# Thread: AC, Dodge, Concealment, Incorporeality and you

1. Originally Posted by Exiledtyrant
If people are walking into end game dungeons with 100-160 ac and only missing 2 hits why do the even bother giving us so many cheap enhancements to stack it? Honestly that makes the earth savant natural AC boost seem like a total joke now. Leveling with 55-61% miss chance has been great for me but if it just becomes a garbage stat down the line I'll just drop it. my sorc had a potential 80 AC at 20 and about 100 ac at 25 but that a lot of sacrifice for no returns it seems.

What the heck do people stack for defense in this game then. Physical damage reduction usually seems harder to acquire past 50 if your not a plate/shield class. So what's left dodge and health? Displacement makes it a 50/50 shoot either way but it was nice having another 50% miss buffer over that .
Concealment, incorporeal and mostly simply moving the character out of the way of incoming attacks (or more to the point, constantly be moving as it's a lot more likely to be moving out of where a mobs attack is being aimed than it is to be moving into one).

2. Originally Posted by Pilgrim1
Oy, I thought I would bring this question up:

Does it matter what order these defenses apply?

For example if you have 20% blur (B), 10% incorpral (A), 25% dodge (D), and say 35% armorclass (C).

expected hit chance with B>A>D>C out of 100 hits:
.8*.9*.75*.65 = hit 35 out of 100 times...

changing the order: C>D>A>B with 100 hits:
0.65*0.75*0.9*0.8 = hit 35 our of 100 times.

Basically what I'm trying to say is that since we multiply our different layers of defense the order should not mathematically madder.
Overall, no. But it is important when trying to determine the effectiveness of each individual element. Though as the effects of 3 of them should be self apparent, only the effectiveness of AC should be in question.

3. Originally Posted by Therigar
Ran the second test of armor class effectiveness in epic Gianthold against CR26 monsters. I took advice from others and removed concealment and incorporeal items from the test character. This let me limit the protection to dodge and armor class only.

In 102 attacks, 23 were prevented by dodge and 51 by armor class. The character ran with 23% dodge and 67% "at level" armor class. The number of misses matches almost exactly with expectations if the order in which defenses are checked is dodge first and armor class last.
It is possible that it isn't this simple though. It could very well be that an attack roll is generated first, with rolls of 1 (or the equivalent)* stopping the process there, while rolls of 2-19 (or the equivalent) go through the whole list of checks and rolls of 20 (or the equivalent) going through the whole list except AC.

*Added since it's been noted that the current system doesn't actually use a d20 roll to determine results.

4. Originally Posted by SirValentine
You mean...in the wilderness? On Normal-difficulty monsters?
You know, I understand people being negative. It shows that they do not understand what is going on with these tests.

Before we can know whether armor provides any measurable benefit in eElite content we first have to know how the defense system works. This means knowing what the order of checks is and whether the Turbine provided "to-hit" formula is actually being used.

Originally Posted by RedHost
The order of operations is entirely irrelevant.
The order of operations is relevant because it affects the actual number of times that a miss occurs for a particular reason.

Originally Posted by Gremmlynn
This tells me that AC is checked first as, if the others are checked first, there would be no need to even make an attack roll.
I hear you, but I am certain that you are wrong.

Originally Posted by Gremmlynn
It is possible that it isn't this simple though. It could very well be that an attack roll is generated first, with rolls of 1 (or the equivalent)* stopping the process there, while rolls of 2-19 (or the equivalent) go through the whole list of checks and rolls of 20 (or the equivalent) going through the whole list except AC.

*Added since it's been noted that the current system doesn't actually use a d20 roll to determine results.
I think that the code is something very similar to this:

If roll=1 then miss
else if roll=20 then hit
else if dodge then miss
else if incorporeal then miss
else if concealment then miss
else if armor then miss
else hit.

My limited testing bears this out although with a glaring anomaly for concealment (the returned value was exactly double what I expected*). Further testing will tell me if that is just a single event or whether concealment is actually being doubled.

What I am absolutely sure of, however, is that dodge gets checked first and armor gets checked last. I'm willing to let the sequence of incorporeality and concealment remain moot since the effect is the same on determining expected armor misses.

*Edit: As mentioned in edits throughout the thread, it turns out that the concealment item I had on in the first test was 20% and not 10%. This means that my results were exactly what should be expected and that the order of checks is dodge, incorporeal, concealment, armor.

5. Originally Posted by Gremmlynn
Overall, no. But it is important when trying to determine the effectiveness of each individual element. Though as the effects of 3 of them should be self apparent, only the effectiveness of AC should be in question.
Exactly.

