Fixing Evasion without touching Evasion? Revisiting Acrobatics/Phase Acrobatics

Hey, I edited this in late. Please have a look.

Put dodge in perspective: If I'm riding the line at 50% evade, and I pick up 1% dodge, how much avoidance does that give me?

Avoidance with 50% ev and Acrobatics: 1 - (1-.5)(1-.2) = 60%.
Avoidance with +1% dodge: 1 - (1-.5)(1-.21) = 60.5%.
See that? 1 point spent on dodge is about equivalent to half a percent chance to evade at the time when you pick it up. Too weak. The circle just needs more.

-----

Frankly, I'd like to see dodge subtract directly from the chance to hit formula, the same way endurance adds directly to reduction. If this happened you'd have to lower the dodge value of acrobatics, to something around 10%, maybe a tiny bit more. That'd do SO MUCH for evasion scaling.

I'm going to go on a tangent here about how additive dodge (rather than multiplicative) would affect rating scaling. Is that too far off track? It'll have an amazing effect.

I'll have to think about losing block. I feel like you're shooting yourself in the foot, though: it caters to 2h builds and bow builds. Bow builds already get by with acrobatics, and pure evasion using GS with a maul just seems kind of ridiculous :b.

Additive dodge is a huge can of worms!

Also, dodge was thrown in to interact with spells, I think -- evasion can't interact with spells because spells have no accuracy rating. That's also why I'd like to consolidate Acrobatics and Phase Acrobatics: Acrobatics feels like a half-measure to Phase Acrobatics' full.

We agree that the 4% dodge is insufficient. Good. I did say bump them by 1-2% each if need be.

Regarding the block removal: you think that by removing block chance, we're actually inviting bow users and 2handed users (typically armour types) to give up both ES and armour?

To turn away from Iron Reflexes?

For that extra life?

I'm not sure if I'm reading you right there. The cannot block merely aligns with what happens when you become so good at evading you don't need to block. In a way, I'm asking you to concoct a justification for that.

Would additive dodge be it? I figure additive dodge rather than block chance wouldn't make any difference, since you're losing that 31% added chance to avoid damage.

Then again, I get this from Malice:

"When an attack is blocked, the game first calculates if the attack would have caused a stun were it not blocked. If it would have caused a stun, the blocking animation is played, stunning you briefly. If it would not have caused a stun, then you get a "free" block with no animation. Faster Block and Stun Recovery and Increased Block Recovery modifiers reduce the length of the blocking animation."

How would no-blocking, extra dodging interact with that?

Um, you wind up getting stunned less often. I think.

I realize additive dodge is complicated, but they did it for armour with endurance. It changes effective health scaling completely, but if done right it'd only actually raise total avoidance by a few percent. You could make things behave such that evasion rating provides linear returns for someone that wants to give up all other defenses, but still keeps its gross over-scaling at low values.

As to interacting with spells, well, its additive vs multiplicative nature shouldn't matter much in that arena. There's nothing else there for doge to compete with in that regard.

I REALLY like this idea. At least give me time to put something more meaningful together. It'll take a few hours.

Btw, I still have a really tough time chewing on the loss of block. It just completely invalidates too much of the skill tree. Turns your shield into a stat stick. Every dual wield cluster and shield cluster in the ranger/shadow/duelist area would need to be re-organized to allows players an easy way to avoid block nodes while still picking up other dps/defensive nodes.

Having more life solves some of your stun problems, if that's what you're after.

Eh, we can drop that idea totally. You're the math person, I just think in themes and concepts. Between the two of us (with you pulling 90% of the weight, I'm sure), I think we can nut this one out nicely.

My 10% contribution would be the consolidation of A/PA and making your idea into a keystone after X number of bridging nodes.

But with the idea of additive dodge, wouldn't that balance out the loss of block *and* make you less susceptible to stuns? That sounds like a fair trade-off to me.

post garbled, must figure out why.

Okay, here's the story.

There's a big danger with making dodge an additive system. Mostly, the accuracy-evasion system has to always favor allowing a low-accuracy attacker to hit a high-evasion defender. If you threaten to mix that up, evasion would suddenly become stupidly overpowered when used by a clever min-maxer. Especially in PvP.

So what we do is look at evasion as a system at the absolute max possible evasion ratings.

