This paper presents the evaluation of multimedia traffic interaction in terms of induced latency. Experiments are conducted in a distance-leaming platform where long-term burst effects are examined together with video in the same channel. The effects of policy on a user's QoE are examined. If a packet exceeds ITU-T recs. G. 114, it is of no use to a timely play-out device, and should be selectively discarded from forwarding mechanisms. This permits the removal of packets by arbitration mechanisms according to what has happened previously to the stream. The interdependent nature of video packets permits the predictive dependants of intra-coded parents to themselves be discarded, resulting in a temporal compromise rather than an accumulation of errors.