This solution is named roughly after the noise it makes, since I ran out of good names. I ran out of good names quite early, and other submissions had such names as "bingle bongle dingle dangle" and "sneep snorp". The point of the solution is to divide the labor between the five input bits as cleanly as possible:

The first bit controls a +3 promotion to the bottom metal.
The second bit controls a +3 promotion to the top metal.
The third bit controls a +1 promotion to the bottom metal.
The fourth bit controls a +2 promotion to the top metal.
The fifth bit is a bit more complicated - if the fourth bit was off, it grants a +1 promotion to the top metal; if the fourth bit was on, it reverses the output.

This makes it hard to explain what the actual encoding is, but if you write it out it becomes clear that bits 3, 4, and 5 pretty nicely map to two digits of base 3 this way. I considered several other approaches, including a more obvious counting mechanism, but none of them seemed quite this clean. I also originally wanted to do this with lead and quicksilver instead of Ravari's wheel, but that was surprisingly difficult, so here we are. The funny result of this choice is that the solution is almost running the disposal at full rate - with all five bits on, twenty-two out of every twenty-eight cycles are used for disposal. That presented some rather funny timing problems that I had to solve to get it this fast.

As for predictions, I have a suspicion that the primary value can go under 400. I know that the way to do computation puzzles these days, especially quickly, is with a State Machine (TM), but I couldn't really think of a good use for one in this puzzle. There are so many possible approaches, though, that it can probably go a good deal below my score of 441. However, 441 is a perfect square, so I'm happy.

--- OVERALL TOURNAMENT NOTES ---

Well, this brings us to the end of my first Opus Magnum tournament. I feel like I've improved quite significantly since the beginning of the tournament, which is funny since by far my best placement was week 1 latency. My goal for next year is top 10 overall. Thanks Haxton for hosting a fantastic tournament, and thanks to everyone else for impressing me so much with a variety of ludicrously successful (and sometimes not so successful) ideas. I learned something every week, and I think this community is fantastic in its intellectual curiosity and friendliness. It's been a great couple months. See you all in the weeklies.

-Rynin