And here is why the order matters.

In the second test in which there is only dodge and armor, if armor is checked first then we expect the number of armor misses to be 102*.67=68.34. We would expect 68 misses but we record only 51. Likewise the number of dodge misses we expect in this situation is 102*.33*.23=7.7418. We would expect 7 or 8 dodges but the recorded number of dodges is 23.

OTOH, if dodge is checked first the expected result is 102*.23=23.46. We expect 23 dodges and we recorded 23 dodges. The number of armor misses we expect is 102*.77*.67=52.6718. We expect 52 or 53 armor misses and we recorded 51 -- very close to expectations.

Since this correlates with the observations from the first test, I'm entirely confident that this is the sequence --> dodge, (concealment & incorporeality), armor class.

6. I have been trying to avoid more sophisticated maths because it is harder for both me and others -- unless they are mathematicians.

The basic premise is that, on average, results will trend towards the expected values. When you study statistics you learn that they normally do not match expected values -- they are higher or lower. This is used to determine standard deviation.

To be absolutely precise we would normally presume 1 as a miss and 20 as a hit. We would then preload our miss and hit bins giving each 1 in 20 events. Now we would apply the defensive chances against the remaining attacks.

In the example of 102 attacks at 23% dodge and 67% armor this would lead us to assign 5 misses and 5 hits automatically and to reduce the observed misses and hits by 5 each.

I could then rework all the numbers.

That is a bit of work and faces a problem -- what if the to-hit isn't really a d20?

In point of fact, we know that it is not. I don't recall the post, but (IIRC) the actual process involves d100 and rolls of 1 and 20 (1-5 and 96-100) are not actually auto misses and auto hits.

So, while it is a kludge to just take raw values it is near enough, IMO.

In fact, although I posted this:

If roll=1 then miss
else if roll=20 then hit
else if dodge then miss
else if incorporeal then miss
else if concealment then miss
else if armor then miss
else hit.

I think that it would be easier to code it like this:

If dodge then miss
else if incorporeal then miss
else if concealment then miss
else if armor then miss
else hit.

The results are close enough to the d20 code that it falls inside standard deviation so is statistically equivalent. And it involves fewer processes so runs cleaner.

And, from a player's perspective, the slight variation will not be noticed over time.

7. Originally Posted by Tanngiostr
I believe AC is not and has not been useful in EE content for a good long while...

I'll stick to other forms of damage mitigation that I know actually function.
I think that most of those posting negatively about armor feel as you do. I think that their assumptions are incorrect. I do think that understanding the balance point for when armor makes some difference in eElite lets the whole player base make better choices about whether to build for AC.

Let's presume for the moment that at 112AC my character's armor is too low for CR60 monsters. Can I determine what the attack bonus is for that monster by testing and can I then estimate where the effective AC will be?

Would that be useful information?

I think it would be.

8. ## I'm so stupid

Just in case you are jumping to the end of the thread without going back through it to see my edits, I have to confess to being incredibly stupid.

In starting this thread I reported on a test (available in video, look above in the thread for the link) in which I tested dodge, incorporeality, concealment and armor in Epic Gianthold against CR25/26 monsters.

In that test I reported my concealment number as 10%. And I noted that there was an anomaly as my concealment misses matched a 20% concealment factor.

Well, it turns out that I made a mistake. I actually had on a 20% concealment item: Greater Nimble Trinket.

What is important about this is that I can now say with a fairly high degree of confidence that the way defenses are being applied is dodge, incorporeality, concealment and then armor.

And, because both tests match so closely to expected results without adjusting for auto hits and auto misses I am equally confident that this means there are no "miss on 1" or "hit on 20" events.

And, what that means is that the 2 misses in the eElite Cabal test are due to armor.

I'll come back to this point.

9. Originally Posted by Therigar
Would that be useful information?
No

Originally Posted by Therigar
I think it would be.
Wrong.

10. ## Summary of Results

1: Dodge is checked first
2: AC is checked after Dodge
3: Concealment and Incorpreality are checked after Dodge and probably before AC
4: Concealment and Incorpreality are checked in unknown order in relation to eachother
5: The Concealment statistic in the tests is stated wrong in some places. Which effects were used for Concealment and Incorpreality would make mistakes like that easier to detect on reading
6: If named items were worn hidden effects may have been present so all items worn should be posted.
7: Denying automatic hits and misses from such a small sample size is not possible

11. Originally Posted by Therigar
. I think that their assumptions are incorrect. I do think that understanding the balance point for when armor makes some difference in eElite lets the whole player base make better choices about whether to build for AC..
as I said before if you want to prove AC is useful at all then Build for ONLY AC and do the testing otherwise your testing does nothing to support AC

