Fixing Evasion without touching Evasion? Revisiting Acrobatics/Phase Acrobatics

Given that it essentially drops life early on, but is enough of a buff later, why not add a simple node with +30 Life, +12% life, Like the Templar has?

Makes a big difference early on, stacks with all the % nodes, is comparably weak yet still noticeable on higher levels; aAnd it doesn't need a new keystone, either.

As for spells, why change the 20%? There's nothing to add it to, anyways.^^

My question is: Armor is effectively normal damage frequency with smaller values and evasion is a lower frequency with higher values.

This necessitates evasion have more hp in order to absorb large damage streaks... how much more?

For a sub-optimal (read: normal) setup with perhaps 40-60% total avoidance, what level of HP is necessary to have similar survivability to an armor build, both going melee?

For a normal character with 50% damage reduction via armor, what level of total avoidance would a normal evasion build have at the same level and gear quality?

I think that any serious discussion about fixing evasion needs to start with some sort of statistical analysis of hits until death and the probability such a streak will occur.

I do appreciate the above posted math options though. They can show how small perks can change the formula for long term mathematical stability.

Here's my contention with that theory, zeto:

Comparing armour and evasion in precisely the same situations isn't a proper way of evaluating evasion itself. If you want to melee and stand toe-to-toe, you should use armour. Whether it's natural armour or IR, that's just how you should go.

When I say 'true evasion', what I really mean is using evasion to supplement a playstyle appropriate to evasion. Whether that's ranged, using whirling blades to keep moving or any other form of non-tank melee combat, it's still different to the stand-there-in-the-thick-of-it-and-pound-away style of armour-based defence.

This is also why I'm not 100% zealous about this idea: maybe the changes to the life nodes and the increase in evasion nodes will suffice. I've yet to test, but I'm getting there.

I suppose I'm trying to argue this from the standpoint of people who still feel evasion is broken, and perhaps such people do believe that evasion alone should be enough to play the same was an armour user. I happen to disagree.

I'm not sure which side would be 'right' though. Maybe GGG intends for anyone wanting to go tanky to need either a lot of armour, a lot of life or a lot of ES. I realise that armour users generally get a lot of life anyway...which is where Zakaluka's input becomes invaluable.

I have a feeling the next question would be would tanky types sacrifice armour and energy shield for such a boost to life anyway? Factor in not only life nods but strength nodes and regen. There's room there for an absolutely killer marauder build.

@ charan, regarding your most recent point, this won't make evasion "tanky", like my armour templar is tanky. Not by a long shot. I've put a very mild increasing return on evasion rating by carefully choosing between 10% and 12% dodge, with other carefully chosen life bonuses.

If you saw the armour-EH curve I posted a while back, the increasing gains are PRONOUNCED. All I've done here is give evasion a tiny nudge in that direction, so there's more of a reason to keep picking up rating instead of life late game.

--------

I've got two approaches here, and they're both calculated fairly precisely. IE, several reasons for each why it costs exactly this many points.

Both approaches eliminate the early-game loss of additive dodge by around 6k or so evasion rating on the character screen. The EHP loss from multiplicative to additive dodge should be gone by the time you hit maps, and it's so minor that it should be unnoticeable earlier than that. That seems good.

Summary: Change dodge to an additive bonus; add some amount of (carefully planned) life bonuses to the A/PA circle. Doing these two things together changes evasion from having diminishing returns on EH, to having increasing returns.

Without at least some small life bonus as part of the A/PA cluster, this change hurts scaling in the early game. That's why both ideas survived with life bonuses remaining in the acrobatics circle.

EH curves, from now and under each proposed system:



These curves are plotted against a hypothetical build with 140% increased life from passives.

-----------------
Approach A: 12% additive dodge, 8% incr. life, +30 life

This solution feels much nicer. Hastily done sketch:



In this context, you get the life for free. After you've spent the 6 points you would have spent anyway on the A/PA cluster, it comes with "uncanny resilience", the new keystone Charan suggested. Minus the blocking thing ^^

I like this idea better, but it can add up to 5.25% total avoidance by the top of the gear curve, maybe slightly more in a shield build. That might be problematic? We're looking at being able to actually approach 80-85% avoidance from just evasion and dodge alone. What problems does that cause?

