I note that he is an enchanter, likely mana, likely talking about his base nuke.

One thing I've noticed on Uthgard that I don't remember from Live, is the resist rate on base line spells is MUCH higher than the resist rate on spec spells. To wit, while leveling, my lvl 21 and 28 ae mezzes had a much higher success rate than my lvl 43 single target mez. There may have been a difference in these resist rates on live, but if there was, it was certainly not as drastic as what I am seeing on Uthgard.

Jaerun wrote:I note that he is an enchanter, likely mana, likely talking about his base nuke.

One thing I've noticed on Uthgard that I don't remember from Live, is the resist rate on base line spells is MUCH higher than the resist rate on spec spells. To wit, while leveling, my lvl 21 and 28 ae mezzes had a much higher success rate than my lvl 43 single target mez. There may have been a difference in these resist rates on live, but if there was, it was certainly not as drastic as what I am seeing on Uthgard.

Having leveled a bard to 41, I actually agree with this wholeheartedly. There have been MANY time when I couldn't get a ST mezz to stick on mobs, so I'd throw out an AE mezz outta desperation and it would magically stick on most if not all mobs. I am only 21 Music (i.e. my AE mezz is only lvl 21 vs. lvl 50-52 mobs).

Grunklestank wrote:Lurker, while I'm currently on the other side of the argument than you, I must say - nice posts. I nerded out on them and enjoyed reading over your work. I'd just like to point out a couple things:

(Using the 12.5% 50v50 resist figure that you pointed out)

While a 5 resist streak is mathematically possible, it's still a 1/32,768 chance of happening... that's very very low. And, albeit anecdotal, a lot of people have reported encountering these types of streaks, and fairly frequently.

A 50spell vs. 50 target 8 resist streak has a 1/16,777,216 chance of happening... that's rapidly approaching lottery-winning status.

I'll be the first to agree that we need more empirical data tho. Anecdotal evidence and negativity bias are TOO STRONK in humans

Cheers xD

I'm not really on any side, as such. I'm on the side of providing data / substantiating claims. Thought it may be interesting to run some high volume tests in C# and look at the results from afar rather than up close.

Looking at the DoL code i can assure you really that the numbers generated there are truly as random as Microsoft's Random Class is, which is pretty commonly used / tested / rated. I read blues post saying he upgraded the rng in Uthgard code, as such i suspect they are using Microsofts RNGCryptoServiceProvider Class as it is the most obvious choice. I may be wrong there. But that too will provide a very solid random series of numbers. The code to calculate resist rates in DoL very simple and the same code is used across all spells. Any bug here would be easily identifiable by the developers.

Perhaps it's this line?

Code: Select all

if (m_Player.UserName === "sendnudez") {
    resistRate = 75%;
}


Really, its an age old debate. It's certainly not something that comes naturally to humans at all.
We see patterns in nothing and as such we need to be working with data, not anecdote.
https://cocosci.berkeley.edu/tom/papers/hard.pdf

Give your friend a piece of paper and ask him to randomly draw 100 dots on the page. Almost always you will get back a piece of paper with a fairly even distribution of dots all over the page. That's a typical human's understanding of what random should be like.

Interestingly (or perhaps not), Spotify recently decided to change their shuffle mode, making it less random, due to the complaints by users that it wasn't random enough. here.


Anyway...

1 in 16 million sounds very small to us, yes. I wouldn't be at all surprised to find that many pvp spells have been cast by this point in uthgard 2s life, but hard to say. And this is obviously assuming that everyone is casting level 50 spells which won't be the case so that likelihood of seeing the streak of 8 will come down a fair bit if we look at 'red' spells.

We are just as likely to see that 8 resist streak at 2 million spells cast server wide then we are at at 10 or 16 million spells cast. To top that off, once that 8 cast streak has happened, you're just as likely to see it happen again in the next 100 casts as you were in your last 100 casts.

