http://blogs.msdn.com/shawnhar/archive/2007/12/14/network-latency.aspx"The time has come", the Walrus said, "to talk of networking "... The network game programmer has three mortal enemies: Latency makes your data arrive late Packet loss makes some data not arrive at all Bandwidth limits how much data you can send I shallenCommunityServer 2.1 SP1 (Build: 61025.2)http://blogs.msdn.com/shawnhar/archive/2007/12/14/network-latency.aspx#6789005Mon, 17 Dec 2007 15:15:26 GMT91d46819-8472-40ad-a661-2c78acb4018c:6789005Ultrahead<p>Talking about Wikipedia:</p> <p><a rel="nofollow" target="_new" href="http://en.wikipedia.org/wiki/Dead_reckoning">http://en.wikipedia.org/wiki/Dead_reckoning</a></p> http://blogs.msdn.com/shawnhar/archive/2007/12/14/network-latency.aspx#6789287Mon, 17 Dec 2007 15:49:48 GMT91d46819-8472-40ad-a661-2c78acb4018c:6789287Ultrahead<p>So, if each frame gets 16 milliseconds in a 60 fps game and Xbox games are expected to work with latencies up to 200 milliseconds, that is to say that it's expected -in average- that every 12-13 frames a &quot;jerk&quot; could eventually happen, what in turn is the amount of frames that the human eye can perceive.</p> <p>Taking 200ms as the worst case scenario, we would be rendering 12 to 13 frames based on our predictions and assuming no changes in the oponents' behaviors.</p> <p>Thus, the question is how much of those 12-13 frames we use to adjust the wrong values to the correct data by using interpolation in a smooth way.</p> <p>This is a very interesting topic, so cannot wait to get that sample!</p>