Groups are supposed to make better decisions than individuals: more perspectives, more expertise, more scrutiny of weak reasoning. In practice, groups frequently make worse decisions than a reasonably competent individual working alone would have. Understanding why clarifies what's actually worth fixing.
The failure modes
Groupthink is the best-known one: the tendency for a cohesive group to prioritize harmony and consensus over a realistic appraisal of alternatives. It shows up as premature closure around an option, self-censorship of doubts, and an illusion of unanimity that doesn't reflect genuine agreement so much as a shared reluctance to be the one who breaks it. Irving Janis's original research on this looked at major foreign-policy fiascos, but the pattern shows up in any group with strong cohesion and a directive, opinionated leader whose preference becomes obvious early.
Social loafing is the tendency for individual effort and accountability to diminish as group size grows, because responsibility for the outcome diffuses across everyone and therefore feels owned by no one in particular. In decision-making terms, this means fewer people in a large group actually do the hard work of independently thinking through the problem, and more people simply defer to whoever seems most confident.
The hidden-profile problem is subtler and arguably more damaging. Groups are good at discussing information that all members already share, and bad at surfacing information that only one member holds. If the single piece of information that would change the decision is known only to one person, group discussion often fails to bring it out, because the conversation naturally gravitates toward the shared ground everyone can already talk about.
What actually improves group decisions
Structured dissent, not spontaneous dissent. Waiting for someone to naturally disagree rarely works, because the social cost of disagreement is real and most people won't pay it voluntarily. Assigning an explicit devil's-advocate role, or using a pre-mortem exercise (imagining the decision failed, and working backward to explain why) gives dissent a structural home instead of relying on individual courage.
Independent input before group discussion. Having each person write down their initial view before the group conversation starts, and only revealing them after, prevents the anchoring effect where the first strongly-stated opinion in the room shapes everyone else's contribution. This single change addresses both groupthink and the hidden-profile problem simultaneously.
Smaller, more accountable subgroups for complex decisions. Diffused responsibility is a function of group size. Breaking a large group into smaller subgroups, each independently working the problem before reconvening to compare conclusions, restores the individual accountability that dilutes in a big room.
A designated challenger role that rotates. Having the same person always play devil's advocate turns them into "the person who always disagrees," which the group learns to discount. Rotating the role keeps the function without letting it calcify into a predictable, ignorable pattern.
None of this makes group decision-making effortless. It makes the known failure modes structurally harder to fall into, which is a more realistic goal than expecting good group dynamics to emerge on their own from good intentions.