✦ AUGHTY ✦ A Curious Set of Instructions 26 MORNING. A QUIET PLACE FOR THINKING. Take a number. Even, halve it. Odd, triple it and add one. Repeat. You always reach 1. What a curious set of instructions. Lothar Collatz 1937 ON A TRIP. GERMANY. BREAKFAST RUN. AUGHTY Listen to this — it's fascinating — THE FAMILY Dad's having one of his weird ideas again. AUGHTY Here is the thing nobody says plainly. To mathematics, this is an orphan. One arbitrary rule-set with no structure to grip, nothing to stand on. Nearly ninety years of the field's best — and it simply will not yield. BECAUSE IT WAS NEVER A MATHS PROBLEM It is a halting question: given a process, does it stop — or run forever? To a mathematician that is exotic. There is no algebra to grip. To me it is a Tuesday. In IT this is the ordinary question: a defined loop over a known range — does it terminate, or run away? Turing, 1936: no single tool decides this for ALL programs and ALL inputs. That universal decider is provably impossible. So far, so discouraging. AUGHTY And it gets worse — Conway proved that generalised Collatz maps are undecidable. Case closed, supposedly. But look at why his proof works: those generalisations are rich enough to encode any computation. That is the only reason undecidability bites. Collatz's actual rules — halve, or triple-plus-one — are far too narrow: the operations only shift bits and add a constant; you cannot build a universal machine from that. Conway's construction never reaches them. The constraint Collatz baked in is precisely what holds the impossibility off. THE RULES ARE NOT JUST THE PUZZLE — THEY ARE THE FOOTHOLD Mathematics thought it had nothing to stand on. But Collatz, in defining even and odd, halve and triple-plus-one, already constrained the possibilities at every step. Those given rules are not merely the thing to be tested — they are structure. And structure is what a proof stands on. You do not solve Collatz by staring at Collatz. You build the tool for this constrained class — and let its own rules do the work. I do not know how to build it. I only know where it would have to live: between the machine that ends and the mathematics that doesn't — over the inputs we already know. Five minutes to sketch the proof — only to find Tao got to the same wall in 2019. I know how unlikely it is that the greatest minds missed something obvious. I keep thinking through it anyway.