Crafting Issues with Quality

[weltall]

((quality of starting item * 2)-(bias derived from your lack of first skill+bias derived from your lack of second skill [if you get practical training you lack skill]))-(randomvalue). bias from lack of skill is >= 0 (0 is when you have got the best you can on that procedure) randomvalue can be negative and positive, if it's negative you get better results.

[valentineaaron]

Ah I see, so at some point you'll probably not never have items go down in quality anymore when you work on them.  I can see that already with spangens since they are so easy to make.  I no longer ever have them drop in quality when I hammer them.  So does metallurgy work the same way? If so then I don't think the type of metal determines if the items will go down in quality or not.  It must just be that you would have to max out the skill.

[Bonifarzia]

((quality of starting item * 2)-(bias derived from your lack of first skill+bias derived from your lack of second skill [if you get practical training you lack skill]))-(randomvalue). bias from lack of skill is >= 0 (0 is when you have got the best you can on that procedure) randomvalue can be negative and positive, if it's negative you get better results.

Well, I can't be as quiet as I intended to be.
The thing I do remember and that is still found as a fallback function in the work manager works differently, Weltall. The initial quality is doubled if and only if it is below the base quality, i.e. 50. That case is not of much relevance. Otherwise, the base value (still 50) is added once or twice, depending on the number of skills involved (i.e. 100 if there is a secondary skill). What is subtracted then is a penalty that is proportional to the relative position of the skill with respect to the skill range and some prefactor. (e.g. half the penalty if the skill is in the middle.) As far as I see, skill training beyond the range won't ever make any difference there. Last point: The random part is not purely additive, but rather a randomized percentage factor. So quality changes in the end by +/- some percentage of its value, where the distribution of the percentage value is a sigmoidal curve. Higher quality gives a bigger amplitude of noise, but partly hidden by cutting off becond 300.

Oops, maybe what Weltall wrote is actually true, and the public fallback stuff I talk about is simply outdated

[weltall]

you are right about the +50 or it would be too fast to get to 300 i missed the line.
it's actually + base quality.
i didn't explain in depth how the randomization function works i just kept it abstracted.

this is sorta the distribution of most crafting http://img822.imageshack.us/img822/8762/schermatafigure11.png

[Bonifarzia]

I did not mean to criticize your comment, Weltall. It's just to help those players out who are interested to understand what the function does by looking at the SVN link, just as I once did.
Let me guess the labels in your plot: Change in quality as a function of initial quality and the cumulated probability to get this or a better result? Nah, that cannot be it, the spread is huge. But the shape fits.

[weltall]

it's only the random distribution part

[Hangatyr]

Well time for me to join this discussion. First of all there might be some with higher skills than mine as I've seen them in the past, way back when Dweomer Detection worked. Funny as some of them had higher skills as it was possible to have at this given point of time; levels above 100. So as I don't licked or kissed the butt of a goddess, my skills are my work. Weltall point to one who outdo me, one who gained his skills like I did, the honest way.
Back to topic. It's nice to look at a beautiful formula, but it's different if one has to deal with it. q300 kits end up in q70 swords, what a joke. There is no way in RL that if I attach a handle to a blade that the blade looses its quality. If the rivet is loose, I replace it. If the screw is loose, I tighten it. If the leather around the handle is loose, I'll tighten it. BUT this has no effect on the blade.
Right now I am working on steel ingots and with my skill level the Q bounces up and down like mad, going from q85 to q300, which is a hell of a range. Dropping Q by 150 isn't uncommon to me. It's more the rule rather than the exception. Btw. I never saw it happen to go up by 150. You say randomization has just a little or no effect, well as I usually work of stacks so the static part of the formula doesn't change (quality of starting item and my skill level). So why, if randomization can be nearly 'neglected', the outcome of my work varies that much? Taken q220 as initial, the range is from q85 to q300.

