Re:GTA 5 breaks entertainment records but Online woes continue
by Negitivefrags on Wednesday October 09, 2013 @05:12PM
(#6523906)
We basically attempted to get the bots to match each metric that we were measuring server side from a smaller sample of players during closed beta. The peak concurrency during closed beta was something like 2000 players, so we multiplied each metric out to get a target. Here are a few examples of metrics:
Chat messages per second
Party creations per second
Average party size
Number of zone transitions per second
Average number of open zones per player
Character saves per second
PVP Match list requests per second
And so on, and so on.
Certainly bots are not going to be the same due to the issues you mentioned, but we do the best we can! We didn't simulate all kinds of connection issues a real player might have, but we did match the non-graceful connection termination metric.
This isn't my line of work but it would seem simulating 50,000 bots in an environment is not quite the same as 50,000 humans from all over the place with their various connections speeds and/or issues.
Plus, you'd have to teach your bots to try to do incredibly stupid things that humans tend to do, willingly or otherwise.