TECHNICAL SUPPORT / WEB APPs BUGS

Since long time, advanced products of the scrollmaker occupation are haunted by a bug which renders the occupation virtually useless if not harmful (because of its deceptive nature) leading to frustration of crafters as well as occupation practitioners. Since the fusion and the demise of advanced occupations, there is few if any point to practise this occupation, at least at grade 6/7 level and for masters in crafting.

The phenomenon of the bug is, that scrollmaker scrolls, when used by a crafting master, appear to reduce instead of increase the success chance of crafting. I made some series of measurements several weeks ago, after I used scrolls to increase success in heavy armour crafting with highest quality materials for a fellow player. The results were detrimental: 4 degrades in 10 craftings, where a success chance of 99% was indicated in the precraft window.

I started a series of tests, easy since I have all crafting masters, based on my daily cash cow, the craftings at the campfire in Dyron. There, in 10 missions, 1110 materials are processed in 67 acts of crafting. I tried it 10 days using q60 scrolls (pretending 98% success in precraft window) getting a 44% degrade rate against a "normal" degrade rate between 15 and 20% (notably, I am crafting with 250 mat at q250 all time though the mission require between 200 and 240).

Obviously, the scrolls are causing a mess. I wrote several tickets, anybody to work on?

Btw., during the last week I recognized a sharp increase of degrade occurrences. During the last 3 days, I measured 30-35% degrade rate (sample=~200 craftings) vs. a "normal" of 15-20%. I am aware that Ryzom random routines are a mess (I never had better than 15% over more than 100 craftings, even very seldom in 10 craftings, but often deviations towards worse, even over large numbers, which puzzles me), and the deviation may be personal mishap inside the standard error margins (though I wonder how). If anything has been done to the random stuff, please re-check.

And, please, repair the scrolls.

Daomei ---

I also craft a lot, and sometimes use scrolls. But I have a situation about a year ago that tells me that it's not as simple as it appears.

First, the display LIES... just plain LIES. I believe that the scrolls do exactly as they say, i.e. that a level 6 scroll increases your crafting skill by 12 points, not your success rate by 12%. That alone should be fixed, but back to my anecdotal evidence.

I had the pleasure of being told by my GL to use the armilo jewelers tool to make myself a set of boosted digging jewels. I harvested a *lot* (2.2K each) of exe seed and amber and made myself a lot of level 60 scrolls.

The theoretical number of jewels needed to get focus boost on a complete set is 400 (a 2.5% chance of the focus boost per action). Of course with a 15% degrade rate some of those boosts are going to land on jewels less than q250, hence the scrolls. Using the scrolls and working very fast I could get between 22 and 25 jewels made before the scroll ran out. I kept records of every degrade and every boost.

It took me 487 jewels to get a complete set of 10 focus boosted jewels. I also got 13 HP 16 SAP and 14 STAM boosts for a total of 53 boosted jewels (11%), well within random error for the armilo tool. I also had 44 degrades (9%), well below the expected 15% failure rate without scrolls. (The tool went from 95 hp to 5 hp and the jewels I was wearing lost 6 points each.) It is above the rate I would theoretically expect for a 12 point increase in skill level, but not statistically so.

I then had an opportunity to use more armilo tools to make a set for my GL. I didn't use scrolls this time, but otherwise my conditions (including the nature of the mats) was the same. Instead of the 2.2K each mat needed in the first experiment it took me well over 3K of each mat to get a full set. So, my anecdotal evidence says that the scrolls work. Perhaps not as well as we thought, but they work.

Since what we really need is data on multiple thousands of crafting actions, the question is not yet answered. However, the devs really should just look at the code and verify that it is at least written correctly, and check to make sure that the RNG is running smoothly. :)

Thanks for the response, Bittty. The explanation that the precraft window display is wrong, makes sense, as well the hypothesis that not success rate, but skill level is increased by n, where n is the increase the scroll is promising.

This would explain all the positive experiences I made with scrolls prior to having mastered crafting skills. Though I did not check and write down success rates I always had the experience of improved success, degrades declining and failures nearly vanishing, even with lower grade scrolls. So had others using scrolls.