For this first exercise I went shieldless, and assumed 193% evasion rating. That's the whole evasion circle and all pure evasion at the start of the ranger area. I also assumed 300 dexterity - that has some slop for extremely good gear, and a 19% evasion rating amulet. Clearly higher than anyone will ever get without a shield. This whole exercise has to be repeated with a shield to make sure nothing breaks in that case.

Buckle in, guys. This is super-wall-of-text mode.

First, we need to know the rating maxes associated with this build. How much evasion could you get if you went completely overboard?

I assumed top-tier items, each with 100% incr evasion rating and +322 evasion. 150%/322 has to be looked at also, but for now I'll stick with this. This character uses healing flasks with 100% reflexes on them. Spreadsheet with the result is here. With this build and my stated gearing priorities, max possible evasion rating is around 36k with that reflexes flask running.

Now pull out the handy computer algebra system, and start crunching numbers. OK, BBcode is having trouble with the CAS code, so that gets left out for now. Damn, that's annoying. This post is all about the math, and I can't post the math.

--- will try to add code in here ---

This is just code for calculating and plotting all the things. We wade through it a bit at a time.

First, we need to know what kind of max avoidance we'll get with absolutely amazing gear, with and without dodge. Maxima says that, at 36k evasion rating against a level 65 mob (570 acc), we will have:
71.9% chance to evade
78.6% total avoidance with 24% dodge, in its current multiplicative state.

Now, to switch to additive dodge and not break the system, we can't add very many percentage points of total avoidance. So, we have to get rid of the multiplicative dodge and subtract a flat dodge% from chance to hit instead, making sure total avoidance doesn't go up by more than a few percentage points with very top-end gear.

The number I settled on for additive dodge was 10%, including the whole circle. Acrobatics would be 6%, each dodge node would be 1%. This seems fair and adequate at the same time. Let me show you why.

1) total avoidance under all 3 conditions



By choosing 10% additive dodge, I stole a little bit of avoidance from the early game and gave it back at the very late endgame. At the very top of the gear curve, total avoidance is about 4% higher than under the current system. Doesn't seem like a massive buff, but I have a prediction: this will completely fix evasion scaling. Watch.

2) effective health under all 3 conditions

The current problem with evasion is this: the more evasion you stack, the less it becomes worth in terms of effective health (or effective healing, whichever you prefer). Armour as a system has endurance to change this fact, changing behavior from linear to upward hyperbolic. Additive dodge does the same thing, even if the percent added is very small like we're doing here:



Again, stole a little bit of EH from the early game, but the overall problem with evasion rating is solved. At the end of the curve it provides increasing gains on effective health.

3) EH per rating point under all 3 conditions

Now, this is just the point-value under all 3 systems. No dodge, current dodge, 10% additive dodge.



If you look closely at this, it means that under a 10% additive dodge modifier, evasion rating surpasses its old value by a factor of about 2.5 in this system. > 15k rating? Evasion rating with 10% additive dodge = 2.5 * evasion rating with 24% multiplicative dodge. And it keeps going up. While before, evasion rating scales down into nothingness.

4) Minor tweak

As I've explained, this modification steals a little bit of effective health from the A/PA ranger in the early to mid game (<15k evasion rating or so) but fixes evasion scaling with super high end gear and endgame builds. Coincidentally, at its worst that loss is about a 12% effective health loss.

I still think pure evasion needs more life. The two ideas go hand in and, and the more life idea counters the early game loss that this idea here introduces.

Some adjustment has to be made to compensate early game. I've already suggested some kind of unique life bonus inside the A/PA circle. In fact, any other kind of unique defensive bonus could cover that loss. We've already had that discusion; this idea does not preclude that one.

The total dodge bonus could be raised to 12% or so without breaking anything, I think. It'd make the increasing gains on evasion a bit more obvious. It also lessens the impact on early game. Pulls the crossover back to about 10k evasion, and about doubles the strength of dodge wearing perfect gear.

Oh, and I figured out your objection about dodging spells. You could do 12% dodge vs attacks, 24% dodge vs spells. Nothing preventing that except developer time.

I'm going to put all the same graphs in this post, at 12% additive dodge. Come back later and see them if you want. The difference is pretty profound. Adding much more than 12% dodge seems kind of dangerous.

I'm altering my proposal to 12% additive dodge / 10% more life. Graphs soon.

I am amazed at how much of that I understood. Thank you for your effort, Z. You're a legend.

When you're done calculating, I'll see if can boil this down to a clean modification/alteration of the existing 6 node structure.

I already have 2 configurations in mind.

You're gonna have to give me until later tonight to work it out.