First published March 13, 2016.
Are overly polite people more likely to betray you? The short answer is no, not as a rule. The idea comes from a real and interesting 2015 study, but the study looked at players of the strategy game Diplomacy, not at strangers at a party, and its findings were far subtler than the headlines that followed. What it did find is that shifts in how someone talks to an ally, especially sudden bursts of positivity, can come shortly before a betrayal. Below is what the researchers actually did, what they found, and how to use the lesson without becoming suspicious of every kind person you meet.
The study behind the claim
The paper is called "Linguistic Harbingers of Betrayal: A Case Study on an Online Strategy Game". It was written by Vlad Niculae and Cristian Danescu-Niculescu-Mizil of Cornell University, Srijan Kumar of the University of Maryland and Jordan Boyd-Graber of the University of Colorado, and presented at the 2015 annual meeting of the Association for Computational Linguistics (ACL) in Beijing.
Some popular write-ups said the researchers "staged a game" at the conference. They did not. They analysed an existing archive of online games: 249 games of Diplomacy containing about 145,000 messages between players. Conference presentation is simply where the results were first shared.
Why Diplomacy is a good laboratory for betrayal
Diplomacy is a board game set in Europe on the eve of the First World War. Each player controls a great power. There are no dice: everything depends on negotiation. Players make alliances in private messages, coordinate moves, and then, at some point, most alliances end, sometimes with one partner attacking the other without warning.
That makes the game unusually useful for research. Every alliance and every betrayal is recorded in the game's moves, and every conversation leading up to it is saved as text. The researchers could therefore compare the messages exchanged in friendships that ended in betrayal with those in friendships that lasted, matching them for length and timing so the comparison was fair.
What the researchers actually found
The results were about imbalances: differences between how the two partners in an alliance talked to each other. Among the main findings:
- Positivity. The future betrayer tended to express noticeably more positive sentiment than partners in lasting friendships. The authors suggested this might be a form of overcompensation.
- Planning. The future victim talked more about plans for the coming turns, while the betrayer talked less about the future, as if the alliance no longer had one for them.
- Politeness. This is where headlines went furthest. Across a whole friendship, betrayers were somewhat more polite, but the politeness imbalance was slight and did not reach conventional statistical significance. As betrayal approached, betrayers actually became less polite, while victims became more polite and made more requests.
- Talkativeness. A growing imbalance in how much each side wrote was strongly linked to friendships breaking down.
In other words, a sudden rise in someone's politeness was more a sign that they were about to become the victim than the betrayer.
Where the "57%" number really comes from
Popular articles in 2016 claimed that in 57% of cases where someone was overly nice, they went on to betray their partner. That is not what the figure means.
The researchers built a computer model that read the messages and tried to guess which friendships would end in betrayal. It was right 57% of the time, compared with about 52% for guessing by chance. That is a modest improvement, and the authors were clear that the signals are subtle and that they did not expect high-accuracy detection. Interestingly, the human players who were later betrayed did no better than chance at seeing it coming.
So the honest version is: a computer could detect weak warning signs a little better than a coin toss, in one online game. It says nothing about the odds that a polite person you meet will turn on you.
What this can and cannot teach us about everyday life
A game where betrayal is part of the rules is not a model of friendship, work or family. In Diplomacy, everyone knows that alliances are temporary and that winning usually means turning on someone. In ordinary life, most courtesy is exactly what it looks like: good manners, habit, or genuine warmth.
Psychology also warns against judging character from first impressions. People vary widely in how formal or friendly they are, and those differences are shaped by culture, upbringing, nerves and even the situation. A shy person may seem cold; a well-raised one may seem extra polite. Neither tells you much about trustworthiness.
What the study does support, in a cautious way, is paying attention to change. In the game, the warning signs were shifts in tone and in balance between partners, not politeness on its own.
Practical takeaways that hold up
- Watch actions over time, not one conversation. Trust is built when what someone says matches what they do, repeatedly.
- Notice sudden shifts. If a person who ignored you becomes very warm when they need something, it is fair to ask what has changed. That is curiosity, not paranoia.
- Look for one-sided conversations. In the study, imbalance was a warning sign. In real relationships, a partner who stops talking about shared plans, or who suddenly talks much less or much more, may be signalling something worth asking about.
- Do not punish kindness. Treating every polite stranger as a suspect costs you good relationships and does not reliably protect you.
- Set small tests before big trust. In business or money matters, start small and let reliability be proven before stakes rise.
Common myths about politeness and deception
Myth: liars are always extra nice. Deception research has found no single behaviour, friendly or otherwise, that reliably reveals a liar. Most people, including trained professionals, perform only a little better than chance at spotting lies from behaviour.
Myth: blunt people are more honest. Bluntness is a style, not a guarantee. Rudeness and honesty are separate traits.
Myth: science has proved polite people betray more. As shown above, the source study found only a slight politeness effect, and in the period right before betrayal the pattern even ran the other way.
The real lesson of the Diplomacy research is a gentle one: relationships give off signals, and changes in how people talk to us are worth noticing. That is a better habit than suspicion, and it leaves room for the many people who are simply kind.
Sources: Niculae, Kumar, Boyd-Graber and Danescu-Niculescu-Mizil (2015), Linguistic Harbingers of Betrayal, ACL Anthology; author copy of the paper; earlier coverage via I Heart Intelligence.

COMMENTS