But what happens when you are lvl 250 in a skill? If you are right, the skill level would be artifically raised to lvl 262-264 with grade 60-70 scrolls. This would, all else equal, result in a success rate somewhere above 90% though not 97-99%.

My experience is that this does not happen. My concern is that something similar and similarly wierd is happening as with the toolmaker's special object. That one is raising the yield of producing by 10% when used, but only if your recipe is below 100%. In case of a 100% recipe, not the expected 19 or 20 products are won instead of the regular number of 18, but 15 instead. My suspicion is that of a sign error side effect somewhere in the code.

To be honest, I have not found the scrollmaker nor the toolmaker stuff in the code when I inspected it last time, and that has not been recently. Maybe I have looked at the wrong places. Even if, I would not guarantee to find the reason of which I suspect that it is not obvious.

I am not sure I agree with you about q60 scrolls, Daomei. My experience of them is the exact opposite of yours. But hey, I could just be lucky, but they do seem to work for me. I have also found q40 scrolls to work better than none when trying for boosted focus gear. I just ignore q70 for the same reason as you state. However, Bitty mentions, due to small sample size we would need to do runs of hundreds of thousands of crafts, at the very least, to get reasonably reliable results, and avoid bias confirmation. The last time I used q60 was for a run of 100 amps with a rubbarn tool. I got 6 degrades and 10 boosts. But a hundred crafts, just like a 1000 or 10000, tells us nothing about rubbarn or q60 scrolls. Though Bitty's explanation about the display does make sense, I suspect that you are correct, at least, about q70, and that there is something broken here, in addition to the display.

If you look at the info on a Masterful writing necessities scroll it says "Modifies your Craft skills by 12 point(s) for two min and 0 sec." I just took that as being true, and found it to be consistent with my success rates. :)

Thanks a lot for your input. Considering your comments, I cannot rule out to have used q70 scrolls, not q60 (though I was sure about or believed so). I am now restarting a test over a longer period, first measuring average degrade rate over some time, then testing by applying q60 scrolls, again.

Still it seems to me that the randomness is flawed, but I shall inquire further on.

Client is calculating craft success% wrong. One possible fix is here

Karu --

I looked at that, and it looked as if the client was *displaying* an incorrect success rate. We are fully convinced of that (and it would be nice to have your patch applied to the client as an update).

We now are wondering if somehow for q250 crafting and q70 scrolls the actual success rate used in combination with the RNG to determine success has been miscalculated.

I'm afraid I am not fluent in C++ or whatever that language is and cannot be absolutely certain what I was reading.

A short intermediate message: I started the measurement of my daily craftings and got the first series of results.

Summary:

My perception that there is some deviation from expected results in disadvantage for the crafters has been disproven. On contrary, the results so far, while well inside the error margin, revealed a deviation favoring the crafter.

Methods:

I chose my daily crafting in Dyron as test setting. Every day, I am taking the 10 missions from the overseer at the campfire. I solely processed materials of q250 (or above) with crafting quality 250, thus an expected success rate of 85%. This set of missions comprises of 71 acts of crafting, processing 1085 materials. The number of craftings necessary may increase, though, in case a crafting ends up in a degrade below the quality required by the mission giver, and has to be repeated.
I measured the outcome of craftings for 2 weeks and 15 sets of missions, between June 27 and July 9.

Results:

In summary, I had to do 1086 craftings resulting in 138 failures. This is a mean percentage of 87.3% success or 12.7% failures, below the precraft expectation of 15% failures. Median was 73 craftings and 10 failures, thus 13.7% failures.

The results contradict my original impression of a bias in disadvantage for the player. No matter whether my perception was flawed at that time, or I had a series of bad luck, the results demonstrate that crafting success over a longer period stays inside the range of expected outcome. Thus, the RNG seems not to be broken ;)

Next step will be the observation of craftings under the same conditions, using q60 scrollmaster product "masterful writing necessities". The tests are underway. So far, they show impressive advantages using the writing set. After having finished this series, I shall test the q70 set (though for a shorter period) to test whether the impression of failure holds for these products.

Second part of the test accomplished, results below.

