On Wed, Aug 22, 2012 at 11:46 PM, Gunnar Hellström <[email protected]> wrote: > I found one: > http://seattle.intel-research.net/~davraham/pubs/Avrahami_CSCW_06.pdf > > It says that average message length was just below 30 characters per > message, and average rate was 6 messages per minute in the sessions they > studied.
For a fast typist, that would potentially result in only three or four extra stanzas per message, if you use conservative transmission interval settings (and transmit only during active typing, as recommended). The message stanza with XEP-0085 chat state, can also combine with <rtt/>, and a second message stanza can include both <rtt/> and <body/>, so at least two of those <rtt/> stanzas are potentially non-additive. If lots of extra pre-existing XMPP extensions are used, then <rtt/> wouldn't even be the most stanza-rate-hogging XMPP extension in a chat client. It certainly can potentially be, yes, but it doesn't have to be. This university study that Gunnar quoted, can be cited, if there's no objection. There are multiple implementers concerned about the bandwidth and/or stanza rate of XEP-0301, and I need to include something here as a counterbalance. (e.g. one of them publicly commented -- Darren of Teligent U.K. expressed this concern) Thanks Mark Rejhon
