Collatz in plain language

Start at 7. Seven is odd, so apply 3n+1: 3 times 7 is 21, plus 1 is 22.

Twenty-two is even, so halve: 11.

Eleven is odd: 3 times 11 is 33, plus 1 is 34.

Thirty-four halves to 17.

Seventeen is odd: 3 times 17 is 51, plus 1 is 52.

Fifty-two halves to 26. Twenty-six halves to 13.

Thirteen is odd: 3 times 13 is 39, plus 1 is 40.

Forty halves to 20. Twenty halves to 10. Ten halves to 5.

Five is odd: 3 times 5 is 15, plus 1 is 16.

Sixteen is a pure power of two, so the rest is obligatory descent: 8, then 4, then 2, then 1.

From 1 the rule sends you to 4, then 2, then 1 again. That cycle is the home the conjecture claims every positive integer eventually enters. If you have never walked it by hand, do it once on paper. The claim stops being abstract when your wrist has done the arithmetic.

Even in this tiny example you can feel the two moods of the map. Odd steps often lift the value. Even runs drain it. Sometimes the drain is short. Sometimes, as with 16, the drain is a clean collapse. On larger starts the lift can go on long enough that a beginner wonders whether the floor will ever return. Then a cascade of halvings arrives and the value crashes toward smaller terrain, though not always below the start on the first try.

People call these trajectories hailstone sequences because they rise and fall like hail in a cloud before falling to earth. The metaphor is cute and surprisingly durable. It also risks making the motion sound friendly. Hail can break windshields. A sequence can climb into thousands, millions, far beyond casual mental math, before it descends. Twenty-seven is the classic classroom bully: it reaches 9,232 before it comes down, and it takes dozens of steps before it gets below its start. If 7 is a short story, 27 is a novella.

Here is what is known in the coarse public sense I allow myself in a plain entry. No general proof that every positive integer returns. No counterexample. Massive computational checks covering ranges so large that ordinary intuition about searching fails. Related theorems about density, average behavior, and almost-all orbits. A living literature. An unsolved core.

The core question is not whether the rule is interesting. It is. The core question is whether every path terminates at the 4-2-1 cycle. Related properties can be progressed while that question remains open. That split matters. Progress is real. Completion is not. If you blur them, you will misread both the math community and your own notes.

Collatz sits at an edge where short rules produce unpredictable-looking trajectories. Dynamical systems language helps. Number theory language helps. Stopping-time language helps. None of those languages, by themselves, closes the conjecture. Tools are not trophies.

Paul Erdős, who lived inside hard problems the way other people live inside cities, is often quoted to the effect that mathematics may not be ready for such problems. Whether you meet that line on Wikipedia, in a talk, or in the folklore of math twitter, the function of the line is a warning, not a movie trailer. I treat it as a warning.

There is also lore about prize money, on the order of five hundred dollars, attached to the difficulty signal. The dollar amount is almost comic next to modern prize culture. That is part of why it works as lore. The point was never that Collatz is a career liquidity event. The point was that a mind like Erdős wanted the difficulty marked in public with a gesture.

I am not Erdős. I do not wear the quote as a dare to overclaim. I wear it as permission to go slow and to distrust my own appetite for climax. If mathematics as a whole may not be ready, then a founder with a notebook should be twice as careful about what he announces.

Terence Tao proved that almost-all Collatz orbits attain almost bounded values, in a precise technical sense you should read from him if you want the real statement. Plain version for this notebook: almost all is not all. He moved a mountain range in the almost-all world. The full conjecture still asks about every positive integer, including any monsters hiding in thin sets that density language can overlook if you get sloppy.

Density results are not participation trophies for the conjecture. They are real mathematical achievements that redefine what seriousness looks like. They also create a communicative hazard for amateurs: you read almost and your emotional brain hears done. I have felt that hazard in my own chest. The discipline is to say the word almost out loud until the emotional brain sits down.

Computational checks create a sibling hazard. A machine can verify a vast initial segment of the integers and find no counterexample. That is important empirical knowledge. It is still not a proof for the infinite remainder. The integers do not owe you that the next block will behave like the last block. They often do behave. Proof asks for necessity, not habit.

Hailstone sequence: the list of values visited from a start under the rule.

Total stopping time: how many steps until you reach 1, when you do.

Syracuse map: a compressed view that folds the even drains into the odd steps, often used when people want to stare at odd-to-odd dynamics.

Stopping time ideas: questions about when a sequence first drops below its start, distinct from the full trip to 1, and historically tied to density results in the Terras neighborhood and beyond.

I list these not to cosplay a textbook, but to mark the words I refuse to use as fog. If I say Syracuse later, I should mean the compressed dynamics, not a spell. If I say density, I should mean a specific notion of almost all, not a vibe of mostly fine.