As reported, my first series of craftings with controlled success/failure revealed that crafting without boosting by scrolls stayed well inside the expected range of 85% success chance, even a bit in favor of the crafter but well inside the error margin. The second series of tests was designed to show whether usage of scrollmaker products q60 (masterful writing necessities) changes the outcome.

Summary:

Usage of masterful writing necessities showed impressive improvement of crafting success. Compared to the results without boost, failures roughly halved. As others noted before, this does not mean an improvement to 98% success for 2 minutes as indicated in precraft window, rather something around 92-93%.

Methods:

As before, all 10 crafting missions in Dyron were obtained and done, with q250 mats and crafting at q250 too. Similar as before, these missions were executed 15 times, between July 11 and July 26. Masterful writing necessities were active during all craftings (in case a crafting fell outside the boost time as the boost had elapsed while crafting started, it was not counted, that happened a few times).

Results

In summary, I had to do 1082 craftings resulting in 74 failures. This is a mean percentage of 93.2% success or 6.8% failures, far below the precraft expectation of 15% failures without boost. Median was 72 craftings and 4 failures, thus 5.6% failures. The variance was impressive: best was 1 failure of 71 (98.6% success), worst was 11 failures out of 73, thus 84.9% success.

It is obvious that q60 masterful writing necessities impressively improve the outcome of crafting.

Still underway is the test with q70 writing sets. This one will be shorter and only a sample or proof of concept determining whether these scrolls seem to
- improve
- deteriorate
- or not influence crafting success at all.

Due to the relatively high cost and time consumption of this test (scroll boost lasts only 1 minute with 4 minutes cooldown). only 5 crafting cycles (355 crafting acts) are scheduled. I shall report soon.

Your data gathering is impressive, but dont forget that as with all % in this game one outcome does not affect the next, so it is always random :(

Here now the results of the last part of the test: The question was how q70 scrollmaker products influenced crafting.

Preface

Some considerations before: We had found (and it prove consistent with the test results) that scrollmaker products are temporarily raising the skill level of the crafter, not the success chance. This is granting the crafter the effect of undercrafting. When crafting at your level, success chance is 85%. I will rise by 1% every 2 levels, thus be 90% when crafting qualities 10 below the own level, and 100% for crafting qualities 30 or more below the own level.

The effect of scrollmaker produce is thus expected 1% for q10, 2% for q20, 3% for q30, 4% for q40, 5% for q50, 6% for q60, and 7% for q70. Durance of boost is shrinking from nearly half an hour for q10 to 4, 2, 1 minutes for q50, q60, and q70.

Failure of q70 scrollmaker produce has been widely observed and reported. The purpose of my test was to check this effect. The tests were planned to take place between July, 27-31, but were extended until August 3 due to unexpected results.

Summary:

First, test results seemed to corroborate the hypothesis that q70 writing sets did not influence crafting success, neither positively nor negatively, in any significant way. But after a couple of days, results revealed a positive effect consistent with the expected boost. It seems that something has been changed.

Methods:

As before, all 10 crafting missions in Dyron were obtained and done, with q250 mats and crafting at level 250. Originally, tests for 5 days, between July 27 and July 36, were planned, but extended due to unexpected outcome. Artful writing necessities (q70) were active during all craftings (in case a crafting fell outside the boost time as the boost had elapsed while crafting started, it was not counted, that happened a few times).

Results:

The test results indicated that there was no effect of the artful writing set (q70) during the first 4 series of crafting: 295 crafting acts during 4 days with 41 failures, an average of 13.9%. Yet, the crafting series at July 31 brought only 3 failures, a better result than any previous crafting day without boost. This made me curious whether that was just a lucky outcome, or possibly a change in the scroll's effect. Thus I decided to test for 3 other days to see whether the effect is lasting.

Indeed, the outcome of the following days showed that the effect remained. During 291 craftings there were 20 failures, thus an failure rate of 6.9%.

It seems that q70 scrollmaster products are working, now. As the test is quite exhausting, I leave it by that, but want to ask the devs whether something was changed in the code so that we may use this stuff now without worries. I also ask others to test whether the observed effect is remaining stable.