Hi all,
Thanks for the feedback thus far, and please keep it coming. Your experiences are very valuable in calibrating the experiments we’re doing and helping us narrow in on the optimal settings for the latency compensation system.
A quick clarification on ping times and latency compensation
I should have explained this clearer initially, so I apologize for the confusion. Ping time is not affected by the latency compensator and there is no correlation between ping time and the experiments we’re doing.
Traditionally, ping is a measure of how long it takes a message to get from your computer to the servers. In GW2, we actually combine this data with a second statistic, which is a guess at how long it will take the server to actually process and respond to your message. So ping measures two things for us: network health and server responsiveness.
A longer ping time in GW2 terms means either your network/Internet connection is slow or losing message, or the server is busy and can’t respond quickly (or both at once). However, the system that monitors ping is completely outside the system that handles latency compensation, so any adjustments we make to the compensator will have no measurable impact on ping times.
There is one way that ping and latency compensation interact, however. If ping time is steady, the LC experiment should make the connection feel a bit smoother. However, if ping time is unsteady or very high, there will be no noticeable improvement, because the message response time will be higher regardless.
Suppose you start with a ping of 100. Our adjustments to the latency compensator should make that ping “feel” more responsive. If your ping goes up to 200, however, the difference made will sort of be swallowed up by the doubling in ping time. If it spikes to 500, as an example, the half-second delay will be impossible to totally hide from view, and the connection will feel worse.
You can think of it like a teaspoon of sugar. If you add the sugar (latency experiments) to a small glass of water (good pings) you get sweet water. If you add the sugar to a swimming pool (bad pings) you won’t even notice the sugar. Also, drinking a swimming pool is bad for you, don’t try it ;-)
It is in some ways unfortunate that so many players are having ping time spikes during this experiment, because it makes it hard to tell if the fine-tuning is working. However, based on our data, it looks like there is an improvement, it’s just subtle, and again it gets swallowed up by the ping problems.
We’re examining our server responsiveness constantly, and there are separate efforts underway to improve that.
The latency experiments are carefully designed to control for ping time spikes and other complications. Suffice it to say it is tricky to sort out the real effects of these experiments, but they are working and making a positive difference. In the spirit of doing good science, I unfortunately can’t reveal all of our testing methods, without risk of introducing bias into the data we collect. So I thank you all for your responses, whether you’ve been benefiting from the experiments or otherwise.
Please keep it coming!