Crafting research

Here's a summary of some research FourteenDays did on the details of how crafting works.

Basic research
For this, FourteenDays simply crafted some items which required only one worker, and watched closely to see how many points they actually produced each hour. These results were done with 5 leadership and all workshop improvements except the highest ones for sawing, sewing, alchemy, and enchanting.

Conclusions

 * 1) It looks like most of the improvements are percentage-based rather than a flat increase, since the numbers are constant when expressed as percentages but not when expressed as raw numbers.
 * 2) The exception may be the Leadership skill. If we assume that the 5 points I have in Leadership translate directly into 5 skill points, then the percentages still remain constant. This is illustrated in the rightmost two columns of the table. Since these numbers look neater than the first set of percentages, I think it's more likely that this is a straight increase, but more data is needed here!
 * 3) It's pretty clear that the third-level improvement is +10% to the appropriate skill. Estimating the other improvements from this, however, is tricky, specially since it's not clear that they stack. Here's my hypothesis:


 * Improvements that benefit both skills (blacksmith tools, sorceress book, tailor work table, carpenter tools): +5% to both skills first improvement, +10% second improvement, +15% third improvement
 * Improvements that benefit one skill: +10% to skill first improvement, +20% second improvement, +30% third improvement

Improvements do not stack. This hypothesis produces nice, simple numbers which agree with the data above 100% of the time, so I think it's pretty likely to be correct, but some cross-checking would be nice.

Co-op penalty
Next I did some research on the "co-op penalty" by measuring the number of points produced when crafting a recipe with additional workers. This was tricky to measure so not all spaces below are filled in. In addition, I've included the worker's base cooperation skill as well as their actual cooperation skill.

Conclusions
The workers seem to fall into four brackets, with each one suffering a different coop penalty. The carpenter doesn't suffer a coop penalty at all. It seems reasonable to hypothesize that because his actual coop skill is above 100, this negates the coop penalty. In fact, if we hypothesize a penalty of

(10-[actual coop skill/10, rounded down])*2%

this seems to work reasonably well to explain the data, but there are still a few points where it doesn't quite seem to round correctly. I'll need to investigate this more.

Update: It looks like the formula worked out by Fexghadi below is the correct formula, as this works perfectly with this data as well.

Additional data
At first glance, it looks like the penalty for multicrafter items is linear and is calculated as follows:

((100 - Actual coop) / 4) * (Required crafters - 1) %

Update: it also seems having more than 100% doesn't give any bonus to productivity.

Open questions
from WikiPim:

The creator(s) have/ has added a nasty twist here. They/ he have/ has announced that longer work on items give a HIGHER bonus EVENTUALLY then short term created items. It is unknown how this has been implemented and if this has the effect it is expected to have. It may also have been taken out later but due to lack of well documented updates, this is impossible to tell. It also may be a very minor effect.

response:

That is very interesting. I investigated several items which took a long time to craft (gungnir, lionheart) and it looked like the number of crafting points remained constant. It looked like there were some changes in the number of experience points, so maybe that's what they were talking about? In any case, this remains unclear.