What the data actually shows
One influential account comes from Robin Dunbar, who argued that language itself may have evolved partly to serve the function that grooming serves in other primates — maintaining social bonds in groups too large to physically groom everyone. In this view, talking about who did what to whom is 'social grooming,' a way of building alliances and keeping a running map of a community's relationships. Gossip, on this account, is not a corruption of language but close to one of its original purposes.
Research on the content of gossip supports the unflattering label being mostly wrong. Studies that sample real conversations find that the bulk of talk about absent third parties is neutral or informational — who is doing what, who is with whom — rather than negative or derogatory. The fraction that is purely malicious is real but a minority of the total.
Work by Matthew Feinberg and colleagues highlights gossip's prosocial side: passing on information about people who behaved badly can warn others, deter exploitation, and enforce cooperative norms. In their experiments, the opportunity to gossip and exclude untrustworthy individuals tended to make groups more cooperative, and people often reported that warning others — even at a small cost to themselves — felt good rather than vindictive. Reputation, spread through gossip, acts as a low-cost mechanism for keeping behaviour in check.
Why this feels different from how it actually is
Gossip has a bad reputation partly because the word itself is loaded. We tend to call it 'gossip' only when it is negative or about someone we feel protective of, and 'catching up,' 'venting,' or 'just talking' when it is neutral or about ourselves. So the category as we experience it is skewed toward the unflattering cases, even though the behaviour as researchers define it is mostly mundane.
It also feels worse than it usually is because the harmful instances are memorable and the neutral ones are forgotten. A single cruel rumour leaves a mark that thousands of small informational exchanges do not, which makes the whole behaviour feel more dangerous than the average instance actually is.
And there is a genuine moral discomfort: talking about people behind their backs can feel like a small betrayal, even when the content is harmless. That discomfort is real and worth heeding, but it is not evidence that the underlying behaviour is primarily destructive — it is evidence that we hold ourselves to a standard about it, which is part of how the norm-enforcement function works.
The everyday behaviour the word describes is closer to social bookkeeping than to spite.
What the research says to do about it
If the concern is the harmful kind, the research points toward attending to function and content rather than trying to stop talking about other people, which is close to impossible and not obviously desirable. The useful question is whether a given exchange shares accurate, relevant information — a warning, a heads-up, a genuine concern — or whether it mainly aims to lower someone's standing for its own sake.
The prosocial-gossip findings suggest that passing on honest information about untrustworthy or exploitative behaviour can protect a group, and that this is one of the legitimate social roles gossip plays. Framing such conversations around accuracy and protection rather than contempt tends to keep them in the useful range.
Because gossip is how reputation circulates, the most reliable lever over what is said about you is, unglamorously, behaviour over time. Reputation research consistently finds that consistent conduct shapes the information that travels far more than any attempt to manage the talk directly.
What the research says does not help
Trying to swear off gossip entirely tends not to work and misunderstands the behaviour. Talking about absent others is a core social function, and the realistic goal is steering it toward accurate, fair, and necessary information rather than eliminating it. Vowing never to discuss anyone usually just relabels the same behaviour.
Assuming all gossip is malicious is also unhelpful and inaccurate. The evidence that most talk about others is neutral or informational means treating every such conversation as toxic misreads the majority of cases and can make people needlessly anxious or secretive.
Trying to control your reputation by spreading counter-narratives or managing what others say about you tends to backfire and is hard to sustain. The research suggests reputation is governed mainly by repeated behaviour observed over time, not by attempts to manage the flow of talk.
The most reliable lever over what is said about you is, unglamorously, behaviour over time.
What this looks like in real life
'Gossip' vs 'just catching up'
The same behaviour gets different names depending on how it feels. We tend to call it 'gossip' only when it's negative or about someone we feel protective of, and 'catching up,' 'venting,' or 'just talking' when it's neutral or about ourselves. So the category we experience is skewed toward the unflattering cases, even though the behaviour researchers define is mostly mundane.
A warning that protects the group
Passing on that someone has repeatedly let people down is a heads-up, not spite. In Feinberg's experiments, the chance to share reputational information and steer clear of untrustworthy people tended to make groups more cooperative — and people often reported that warning others, even at a small cost to themselves, felt good rather than vindictive. Framed around accuracy and protection rather than contempt, gossip stays in its useful range.
Real numbers in context
Estimates of how much of conversation is social in nature vary, and you should treat any single figure with caution, but multiple studies suggest a large share of everyday talk concerns social topics — people, relationships, and what others are doing. Within that, the recurring finding is that the clear majority of gossip is neutral or informational rather than negative; the purely malicious share is a minority.
The honest framing is qualitative more than quantitative. The robust pattern across the research is directional: gossip is near-universal, mostly mundane, and serves real social functions around information, bonding, and norm enforcement. Precise percentages differ by study and method, so the safest claim is the direction, not a specific number.