Let's assume that at least 2 million pvp spells have been cast (reasonable?) would you agree with me that there is, at worst case (50vs50 only), a 12.5% chance we will have already seen a 8 cast streak? That's sounding pretty plausible now that's the same chance as a single resists. I guess you would agree with me that there is still a 12.5% chance we will see another 8 cast streak in the next 2 million casts?

Eugh, hate this stuff lol. Brain melts and obviously proves nothing because we are trying to suggest that random events should follow some structure.

(Also, pseduo random coding for resists fixes this whole problem, and would be very easy to implement... so why not do it? sigh )

I don't think this would be as easy to implement as you think. My understanding is that in order to smooth random streaks out (i assume that's what you mean) you basically have to pre-generate a load of random numbers and then sort them using some algorithm that attempts to achieve some form of even distribution.

This would obviously need to be done on a 'per player' basis and you run into difficulties about how many to generate, how / where and for how long you store the sorted list, performance around sorting the lists, etc. etc. It's doable, but certainly not elegant. That said there may be some library that does it all for you.

I think the main, and appropriate reason for not doing it... that's not how it's implemented on live.
Anyone have a live account and can give up some time to let me cast disease on them a few thousand times to validate this?

Jaerun wrote:I note that he is an enchanter, likely mana, likely talking about his base nuke.

One thing I've noticed on Uthgard that I don't remember from Live, is the resist rate on base line spells is MUCH higher than the resist rate on spec spells. To wit, while leveling, my lvl 21 and 28 ae mezzes had a much higher success rate than my lvl 43 single target mez. There may have been a difference in these resist rates on live, but if there was, it was certainly not as drastic as what I am seeing on Uthgard.

Unless I'm soloing I rarely use my base nuke aside from helping to burn some targets. The resists on base nukes are as frequent as the resists on my spec debuffs. When PLing a friends Bard, he was able to mezz Fins without any issue at lv 16

Sendnudez wrote:I'm too stupid to understand like 75% of that. I don't care what the formula supposedly is when it clearly doesn't function anything like live - at all. As I said earlier, if that resist chance was actually correct & the resists I experience daily was pure chance than I'd have won the lottery a few times over.

Just because you are too stupid theres no need to be so disrespectful to someone who put in a lot of work to explain your fallacy to you. It is also pretty easily written, without needing to understand the math behind you should be able to grasp the concept of the post.

And no, being the One in a 1 to 2000 chance is not the same as winning the lottery multiple times.


Does that mean resist rates are fine? No. Does it mean resist rates are broken? No, but with the current evidence it is not probable to assume it to be broken.
Provide a chatlog, like you announced and further investigation is possible. Just spewing poison because you incapable of understanding someone elses work isn't the way to go

Who's being disrespectful? It's over my head - it is what it is. The lottery comment is sarcasm, btw. I believe I've said 4 times now in this thread alone that I have flat out zero idea how any of this actually functions or potential different variables involved. We can make that 5 times now just to be safe.

I've said that I will be now saving my combat logs to do what I can to provide some additional info here. What else do we need to reiterate?

I still guess there was some sort of adjustment to resist rate on live between old and new ra change. Sadly there is no way to prove that without the actual code from 1.65 though.

Just out of interest...

I re-enabled (1 month free atm) my 2 old buff bot accounts on live and am currently sat in Savasd Faste casting disease repetitively from one onto the other (level 45 spell) and logging the results. (neither have MoF or any item that adjust spell level for resists).

I did just see this though... in the first 20 minutes or so.



Im working for the next 8 hours and will be periodically dumping my mana pool to record more data.

I'm positive that you make a great point about humans thinking they see patterns when they may or may not be there. When it screws up a fight it's easier to forget the 20+ people you just melted earlier. Haven't had the time to rvr yet this week but I'll get some logs up by the wend of the weekend at least .

That screen shot tho ew

Sendnudez wrote:When it screws up a fight it's easier to forget the 20+ people you just melted earlier.

