The schedule is all, the schedule is the essence of professor Pareto's transmutation. We count the number of atoms we must handle (five), grant each atom an indivisible set of 4 instructions to move the atom, and include 2 more instructions to account for the two pivots. Another 2 instructions account for motion to the output glyph, for a total of 24 instructions to schedule. We distribute these sets across the arms in a cost target. To do better than this requires a multitasking multiarm, non-resetting multiarm, or a work-sharing track.

But if I get one more regrab problem with multiarms I swear to Asmodeus I will feed the offending arm to the Glyph of Disposal.

The other big limitation, besides geometry, was access points. Sometimes, we could count access points on arms and determine that an arm with three jobs couldn't perform all three in 4 instructions each, inflating the schedule.

Here, there are five 4-instruction jobs to split across four arms, so one of them must take two. We pack all of the pivots onto an otherwise not-very-busy arm, so one arm can alternate between moving silver and quicksilver, for the minimal 8 instructions.

Extra arms are more important than anything else. Every 20 guilder increment spent on multiarms or extra Glyphs of Bonding could instead be an arm that solves a scheduling woe. We only use extra bonders for the very fastest machines.

My production is that professor Pareto's solution book has a frontier smaller than 6,000 units, but that we only slide below that benchmark by pooling our efforts.

- Alchemist Kazyan, in correspondence with Alchemist Jasper