Hey guys, just somethings you might like to consider from a mathematicians perspective:

Flat Hp works on laws of average because your HP is a fixed rate, and the hits is almost flat as well (i'm assuming we aren't going to talk about critical chance and damage, because my math isn't up for it.) so lets take one of these situations:

you are forgetting one important thing though: dodge is a chance, so you can't just look use averages. "Law of large numbers" states that experimental values (the values recorded) only approach theoretical values as the repeat of the experiment approaches infinity. 15 hits is not infinity.

Instead lets approach it from a statistical point of view and recognize dodge as what it is: a binomial probability. 6.75% dodge is a probability of success. Lets set a fixed number of trials at 6, and any successes k>0 means survival. I use a ti-85 "binompdf(6, .0675, 0)" to check for chance of death = .6575%. What does that mean?!?!

Only in 66% percent of the time does having

__only__ 6% dodge kill you. that means 34% of the time does a measly 6% dodge result in one OR more dodges saves your life. WOOOH!

Wait, lets change the scenario and make the dodge even better. Lets grab a Ninja Tabi and estimate your dodge chance at 1 in 5, 20%. So now use binompdf(6, 0.20, 0) = 26% of you never dodging. Which means in 74% of the possible cases you survive the 15 hit encounter. And in 34% of the cases (binomcdf (6, 0.2 , 1) ) you dodge more then one hit, and can actually win the fight. Crazy huh?

Now the fact is this is a statistical chance game. If you get ganked in the jungle and you die in fifteen hits, screaming "OMG my binomial statistical chance of survival was over 70%" isn't going to get you anywhere, where as HP might have saved you.

The point is, the small dodge chance is actually more then you think it is. So, where as flat HP gives diminishing returns over time, dodge actually holds up real well. If you guys have any questions on the math, I can explain it more