This puzzle started with a simple premise: let's have some fun with dispersion in a cabinet. If memory serves, there are no cabinet dispersion puzzles on the leaderboard, so I figured, why not change that?

I almost didn't pick instructions as a metric, but after doing half an instructions solve my gut feeling was that the puzzle would be great for I. So I swapped it over, and it's blown me out of the water. There's so much variety in the awesome solves I've seen, and it's remarkable how different the puzzle would be if you mirror the output. So all in all, the decision paid off.

But I do have a little something for cabinet instructions optimisation, so I'd be remiss if I didn't share my own solve.

It took me longer than it should have to come up with the motif of two hexarms, with a bit of help, pulling all the cardinals off of dispersion. Once that was found, I spent a long time playing around with various ways to try to time the quintessence nicely and avoid crashing back into the input. I never figured out a way to do it elegantly, because the depression glyph is just too sensitive to suppression in this approach, and that messes with the timing.

Instead of timing it carefully, a simple hexarm will do, with its own helper to ensure that quintessence never suppresses dispersion.

Using a hexarm pays itself off anyway, though, because it makes the assembly incredibly elegant and easy. The simplicity makes room for an extra triplex bond, which means that only 3 instructions are spent on assebly, with a single pivot to let the fire pair access the bonder.

Except...

Do you ever wake up in the morning and find yourself swallowed by a pit of depression, unable to continue your journey? That's exactly what happens to the quintessence on cycle 69 (I assure you, it is NOT nice).  The next quintessence arrives at the fire pair too late, and the upper arm is about to grab it and cause an explosion...

So I have an entire arm dedicated to moving the product temporarily out of the way of the upper arm's grab.

This is maybe my least favourite arm out of all my solutions. It just feels like it shouldn't be needed, especially given how the solution eventually settles into a rhythm with dispersion firing every four loops. But the more consistent the outputs and rhythm, the less quintessence make it through to pick up products. So there's seemingly no way to win.

P.S. This is actually my best IR solve---it's possible to give half of the piston's duties to another arm and get it down to 7P. This is cheaper. But I seem to have misplaced the 7P solution, and it is slightly less pretty anyway.

P.P.S. But wait, there's more!