This paper presents an algorithm that detects misunderstandings in collaborative work it a distance. It analyses the movements of collaborators' eyes oil the shared workspace, their utterances containing references about this workspace, and the availability of 'remote', deictic gestures. This method is based on two findings: 1. participants look at the points they are talking about in their message; 2. their gazes are more dense around these points compared to other random looks in the same timeframe. The algorithm associated the distance between the gazes of the emitter and gazes of the receiver of a message with the probability that the recipient did not understand the message.