Anyway, this approach feels more evade-y. Right that the EH curve is slightly steeper, relying more on evasion and less on life.

Approach B: 10% additive dodge, 32% increased life, +40 life

This is probably the more awkward implementation, but it adds less total avoidance to the top of the gear curve. This one only adds about 3.25% at max gear for the build I looked at.

It's actually less powerful than the other option even given the bigger life bonus. That's because it costs more points. I don't have an illustration for this one. Only change the following things:
- Acrobatics becomes 6%, dodge nodes are 1% each.
- Rephrase Phase Acrobatics: "Spell damage can be dodged. 2% additional, and double chance to dodge spells." Or maybe just to make it less convoluted, just say triple chance :)
- Add two new nodes at the end of the circle (connecting north from the last dodge node) that read: "16% increased life. +20 life.".

The player grabs the two life nodes in place of two 8% incr. life nodes he otherwise would have. If you change the layout by making the cluster smaller, those two points become too powerful.

The plot at the top accounts for the fact that option A still gets to buy two 8% incr. life nodes, which this one had to give up. You can see that their EH curves still track each other very closely.

Now, to why you might want to do it this way, even though the other option is more elegant: this approach adds less avoidance. In my hypothetical build, at max possible gear, this approach only raises avoidance by 3.25%. Raising avoidance too far is risky. That's why this proposal stays here.

-----------------------

Either way, the end goal is accomplished. Taking A/PA and dodge improves your defenses significantly, and also improves evasion scaling, giving evasion rating slightly increasing returns the moment you complete the dodge circle. I like option A.

To wrap it up, and show development of this idea, here's the progression on EH-per-rating, again under all 3 systems:



Charan, thanks for all the help and ideas. Sorry I kind of took over at the end there.

If I didn't want you taking over, Zakaluka, I wouldn't invoke your name. This is what you do, and you do it very well.

I anticipate GGG's response...

So, there's something confusing about all that. I went back and looked at replies after posting.

When I plotted curves for additive dodge, I ran into a scaling problem at low level. Low level characters take a sizeable survivability hit, and I needed to find a way to solve that. Keeping the original idea for multiplicative life solved the problem with early game scaling.

But the cyborg said, "hey, why don't you just use a discipline and training node?"

Duh, that makes way more sense now. Okay. Everywhere I said "8% more life" you can just put in "12% increased life/+30 life". Everywhere I said "10% more life" you can just put in "16% increased life/+20 life". Scaling problem solved.

There was more to it than the low level scaling issue, though. Part of the original idea was to add some sort of significant life bonus to the A/PA circle.

I think I edited the last post to reflect all that, but I'm a bit tired.

I'm fairly sure GGG are going to keep A and PA separate, just try to make it all a bit more lucrative.

So what you might see is the current setup, but with a discipline and training node branching off one of the +1% dodge nodes.

Would that suffice?

Okay, DODGY PHOTOSHOPPING ALERT!

Um, yeah, I just did the combination thing because you suggested it and it front-loaded some more benefits into acrobatics.

No biggie.

I'm not sure though if only one is enough, given that placement. It screws up the skillpoint cost of the whole thing, and you land between option A and option B (a less powerful mix of both). Might bring back the low-level scaling issue.

If it's just one D+T node, the whole circle has to cost the same number of points it does now. Maybe give Acrobatics 9% dodge, and have only 3 dodge nodes. Then branch a D+T node off that whole mess.

See above. So dodgy. But fun to do.

Still insufficient? I'd be happy with that change. The extra 1% node is just to complete the circle, really.

Oh gosh, I'm tired. The life bonuses are really important to early game, but the dodge bonuses are really great at endgame. It might be fine. I'm too tired to put the numbers in and go around the circle again, though.

My first reaction was "it needs more life, or the whole thing needs to cost 6 skillpoints again." But if you're adding another dodge node, that's really attractive to me. I like endgame bonuses.