Much research concerning asynchronous online discussion reports quantitative data such as frequency of posting, time online, and number of characters or words. To effectively understand online discourse within education environments we need to go beyond such measures and study the interactions between messages and how these interactions contribute to the construction of collective knowledge. Complexity science suggests that the emergence of new group understandings requires both redundancy (agreement) and divergence (disagreement) in the interactions between contributed ideas. Studies, focussing on graduate course asynchronous discussions spaced almost 20 years apart, and employing a coding system developed from Fisher’s interact system model (ISM) have shown a lack of messages expressing explicit disagreement. Online discussions, with their limited channels for mutual social support, appear to encourage a student tendency towards being "too nice". Course participants camouflage disagreements in ambiguous postings that, while avoiding potential offence, do not effectively contribute to the progress of debate.