Recent Failure Rates

[Verdandi]

2/5 on helanic tundra bracers
1/1 on helanic tundra boots
1/1 on artificers bracer
6/8 on tae ew shields

...so far it's been normal for me.

[Armaged]

Sunday:

Made 5 out of 6 Nightmare Compound bows with 250 fletching and Trophy. I was loving the RNG.

[Kahlmer]

0/3 on earth bows using geerlock.

[ChanterCrafter]

Ok after more info...

Ok so after a string of failures on earthweave and firestrand pants along with reports from others regarding problems...here's my hypothesis.

A while back SoE or VI or whoever was changing how tradeskills work...so that the higher your skill was above the trivial of the item, the more likely you were to suceed.

However, I think in the changing formulas for successes and failures, they didn't properly calculate tradeskill modifiers.

Somehow it seems that if the modifier would push your skill past the 252 mark, it turns the modifier negative. JCM AA works differently, however....so I think the new calculations simply overlooked JCM...causing it to have no effect.

It seems the best path atm is to not do any highend combines unless you have 250 raw skill. Plus do not wear any items that adjust your skill as it seems such items can have the opposite effect than intended.

So far SoE has been silent about this, I'm not sure if this was just an intended increase in high end failures or a mistake.

[GraemeFaelban]

I failed on 3 out of 5 Blessed Fishing Rods, my normal failure rate had been about 1 in 10 prior to that. On the other hand, I have had almost no failures making gate potions. In both cases, I had a +5% mod item equipped. So, really, no useful data to add one way or the other.

[Thicket Tundrabog]

I've also noticed some weird results with skillups. I went from 227 to 250 this past weekend in fletching.

It took me just over 800 combines to get 8 skillups. One skillup in 100 combines is pretty pathetic. It happens sometimes, but not usually for a long streak. My wisdom was 254. I intentionally chose that wisdom because of problems some people have had when they exceed the maximum wisdom cap.

I did another 80 combines and got five skillups to 240. Then I noticed that I'd forgotten to take off a piece of wisdom gear, and that my wisdom was maxed at the 255 cap (I know you can get higher with AA). My last ten skillups, with maxed wisdom took just less than 400 combines.

I suppose it could be the RNG, but I can't help but think that the skillup algorithm has some flaws in it. Perhaps there are 'sweet spots' where you get the most skillups (and maybe successes).

Thicket

[Taushar]

My wisdom was 254. I intentionally chose that wisdom because of problems some people have had when they exceed the maximum wisdom cap.

Can you elaborate on this? If this is superstition, I can understand but it's a prove fact (from developers) that stats above the "max cap" for under those with AAs to be only beneficial. I also have seen this personally with 320 wisdom that the skill ups are much faster than at 255 or 254 or whatever. I would not spread rumors like that unless you clearly state that it is a personal superstition not a fact.

Taushar

[Itek]

Groan.

There is no "geerlok nerf" stealth or otherwise.

YESTERDAY, as in the day before today, I went to skill up on smithing for coldain shawls.

Here are the relevant data points.

Skill 111 --> 122 on embroidery needles, did 33 attempts, got 24 successes

122 --> 132 on Fay needles, 22 attempts, 14 successes
132 --> 135 on ornate gold bracer, 6 attempts, 5 successes
135 --> 139 on ornate elec coif, 19 attempts, 17 successes
139 --> 146 on ornate gold coif, 15 attempts, 14 successes
146 --> 155 on ornate gold tunic, 52 attempts, 51 successes

That's with 255 Int, and geerlock.

The "attempts per skillup" and "successes per attempt" are BOTH well within the -expectation- for the skill level range and the standard "success" formula.

Skilling isn't broken, success formula isn't broken, geerloks aren't broken, 255 Int/Wis/Str isn't broken.

JCM 3 ..... that might have been FIXED.

OLD version JCM 3 reduced the "failures" by 50 percent from the expectation of the "success formula" that most people use. (success = 51.5 + modified_skill - (0.75 * trivial))

NEW version JCM 3 seems (note: SEEMS) to be a more accurate reflection of the DESCRIPTION OF THE ABILITY in that it reduces the "chance to fail" by 50 percent.

e.g.

Old Failure rate 2 in 20. (formula success 8 in 20 plus a 50% bonus)
New Failure rate 6 in 20. (formula failure 12 in 20 divided by 2)

Yes, that's a HUGE change.

Every enchanter who didn't see it coming.... wait. We _ALL_ saw it coming. We saw Dire Charm nerf coming. We saw KEI nerf coming. It was BROKEN. It was a BUG. It's been FIXED. Get over it. Those that spent 18 AA points reaped the benifits while they could, those of us who didn't yet now can decide if it's truely worth 18 AAs.

For those who like math ...

