New Math Explains Why Group Decisions Get Stuck on Old Drafts

Picture a committee, a jury, or a team editing a strategy document. In this new paper, two researchers study a style of group decision-making called "sequential deliberation": at each round, two people chosen at random negotiate a small revision to the group's current answer, and that revised answer becomes the starting point for the next pair. Earlier theories assumed people treated the current answer only as a fallback — what happens if talks fail. This paper adds a very human ingredient: anchoring, the way an existing draft quietly drags everyone's opinion toward it.

Their model puts each person's position on a single line (say, a budget number or a left-to-right political leaning). Each round, both people drift a little toward the current draft, then split the difference. The result is a clean trade-off. Strong anchoring makes the group slow to change, but it also lowers the "social cost" — the final answer ends up closer to what the group actually wants. Weak anchoring lets positions move quickly, but leaves the group further from the ideal outcome. In the worst case, the group lands about 21% off the best possible answer; as anchoring gets very strong, that gap almost vanishes.

The math also identifies a unique "deliberative fixed point" — a natural resting place where the group stops drifting — and shows that stronger anchoring clusters everyone tightly around it. That has a real-world read. It explains inertia in committees and long-running policy debates. It also speaks directly to AI-assisted work: when a chatbot writes the first draft of a memo, that draft anchors every human who edits it afterward.

The honest catch: this is a mathematical model, not a study of real people. It assumes two idealized negotiators, one single issue, and no ego, office politics, unequal power, or manipulation. Anchoring can stabilize a group, but it can also lock everyone near a mediocre first draft — and the model can't yet tell you which you're getting.