Maeloch
Part of the furniture
- Joined
- Jan 21, 2004
- Messages
- 2,392
- Thread starter
- #31
Well, the 1s resolution of the chat.log is not particularly important when you're running the test over 10mins or so.Chilly said:from the logs I've seen and from the innacuracy of the game logs (max resolution is 1 second) and the complete lack of error correction used in analysis of these data, I can safely say I don't believe a word of any of it.
Until someone does it properly with a couple thousand casts and has a pingplot running so they can be 100% sure of lag issues, there is little point even looking at the results.
A quoted answer without an error is meaningless.
Re: error. You can see it's there, although the reproducibility on the first set of tests is pretty high imo. Dunno if 4 runs is enuff to stick +/- error bars, someone with stats expertise would know that, I don't. I'm not gonna run the test 10 times for each data point or any crap like that, my intent was just to do a quick check, not to come up with a formula at the end.
One priviso tho', the NS test only has 200-300 casts as you run OP pretty soon NS (even grey NS is a 1x lvl spell), so I'd trust those figures less than the dam shield ones.