See picture in the link...

http://i.imgur.com/DpxsT.jpg

oh realize the pic doesn't show country so to help you out a bit...it's Express country #153. Created it to test some stuff.

There's 11 hours left in Express before the server gets wiped.

Me/oldman were SSing each other to test the break formulas and get some gaussian data.

http://www.eestats.com/express/country/153

teamwork on solo server? DELETE??? :P

testing the pang factor?

I have tested this out up to I think a 500 million jet PS down to the exact break and have determined that if you account for the 10% you will never fail. Not with a dict though, now that I think of it though.

Well, we concluded that if you undersend by 1-2%, you still have a slim chance of breaking. About 1 in 8 hits, give or take a bit.

So, it does seem like the gaussian distribution is working fine, but possibly not centered on the exact break. If it is, we got unlucky with only 3 hits breaking out of 16 hits on undersending 1% as opposed to 6 hits breaking out of 10 when we oversend by 1%.

Sample size is very small, but still...

Realistically, this is just mathematical knowledge and doesn't really give anyone an advantage, most people will still oversend by 10% (and not 1%).

Maybe qzjul or pang can give some insight.

hm that's nifty; i don't know how that would happen, probably some rounding somewhere; I can put in a check for that i guess