Craft is a skill reflecting your ability to make things. It uses the Intelligence stat as a modifier
What can you do with Craft?
There is a flat mechanic in craft for producing half your check result in gold pieces per dedicated week of work.
If you actually wish to craft an item, there is a fairly complicated process involving taking it's value in silver pieces. You then make a check and multiply by the DC of the item and that is your 'progress' in silver pieces in making the item.
Which of these have ground for use?
I have mixed feelings. I can certainly tell you that after a certain level of proficiency I've never failed to complete a drawing. Some have not turned out the way I like, but it wasn't due to random variance in skill. It was due to either a lack of skill, or a new technique.
With something like alchemy or baking, a certain random element is good, because there are vagaries of temperature, air pressure, and other variables that are outside of control.
I have never been a blacksmith or a fletcher. Perhaps a trained blacksmith (1 rank + 3 class bonus + 1 Intelligence) fails and ruins 25% of the longswords they make.
What is it we gain by having this skill?
It is a quick system to generate an earned integer of gold pieces every week. It allows characters to produce basic equipment at a reduced cost. It also allows characterization, defining a character beyond class roles or things they can do in combat.
What do we lose?
I'm not sure - no, wait, I am sure that no dungeon master anywhere ever has tracked silver pieces per day multiplying them out to determine when the item is completed. If you ever have, please let me know
Also, I'm having difficulty seeing the system as one that intends to determine the goodness of a product with high granularity. An argument was presented recently that in this or in a profession skill the degree of the roll indicates the goodness of the result.
This is objectively incorrect.
The system clearly intends the roll to indicate the speed at which the task is accomplished, and the goodness of the task is set by the creator.
I'm going to create a masterpiece of a painting - I set the value at 10,000 gp and the DC is 20 (for highly complex) and then I'm off. Even with one rank in craft if I keep plugging away at it long enough it will be accomplished (even if I waste a metric ton of materials)!
Making a bunch of arguments over whether that's accurate to real world examples is pointless. The real question is:
How is this implementation of the skill improving my game?
I do not believe that it does so.
Conclusions & Suggestions:
The granularity is too high. Also, it doesn't represent how actual craftsmanship works (in that you are not likely to fail in many cases).
I am fond of a 'non-weapon' proficiency system for such things - allowing a level of proficiency, and then further focuses of mastery and grand-mastery.
Most systems run such a skill as a great/pass/delayed/failed/disaster system which is something I find perfectly workable for alchemy, but as far as forging and other type of craft activities, I much prefer two to five levels of granularity and not forcing checks in order to produce - just requiring the time and materials seems to be enough.