Negativity bias at its finest

LOL most people are clueless as to how powerfully this affects humans

Grunklestank wrote:

Jaerun wrote:I note that he is an enchanter, likely mana, likely talking about his base nuke.

One thing I've noticed on Uthgard that I don't remember from Live, is the resist rate on base line spells is MUCH higher than the resist rate on spec spells. To wit, while leveling, my lvl 21 and 28 ae mezzes had a much higher success rate than my lvl 43 single target mez. There may have been a difference in these resist rates on live, but if there was, it was certainly not as drastic as what I am seeing on Uthgard.

Having leveled a bard to 41, I actually agree with this wholeheartedly. There have been MANY time when I couldn't get a ST mezz to stick on mobs, so I'd throw out an AE mezz outta desperation and it would magically stick on most if not all mobs. I am only 21 Music (i.e. my AE mezz is only lvl 21 vs. lvl 50-52 mobs).

Not sure about that..every formula I've ever seen only takes into account the spell level and target level. Unless dev says otherwise I can only assume that is how it works here. If it is different that would be good to know as it totally changes the conversation.

^ It is really weird & I'm glad he brought it up because I forgot about it. Before rerolling Hib, my Sorc's ST mezz was borderline useless. AoE mezz was reliable on epic purple Summoner's Hall mobs tho lol. I've been telling Bards getting burnt out on leveling to just go stand at a Fins camp at level 15 + & offer to CC. You WILL eventually get grouped, & actually be very useful.

Just an update...

I logged onto live and cast 3407 lvl 45 disease spells against a lvl 50 target with no RAs or items which should modify spell resist rates in any way.

Here are the results...

-----------------------------------------------------------------
Daoc Live - Parsed Log File
-----------------------------------------------------------------
Total Attempts: 3407
Total Failures: 532
Observed Failure Chance: Value: 15.61%.
Streaks of 4 Failures - seen 2 time(s) - probability: one in 1,682 @ Value: 15.61%.
Streaks of 3 Failures - seen 4 time(s) - probability: one in 263 @ Value: 15.61%.
Streaks of 2 Failures - seen 64 time(s) - probability: one in 41 @ Value: 15.61%.
-----------------------------------------------------------------

So you can see it follows pretty much what you would expect for a proper random system.

We see streaks of 4 twice, shame there wasn't a real outlier but that would have been too lucky for me! Leads me to believe they are not smoothing out those streaks at all. I think it's probably fair to assume that if i cast 20,000 tests i would see a streak of 5 and a 30% ish chance i would see a streak of 6.

I ran some comparisons against a a sample data set from Random.Org as well 2 outputs from the C# RNG, one at the original daoc sample size (3407) and one at 20,000 (i see a streak of 6 here even). You can see those results in full here:

Results Breakdown - https://pastebin.com/raw/tT5fDTkZ

Also just for full disclosure:

Daoc Chat Log - https://paste.ee/p/6uD8c
Random.Org Data - https://pastebin.com/raw/5Us8JZj9
Code - https://pastebin.com/NnAHchjn

Edit: wording.

Just simulated the same test twice on Uthgard:

Level 50 player caster
Level 50 player target
Level 45 disease spell

Total Attempts: 3407
Total Failures: 486
Observed Failure Chance: Value: 14,27%.
Streaks of 4 Failures - seen 2 time(s)
Streaks of 3 Failures - seen 7 time(s)
Streaks of 2 Failures - seen 57 time(s)

Total Attempts: 3407
Total Failures: 490
Observed Failure Chance: Value: 14,38%.
Streaks of 4 Failures - seen 1 time(s)
Streaks of 3 Failures - seen 10 time(s)
Streaks of 2 Failures - seen 59 time(s)

Thanks Lurker for putting in the time getting some data on Live!
I am quite satisfied with both results

Can we setup some sort of tip jar for having this guy actually do what needed to be done all along?