently produces a design with the disadvantages of both.
• Escalating without a recommendation, which asks someone with less context to make a decision.

The framing that matters

Sort by reversibility. A decision that is cheap to reverse should be made quickly by whoever is closest, with a bias to action; an expensive one warrants the evidence-gathering. Most stalled architectural disagreements are about reversible decisions being treated as irreversible, and pointing that out is frequently the whole intervention.