I like it as well when people discover alternate rotations, as long as the harder/riskier rotation has an adequately higher reward compared to the easier ones, I think it's worth keeping them around.

I do think DK Monk should be nerfed, the execution requirement compared to a standard rotation I think shouldn't be a loss of only 1-2%, same deal with Yukikaze. I like that they exist, but the loss should be greater as the risk of learning the "proper" rotation is made up for just in crit luck.

TK Monk back in SB went in the other direction; harder to execute than the standard rotation, but the reward for pulling it off felt good proportionate to the standard rotation. BLM has also historically done this a lot, even now we have advanced rotations that cut out uses of Fire III and Blizz III as they're the weakest parts of the BLM kit, instead relying on Transpose for swapping.

If the non-standard rotation is harder than standard, I think it should be up to a 3% or so gain, it'll depend on how hard it is to execute compared to standard, but 3% sounds right without it being massively imbalanced from what Square likely balanced around. It's a little trickier to decide on how big a loss it should be if non-standard is easier, since that comes down I think to how hard it is to execute or how much less risk is involved. DK Monk I think needs a bigger loss than 1.7%, probably closer to 5-7%. Enough to still be viable, but should still encourage using standard.

Skill output I think should be balanced logarithmically; twice the input/thought shouldn't be 100% extra output, but more like 20%. Going from an already difficult standard BLM to learning Transpose lines, it sounds fair to only gain about an extra percentage or 3.

Also, how much of a gain was it to use TK Monk back in SB? I can't remember, I just remember being bummed that it was removed in ShB.