The biggest change is from 100% success to 75% success. This occurs for those with 252 effective skill at...

50% = 51.5 + 252 - (3/4 Trivial)
(3/4 Trivial) = 254
(3 Trivial) = 1016
Trivial = 339

Things with "effective trivial" well over 250 suddenly aren't "no fail" with JCM 3 anymore. I think for our own bodily safety the enchanter community should collectively take it's lumps. There are a HUGE number of tradeskillers wondering where Baking Mastery 3 is for Druids. Alchemy Mastery 3 for Shaman. Smithing Mastery 3 for (insert fist-fight from the melee classes here).

Oh, and for the guy asking "why are you worried about making sure your Wis is 254 exactly" ...

It's called "off by one errors" and they occur frequently in programming. (Details about arrays, indexing, black box testing, test cases, etc etc etc until your eyes bleed available in your nearest Software Engineering textbook.)

He's mistaken. They no longer use a single unsigned byte to store the values. (unsigned byte = EXACTLY 256 values from 0 to 255, which they OLD Everquest, i.e. pre-Luclin, tended to use a lot) At this point it's superstition. But in the era of "high resist value wraparound" it might have been a valid strategy. (Back when having a resist over 150, or 200 depending on whom you asked, started having NEGATIVE returns. i.e. if you had a 200 resist you got hit for MORE damage than someone with a 190 resist.)

[Ortrillian]

JC has not been fixed. Guildmate did 6 combines and got 3 successful combines. That is just pathetic for 250 JC + JCM 3. 3/6 is around what I would succeed on with just my 250 JC (clerics dont get JCM #$&#*$&). 75 - 80% is acceptable 50% is not.

*edit: Also, If the previous success rates were all "Bugs" then SOE should have the nads to say something in the patch messages so people can know wtf is going on. Ninja nerfing/"fixing" things just makes people upset.

[ChanterCrafter]

Hmm

Well more failures to report.

There is a problem...not just with JC but with fletching and tailoring as well.

It's not just a minor streak of bad luck.

The gearlock thing is just a guess, since guesses are all we have to go on at this point.

I admit some people are not experiencing higher failures...but many are. So that just points to something we don't understand regarding the higher failures.

The higher failures could be caused by particular focus effects, or certain combination of character stats, or race, or server, or who knows what.

It's not some conspiracy theory that people are making up. It's clear something is wrong in crafting or has changed. Yet because SoE has not mentioned anything about it, crafters don't know why the failures are occuring and how to minimize them.

--Maob

[Albur Bitterman]

More thoughts

I'm thinking that if there is a problem (if!) it might be with geerlok/trophy use and high (near 250) skill. Most of the reports (Itek, Scarbrowed) that report normal rates of success have done so at sub 200 skill.

Albur Bitterman

[Thicket Tundrabog]

In response to Taushar, my comments are neither fact nor superstition. In my opinion, they fall in between.

I recommend reading the post "Any idea what I could possibly be doing wrong?" started by Ornette Coalman. The post is on this board, but a few pages old.

Ornette averaged 189 combines per skillup for 30 skillups with 355 str/wis etc. These terrible results are far beyond statistical reasonableness for the random number generator.

There have been other examples that can't be explained by the normal vagaries of the RNG.

I recall these discussions in the days where 255 was the physical cap. Some folks experienced horrid skillup rates when they 'exceeded' the cap.

Poor skillup rates were the exception instead of the rule, but they did happen.

The cause of statistically significant poor skillup rates is speculation. To my knowledge it has never been discussed or acknowledged by Sony developers or representatives.

My own mathematics background causes me to speculate that skillup algorithms may not be smooth. They may have dead zones and sweet spots (highly technical mathematical terms... LOL).

Therefore, in my opinion, poor skillup rates that can't be reasonably explained by RNG statistics are fact (i.e. verifiable, repeatable).

The reason for the poor skillup rates is speculation (i.e. a theory, if true, could explain the observed facts).

Thicket

[Itek]

Originally posted by Ortrillian
JC has not been fixed. Guildmate did 6 combines and got 3 successful combines. That is just pathetic for 250 JC + JCM 3. 3/6 is around what I would succeed on with just my 250 JC (clerics dont get JCM #$&#*$&). 75 - 80% is acceptable 50% is not.

*edit: Also, If the previous success rates were all "Bugs" then SOE should have the nads to say something in the patch messages so people can know wtf is going on. Ninja nerfing/"fixing" things just makes people upset.

Groan^2

3 of 6. Wow, holy statistical sample size batman.

If the expectation of 252 JC is 3 of 6 then the expectation of JCM 3 is 4.5

3 of 6 instead of 4 of 6. Wow, that's a nerf. I'd sue.

