I wrote most of this late last night, and I get long winded. Sorry for wall of text, but this is the area of the game I'm spending most time thinking about lately, so I hope it's all relevant.
I'm relatively new to the game, and found the roles in Dwarf Therapist early. I thought "Cool, it's all there". I always looked askance at the creativity x2, but everything else seemed to fit with what the wiki had to say. But the more I read on the forums the more it seemed like what you call it: a skeleton to build on, useful but far from complete and not accurate. I was excited when once you explained how to use custom roles in views, and wanted to redo the default ones too, but that was subdued by frustration. I seems there's too little information to go on yet, if the goal is production speed and quality.
Most of the information I've found on how traits and attributes affect job speed, quality, persistence, etc is very speculative, with one assertive comment or observation immediately countered by another. One guy puts every consideration on this or that while another affirms it really doesn't matter. It seems faith-based, and I can't get on-board with that. I've seen a couple of posts that positively identify a trait as doing something specific, such as traditionalists engraving more historical figures and self-disciplined dwarves taking shorter breaks and holding longer in a fight. Makes perfect sense and may have been the case in someone's game, but where's the measured evidence? These could conceivably be implemented but not functional, like the musicality attribute. Or Toady had a completely different vision for these than the speculating players. Most probably
do have some significant function, and ignoring them in favour of something with proof (such as creativity x2, though I'd like to see that experiment for myself) would greatly skew results. That's why we currently see a few dwarves so good at everything and the rest show as fairly useless, when they might not be. The few that still jump out for certain roles after the godly choices usually weigh in because of preferences that are wholly unjustified, so I put prefs at 0.0 weight and do that part manually even though they are my most significant consideration.
So far I've mainly seen evidence for preferences. They make significant difference to quality early on, but almost none with legendary skill. And in addition to happy thoughts upon noticing/consuming things they like, they also seem to be top choices for moods, mandate and trade ban materials and items. So war hammer, adamantine, door, or gold/statue loving dwarves will be nudged to their future moods in hopes of useful artifacts and legendary skill. Steel, flux/obsidian, and oak wood dwarves will, depending on map materials, have pre-determined careers as armour smith, mason/engraver and carpenter because of their early game advantage and long term utility. Dwarves with common preferences, good socials and essential traits gets to be mayor. But other than that, I can't find further info on preferences.
Happy thoughts seem to be a big consideration for some current default Dwarf Therapist roles. But will a smith admire the anvil used to build the forge? Will a farmer persevere against the death of a husband because she likes plump helmets and barrels? Does using/producing preferred items trigger happy thoughts, and isn't keeping dwarves happy more important than a few extra high quality products? This is a question I'd really like answered. Furthermore, while preference Plants(Alcohol) and Metal might be convenient for setting up roles, it doesn't make any sense to give wild strawberries equal weight to sweet pods for a farmer or cook, or rose gold being no different than bronze to a weapon smith. Some materials and products are clearly more appropriate. Others will be more readily available, more precious, or entirely depend on availability on the map, making them the right choice to determine weight.
Attributes supposedly have an affect on job speed and product quality, but nobody I've read has demonstrated that advantage to be dramatic or even true. With so many players having no problem producing for the fortress even blindly assigning dwarves, this seems of little real consequence. But attributes are very significant for military where it's life or death, especially with the attribute cap depending on their starting stats and both resistance and recuperation not being trainable.
Traits are the biggest mystery to me. Aside from doctors and nurses getting happy thoughts from helping others, and mayors potentially missing out on an entire social skill if a trait is too high or low, there's no definitive relationship to roles. I'd like to assume they are all implemented and integrated with appropriate skills/jobs. But even if they are, it's not clear how.
As I mentioned before, this is all complicated by individual situations, embarks, and player preference. Everyone will have a different idea of optimization. I might like to get the most masterwork gold statues possible, but you might want to make the best out of a limited platinum supply. I might be thinking long term benefits and not care about skill levels, while you might want the best dwarf for the job
right now. Default roles will never please everyone. But until we figure out all the variables it does narrow utility and expectations for roles. Therefore, for me at least, it gives very specific focus, priority on what I can achieve with roles.
1) Targeted artifacts (if I'm not wrong about prefs vs moods).
2) Maximize high quality weapons and armour.
3) Designate specialized, multi-disciplined, important, time intensive or time restrained jobs like goldsmith, CMD, mechanic, butcher/tanner, farmer.
4) Males aspire to be elite gutbusters or become regulars.
5) Industry specialists (with focus on happiness from mat/item/tool prefs?).
6) Industry generalists by sector, and various non-critical jobs.
7) Reserves/Haulers.
This is much simpler than trying to find the perfect dwarf for everything, but it's not quick. It's actually compatible to the most
thorough dwarf optimization system I've found. I can be a perfectionist and I'm trying to accept that for 90% of dwarves it doesn't
really matter what they do. Even with that pressure gone, though, I'm compelled to optimize it and need to set up these custom roles. I spend too much time paused after my first and second migration waves bouncing back between pros and cons and doubts as I juggle job matches between limited "suitable" dwarves.
These are some of my ideas for roles, beefs and concerns, knowledge and wishes. If anyone can point to solid info about attributes, traits or prefs here it would be very useful. If I've got anything wrong, don't let it slide and cause confusion; correct me. Also, with the ability to change these stats with DFhack plugins, I'd love to see some experiments for the unknowns. I only discovered those scripts on the weekend, so I won't be doing my own science just yet.