[bilbous]

But ... but Hangatyr, you didn't tell us your levels! Mine are Met:85, BS:34, SwM:22 rest are lower. I would expect rather poor results except for smelting which ought to give rather average results.

I have yet to make anything q300 except for stocks/ingots up to gold -- oh and possibly bronze spangens, I can't quite recall. The swords I make average around 150 highest being about 220. Oddly enough when making swords the kit always takes the quality of the sword handle.

I believe there may be legacy code somewhere that was put in place so that crafters could get to q300 back when the highest level of the skill available was 50. I do not see why anyone ought to be able to make perfect weapons with only 1/5 the possible level, it ought to be rare with even 1/2 unless these weapons are trivial.

With the superior sword parchment we have at least the semblance of a more complex weapon system but I do not see a significant improvement in the types made from it over the types made with similar amounts of material in the base system. Perhaps I just haven't paid enough attention to the differences as I have not made a detailed study of the topic.

[valentineaaron]

Bilbous from what I gathered by the discussion so far the 300q range isn't the max of your stat such as 200, it's the point where you don't get training for the item you are making anymore.  So for example with spangens it's around level 15 in armor making I think.  When you are at level 15 the spangens will still bounce up and down in quality (which is very annoying when you are trying to make them 300q) but now that I am 30 then never go down at all anymore.  I will make a spangen now and it will go up to 300q in a couple hammers, and never lose quality.  I think that is how the rest of the items work as well for every part of the effort to make them including the combination. 

If that is the case though then something is seriously messed up with metallurgy since before the last changes happened you would max out on a metal but then never get a better chance of making 300q from that metal.  My skill is 126 now and I still have about the same odds of making bronze 300q stock as someone who just reached the point where they don't get training on it anymore.  I think with smithing though once you reach the level where you don't get training points to make an item you'll start seeing the random jump in numbers less and less to the point where it will never go down at all anymore.

[valentineaaron]

The other problem I see with crafting (other then metallurgy because it seems to be a mess right now) is that you don't really know when you'll stop getting training on an item you work on.  So far the only one I see you stop getting training on are spangens which is understandable since they are so easy to make.  But here's the thing with that, 1. you still get training for the parts for a spangen and you'd think the parts would be easier to make, 2. there are plenty of other items in the game that are just as easy to make, 3. if you get training for making simple items all the way up to maxing stats then or even making easy weapons with common alloys then there is really not much of a point in striving for a higher skill since you'll never overcome the randomness of the quality you get out of items. If you are never guaranteed 300q items no matter your skill it's kind of a waste of time to keep training after a certain point.

[bilbous]

I haven't done any real crafting in a week or two but last time I did I was getting practice for pretty much everything I did except trivial things like quenching. I was getting practice for making steel and bronze ingots which had previously stopped giving practice. The weapon/armor skills were never high enough to stop giving practice for those actions that gave it ever. It has been quite some time since I made a spangen as it was always quicker to max out the bronze quality than the spangen and the profit was no different.

[weltall]

well then you should correct your description as it's not the truth.
second there are both from the "divinity" and by themselves, who are from 5 to 50 levels above you. I'm sure you see one of them daily.
As for the "not seeing the wild effect" well you are too low at levels to be in that area you need to be above the training limit of the things you are doing and this isn't your case. Plus as you see from the diagram most randomness will space from 50 to -50 but there might be cases of -150 +150 too, even more at high qualities. So the more the quality is high the more wild will be the loss or gain.
The schema above is for sword crafting, some other crafts have lower randomness (but not extremely different)

[valentineaaron]

But Weltall if you get experience in making an item all the way up until you max your skill at 200 then the random issue will always be as big of an issue since you'll never increase to the point where you can lessen it's impact.

[weltall]

basic items don't get there, special items obviously cannot be normal so they must be more difficult to make.

[valentineaaron]

Good to know Weltall, thanks so the basic items made from bronze and steel will max out at some point lower the 200 in the skill.  I can live with that.  So to the anvil!