12. I remember that a 5% minimum miss chance was stated by the devs to exist in the enemy attack formula along with a 5% minimum hit chance. I think this was on MotU closed beta forums. The devs also said that the enemy attack formula was not regularized to a d20 but player character attack rolls were. The 5% minimum hit chance is possible to test in low level quests on a high AC character

@The_Troll If we could figure out the enemy attack formula or find the original dev post on it, it would be possible to determine the enemy attack bonus. Further if we could determine the enemy attack bonus we could determine the defense chance a given AC would have against that attack bonus and therefore determine an effective AC. Your previous post does not inform readers of what Therigar is saying that you are responding to. It also does not add to the discussion in any way other than stating that you disagree with him. I disagree with you that it is not useful information. In fact I believe that if we could discern enemy attack bonuses it would be extremely valuable to build planners and any player who runs content where enemies use physical attacks. I agree with you that it is probably extremely difficult to calculate the required numbers but most of that difficulty would be in the data collection step. I believe that if we are able to determine effective ACs against enemy attack we well not be able to build the calculated effective AC for EE enemies.

13. Originally Posted by Book_O_Dragons
I remember that a 5% minimum miss chance was stated by the devs to exist in the enemy attack formula along with a 5% minimum hit chance. I think this was on MotU closed beta forums.
This could be right. At the moment the limited testing does not support this. It could be because of your later observation that monster to-hit isn't regularized back to a d20. It could be that the sample size is simply too small yet.

I did find this interesting. One of the early runs when I was testing out how to record video is a run into eElite Cabal with the full gamut of dodge, concealment and incorporeality. In 49 attacks there are 4 that are misses for reasons other than dodge, concealment or incorporeality. This is roughly twice the expectation even if there are auto misses as in 49 attacks we would only expect 2.45 misses.

This seems to support the point of view that armor is the cause of misses in eElite content and that monsters are not hitting on 2+ as is routinely claimed.

Again, the number is too small to be definitive and any conclusion drawn is equally valid.

I'm currently working up a cleric to L20 on a second account. I'll be able to hand it over to my son and have him run healer on me to test this a bit further. But, the character is currently L16 so will need a few days to reach 20.

14. Originally Posted by Book_O_Dragons
1: Dodge is checked first
2: AC is checked after Dodge
3: Concealment and Incorpreality are checked after Dodge and probably before AC
4: Concealment and Incorpreality are checked in unknown order in relation to eachother
5: The Concealment statistic in the tests is stated wrong in some places. Which effects were used for Concealment and Incorpreality would make mistakes like that easier to detect on reading
6: If named items were worn hidden effects may have been present so all items worn should be posted.
7: Denying automatic hits and misses from such a small sample size is not possible
Good points and summary. I am fairly confident that the order is dodge, incorporeality, concealment, armor. But the number of samples is small (only 1 with all 4 defense elements) so definitely not conclusive.

The sample size really isn't that small. I've recorded over 200 attacks and the results don't support auto miss/auto hit.

I'm leaning towards the no incorporeality and no concealment testing at the moment to get a better picture of whether armor makes any difference in eElite.

15. I did some testing awhile back (when citw was still the only "endgame" raid is the most accurate time estimate I've got) on the order in which incoming attacks checked defenses. Dodge is checked first, AC is checked last, incorp and concealment are checked in between but couldn't determine the which was checked first, even after 1000 hits. I'll have to do another batch of numbers over the next few days to provide evidence, though, as I can't seem to find the notebook that data is in.

I logged about 550 attacks from the two melee hobgoblins in the start of EE cabal. I'm old fashioned and that data is written down in a notebook, rather then plugged into an excel file. Since this is just preliminary data, and I'm ultimately doing this to satisfy my own curiosity, I'm not spending the effort to screenshot everything or take a bunch of videos. If you don't want to believe me, then that's fine by me. Here's the data:

181 AC, 10% incorp, 11% dodge, no concealment.
126 attacks from the clovenjaw warlord. 13 incorporeals, 15 dodges, 48 misses, 50 hits.
436 attacks from the clovenjaw veteran. 43 incorporeals, 51 dodges, 166 misses, 176 hits.

Seems fairly obvious that AC is working to some extent. If we assume AC is checked last, we're talking just shy of a 50% miss chance. If it were checked first, it becomes ~40% miss chance (but dodge and incorp numbers support AC being checked last). Either way, it's more then enough to rule out missing just on a 1.