You also neglected to state that while "75-80% would be acceptable" that PRIOR to this if JC 252 was 50% JCM 3 was _100_%_!! You didn't think MAYBE that was unintentional? (Read: BUG!) Not all bug fixes are announced. They never have been, they never will be. When they fix exploits they don't announce it. For the VERY GOOD reason that publishing OLD exploits that have been plugged encourages people to go looking for similar exploitable situations. It's a precept of game design theory.

ChanterCrafter -

Sorry, most of the people hollering "NERF" the loudest are the JCM 3 crowd. I did say "seems" that is where the numbers have taken the biggest turn. Now for the standard line. "How big a sample have you done?" Let's face it "1 in 1000" level unlikely situations happen almost HOURLY on EQ.

In my smithing run I had to make some metal bits. In the sessions prior I had not failed a metal bit since the "reduced low trivial failures" patch. I failed 2 or 3 metal bits. And a couple of rings. (Though to be fair I expected to.)

Albur Bitterman-

Aye, it might be. But it's unlikely. (read to the end of my post) They no longer use "single unsigned byte" to store values. There is no more "wraparound" of skills.

As a more general solution to the problem... They might have simply added 30 to the "effective trivial" of all the combines people are failing at more. Which would reduce successes by 30 percent for most 252 effective skill users.

That thought make anyone happy? It explains the failure rate increase, across a number of skills, and only effects the high end combines. Ahh, mystery solved. Nothing is broken. They just fastened tradeskillers to the wall. Happy?

(Note: HEAVY sarcasm in last paragraph. That's probably NOT what has happened.)

Thicket Tundrabog-

Everyone discusses the "hell levels" of skilling 180-220 and it's pretty much an agreed on thing.

Yes, you can write "non-smooth curve" algorithims. But people have come up with their own "expectation" levels for various skill ranges.

For me when doing skills lately for coldain shawls I have noticed that (very roughly)

0-50 one skill in two attempts
50-100 one skill in 2-3 attempts
100-120 one skill in 3 attempts
120-140 one skill in 4 attempts
140-160 one skill in 5 attempts
160-180 one skill in 6-7 attempts
180-220 one skill in 20-40 attempts
220-250 one skill in 10-15 attempts

But, again, as I said previously, the old "off by one" errors for 255 vs 254 Int/Wis has been made a non-issue by the way they store the data. Other than superstition there's no need to jump thru hoops to avoid 255.

[Tashim]

The reason that most information comes from JCM3 enchanters, is that is the easiest to test.

The only problems that I've experienced, are with jcm3+250skill (with and without geerlok). I would make hundreds (probably thousands) of gem studded chains for many different tailors. In fact, those were the primary reason I spent the 18AA on jcm3.

Prior to jcm3, I would average (over the course of several hundred attempts) around 50% success. That has been reported by other enchanters as well. Personally, I think thats BS, after making something thousands of times, you'd think you'd have a better shot at it than 50/50.

Then along comes jcm3; According to the description, it was supposed to reduce failures by 50%. So, in the past, where I had made 100 per 200 attempts, I then expected to make 150 per 200... But it didn't work that way. Amazingly enough, I was making 195 per 200 combines. From 50% failure to 2.5% failure. wow. Definately worth the 18aa, imo.

But now, things have changed again. Now, when I attempt 200 gem studded chains... I fail half of them. Consistantly. After 600 (200 without +skillmod) combines, I stopped making them, waiting for this obvious error to be addressed. I am still waiting. Personally, I feel that SoE has grown by amazing amounts, in terms of their communication to players; But in this particular case, they've failed pretty miserably.

I'll agree that jcm wasn't working as intended... but there is no way in hell that its working correctly now. Perhaps there is a seperate issue with using +skill mod items at 250, but I've not seen anything quite so prevelant as the jewelcraft failures. I do not often make hundreds of non-trivial combines with other 250 skills, so it is difficult to compare them.

Personally, if jcm is going to only work on trivial combines... then it should state that in its description. If its going to have 0 effect at 250 skill... then it should be removed from the game and have the points refunded to all enchanters who have it. Regardless of what SoE decides to do with it, they have a responsibility to communicate that to us.

I've sent /feedback on it, and even tho those don't get a response, they do get read. I'd encourage everyone to use that feature when you find a reproducable anomoly like this. Be brief, clear, and (above all) polite! Examples and numbers are good. Many people only bother to complain by the time they're too fed up to talk rationally, but if you take a step back, realize that you're playing a game, and have the mindset that you want to help improve this game (rather than bitching at the game developers), you may find that your issues get addressed. I certainly have.

[Sam DeathWalker]

And...

Any new infos on this after the patch. I did fair on some low level BD combines, but to scared to try bloodmetal ears again, although my business is going down without new items coming in.....