In this section, I would like to preempt reception to this document, talk about areas of development, and outline problems with theorycrafting. I think the primary question people ask when reading this piece is “What does it mean my BiS is?” or “What does it mean for substats?”. To the former, this piece is a precursor to answering that question. Well-developed simulators are the most accurate theoretical approach to determining Best-in-Slot because the gains from increased GCD speed are not straightforward for every job. For example, a job like Warrior can make comparatively more gains from a faster GCD if it nets them more attacks under Inner Release. As for the latter, stat-weights were always a misnomer.
How much of a stat you begin with and the type of stat will ultimately determine the value from adding more of a given stat. For example, your first tier of Determination will always be worth more than any subsequent tier of Determination. That is because Determination has diminishing returns as you stack it. Since the gains from attributes and substats changes as you add more, stat-weights are inaccurate despite its obsequious following.
Right now, the most incomplete areas of research are pet mechanics and order of operation. The former is largely stifled by our inability to directly view pet stats. There is conflicting work between players that swear pets mirror their masters with only a percent cut, during calculations, distinguishing them. Others claims that pets have their own starting stats and only appropriate their master’s gear. In my opinion, the majority of tests tend to favor the latter.
Next, order of operations is vaguely understood, but still indefinitive. There are certain floorings we do recognize, but there are times where our calculations will be off by one point – this suggests that some rounding error has occurred.
I think the big challenge for the theorycrafting community will be documentation, transparency, and presentation. Right now, we document testing using Discord pins, but it is unsustainable and still somewhat chaotic. Ideally, tests would be reviewed by multiple, informed, peers and publicly posted. However, this is a game and so I expect any review to be brief and ineffective – understandably so. Compounded by the community’s propensity to latch onto spavined information, you can see why we are hesitant to post our findings periodically.
Finally, I hope that some of the changes and additions to the previous How to be a Math Wizard helped readers digest its content. There are congruences I did not elaborate in this piece that may have helped with clarity (e.g. most functions use the format y = [ A × (stat-B) / C + D ]), but I felt the formatting conflicted with the subtext needed to clearly explain each formula. That’s all I have to say on the matter.