Managing disagreement in conversation requires subtle linguistic and pragmatics skills. One key dimension is the degree of `knowingness' with which people present their stance on an issue. It has been hypothesised that framing stances as `knowing', i.e. with higher implied levels of speaker certainty limits the potential for challenge by others. We present the first experimental test of this hypothesis. Using a text based chat-tool paradigm and a debating task we are able to systematically manipulate how `knowing' people's turns appear to one-another. The results show that `knowing' stances tend to close off discussion leading to less carefully formulated, truncated turns, but do not reliably affect the range of solutions considered. Unknowing stances, by contrast, do not affect turn length or formulation but do encourage more deliberation and include more signals of certainty in the message contents.