When to reach for it
- The stochastic diversity of models should be leveraged.
- The result must be robust against individual logic errors in single runs.
Self-Consistency.
Vote across many minds. Settle on the majority.
The system generates multiple independent reasoning outputs for the same prompt, and then merges the results via consensus or voting for a final answer.
When it backfires
- The task can be deterministically validated anyway.
- The cost per request is strictly limited.
The tradeoff
Significantly more robust answers against false paths, paid for by multiple inference costs.
What it actually does.
Several samples vote; the majority answer wins.
Two ways this pattern will hurt you.
Majority agreement on the same hallucination
All reasoners share the same training bias and agree on an incorrect answer. Majority voting amplifies the error.
Fix · Use diverse prompts or models for each reasoner, or add a verifier that checks against an external source.
Even number of voters with no tie-break rule
A 2-2 split leaves the majority node without a clear winner. The system hangs or picks arbitrarily.
Fix · Use an odd number of reasoners, or define a tie-break (e.g., highest-confidence vote, or escalate to human).
Where Self-Consistency is native.
Microsoft Agent Frameworkmulti-sampleNative
Anthropic Cookbookdocumented as a recipeAdaptable
Next pattern