The service model that a network provides has a profound effect on users expectations. For example, users of the modern phone network expect that their phone call will very rarely be blocked (i.e. not get through), and furthermore that when they get through t o the intended recipient, that the call will practically never be dropped unexpectedly except by the receiver hanging up. In contrast, mobile phone service levels are actually acceptable to users even though call drop out and blocking are both frequent occurences in todays service provisions. In other words, it is very important to consider users' expectations and also the added value of an overall network (who can reach whom, when and where), when considering quality of service requirements.
By contrast, today's typical Internet user expects that they can always start an application, but that the quality of communication has no guarantee, and can vary between almost no throughput and very high delays, right through to perfect communication, with no apparent correlation between the behaviour and their own actions.
This lack of expectation of quality has led to user acceptance of quite low quality audio and video communication, both in base level, but also in high variation of media quality and delay. This acceptance would seem quite surprising when compared with early experiments with video over phone type networks, but is not so surprising given existing Internet users previous experience of the highly variable performance of traditional Internet data (non multimedia) applications.