Assuming AC is checked last suggests a to-hit of ~170 in both cases, but I won't trust that number until I gather data at other AC values to see if it matches up, as I don't trust turbines to hit formula (said formula is posted on the wiki page for armor class, if people don't remember it).

I'll be sharing a bunch more numbers in a couple of weeks, once I've had enough time to gather enough data to satisfy me.

16. While I fully support anyone who tries to deeply understand the mechanics of the gameI like and shares his knowledge, I have several questions regarding stuff in your conclusions and data that I dont understand:

1. Why do you actually care, what works first? I seriously dont understand why? If you have 25 dodge, 30 incorp, 50 displacement, it simply doesnt matter in which order do they work. The endresult is always the same. And since all these are fixed chances of damage avoidance, you can use simple math to find out the chance of your current AC against your current foe.

2. While getting no joy from supporting teh-troll over ya Therigar, I do agree that AC is useless in epichard and epicelite. Defense chance is At level and levels of mobs in EH and EE are so much above character level, that if you are getting hit 94% of the times with only AC on... then CG on that 1% that is amybe coming from uberboosted AC. And the fact that it works on Epic normal... well... ya know... on epic normal, I dont have to care about it in the first place.

That said, for me, the fourth corner of defense is PRR.

17. For me the issue is much simpler.

Let's say that your findings actually show that AC can matter if you get it high enough, (So far it's not looking promising).

Does it matter if an AC of 250 will actually protect you a bit, if only a very small selection of builds can actually reach high enough AC to matter?

The reason the Dev's gave for the combat change was because AC was worthless for all but a small selection of builds.

And then they jacked up the Challenge Rating of mobs in Epic Elites to the point where we're back at square one.

What was the point?

18. Originally Posted by Book_O_Dragons
If we could figure out the enemy attack formula or find the original dev post on it, it would be possible to determine the enemy attack bonus.
It can be found at this link.

DDOWiki has a tad more concise summary located here.

19. Originally Posted by viktorserak
Why do you actually care, what works first?
Good question.

It has to do with player perception. If armor is checked first then players will see lots of Miss messages on their screen. They will think, "Oh, look how important armor is to protecting me." And, they will easily see that armor is giving them a lot of help.

But, if armor is checked last the players will see lots of Dodge, Incorporeal and Blurry messages. They will think, "Oh, look at how seldom armor is protecting me." And, they will conclude that armor is of little or no benefit.

Because the order is dodge first, incorporeal and concealment next, then armor last what we have is the second situation. We get people like Teh_Troll making claims that no amount of armor works, that players get hit on 2+ regardless. But, that simply is not true.

It is a disservice to the player community to allow the false information about armor to permeate the forums and go unchallenged. Players need to understand that armor class does matter -- even in epic elite content.

So, it matters to me because we already allow ignorance and innuendo to influence how and what we think. By presenting the facts and making them available to others to double check for themselves we can definitively arrive at the truth of how these things work.

That won't stop some people from continuing to spout falsehoods. But it will allow those who are willing to listen to the truth to discover and understand what the truth is.

20. Originally Posted by Archangel_666
For me the issue is much simpler.

Let's say that your findings actually show that AC can matter if you get it high enough, (So far it's not looking promising).

Does it matter if an AC of 250 will actually protect you a bit, if only a very small selection of builds can actually reach high enough AC to matter?

The reason the Dev's gave for the combat change was because AC was worthless for all but a small selection of builds.

And then they jacked up the Challenge Rating of mobs in Epic Elites to the point where we're back at square one.

What was the point?
I think the point is that more than a small percentage of builds can reach enough armor class for it to matter.

I know that for the longest time I thought the way you are thinking and I had the exact same thoughts about it not mattering. Then I was shown a way to get a much higher and sustainable armor class on my character. I saw that there was a way to get into the 130+ range on armor class.

For me the issue is that players often take the "AC of 250" line. But, that is a gross exaggeration. AC in epic elite content actually starts making a difference closer to 120. At 140 it is probably stopping 25% of the attacks. And, 140 is attainable for a lot of builds.

The reason we don't have the impression that 140 makes a difference is because we don't see a lot of Miss displays. And the reason we don't see those is because armor is the last thing checked.

So we think armor is doing very little (the "everything hits on 2+ anyway" attitude). In reality it is doing just exactly what it is supposed to be doing. It would be more obvious if it were being checked first. But it isn't checked first.

That is why understanding that AC gets checked last is so important.

If dodge were checked last people would be saying dodge is unimportant. But, because dodge checks first we see lots of Dodge messages. So we think, "OMG, dodge is key."

It is all about the difference between impression and reality.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•

This form's session has expired. You need to reload the page.