Jargon can be precise. Jargon can also be perfume on a weak claim. On January 10 I forced myself to explain the problem the way I would explain it to a smart friend at a table with no whiteboard markers. If the sentence only works when it wears symbols like armor, maybe the thought is not ready.

Plain language also protects the later architecture from premature worship. Look-ahead, energy, obstruction, Omega notations that show up in research notes: those need definitions and citations. They do not need to enter a first explanation of Collatz as if they were part of the original rule. The rule has no look-ahead. The rule is even and odd. Everything else is human scaffolding.

I will say this once in this entry because the project needs it on record: I do not claim that look-ahead thinking or Omega-style bookkeeping is a novel discovery of mine. The broad neighborhood has prior art. Terras-type stopping and density ideas are part of the landscape. My work, when it is work, is local and accountable: particular formalizations, particular failures, particular small results. Plain language keeps me from sounding like I invented winter.

Tao has also spoken to the danger the problem poses for people who try to solve it. Obsession, false proofs, years lost to a tunnel with no exit signs. I am one of those people, carefully. Not as a prize hunter. As a student who needed a clean problem and an honest notebook.

Carefully means I keep the wall: I have not closed it. Carefully means I read before I brand. Carefully means I let negative results kill strategies without killing curiosity. Carefully means I remember Las Vegas bills and Brazilian family and faith practices that do not revolve around my intellectual romance. The problem is allowed to be major. It is not allowed to become the only room in the house.

Student posture also means I accept humiliation as tuition. I will misunderstand a paper. I will reinvent a wheel and then find the wheel in a 1970s reference. I will write a note that future me deletes for being wrong. The plain-language entry is a place to return when the research notes get loud. When in doubt, come back to 7. Walk it. Remember what is actually being asked.

Try 9 quickly. Nine is odd: 28. Twenty-eight halves to 14, then 7. Suddenly you have joined the road you already walked. That joining is part of why the system feels like a watershed map. Many starts pour into shared downstream segments. The conjecture says all watersheds eventually reach the same sea cycle. Seeing merges can trick you into thinking the global claim is obvious. It is not. Shared downstream roads are compatible with both universal return and with a hidden rogue river that never merges. Computation has not found the rogue river. Proof has not ruled it out.

Try 27 if you have an hour and a willingness to be patient. Or trust the well-known peak at 9,232 and the long climb, then verify a portion yourself. The point is to feel that plain rules can generate long weather. Without that feeling, almost-all theorems sound like bureaucratic fine print. With that feeling, they sound like hard-won light in a storm that still has unmapped cells.

A philosophical reading about return can coexist with an honest math posture. It cannot replace the missing lemmas. That sentence is the hinge between this plain entry and the rest of the project. If you only want the conjecture, you can stop at the rule, the walk of 7, Erdős as warning, Tao as almost-all mountain, and the open core. If you want to know why I stay, the other entries cover motives and origin and later technical bets.

I will not end this with a sunrise. I will end with the assignment I gave myself on 2025-01-10: be able to state the problem without lying, walk an example without theater, name the best-known warnings without trying to outrank them, and keep almost separated from all as if my intellectual soul depended on it. Because it might.

Objection: if every tested number works, it is true. Response: tested is not proved. The integers are not a finite product you can QA through a release checklist.

Objection: just use a computer for bigger ranges. Response: bigger ranges strengthen empirical confidence and still leave an infinite unchecked set. Also, a counterexample might be unimaginably large. Computation is a flashlight, not a sun.

Objection: it is probably true, so move on. Response: probable is a stance, not a theorem. I am allowed to believe it is plausible while refusing to launder plausibility into closure.

Objection: why waste time on a problem that might be unready. Response: unreadiness is a warning about tools and arrogance, not a ban on careful study. Study can mean mapping failed strategies and learning the literature's edges. That is not a waste if you tell the truth about outcomes.

Objection: you are not a professional number theorist. Response: correct. That is why the posture is student, the claims stay narrow, and the notebook stays allergic to coronation language.

A cheap heuristic everyone meets early: odd steps multiply by roughly 3, even steps divide by 2 enough times to compensate on average. If the average number of factors of 2 after a 3n+1 step is high enough, orbits should drift downward in some statistical sense. Heuristics of that family help explain why people expect the conjecture to be true. Heuristics also fail as proofs because the worst paths may refuse to look like the average path.

That is the whole amateur tragedy in miniature. You glimpse a drift. You feel enlightened. You write a note that accidentally assumes every orbit will sample the average. Then you meet families engineered, or at least observed, to produce ugly streaks of shallow collapses. Later work in my project collides with that ugliness hard. This plain entry only needs the seed: average intuition is not universal control.

If you want a practical exercise, write down the odd terms of the walk of 7: 7, 11, 17, 13, 5, then 1 after the power-of-two collapse finishes its job. Watch how the 3n+1 lifts are interleaved with different lengths of halving. The variation in those lengths is the weather. Proof would need a coat that works in all weather, not a postcard from a sunny day.

Every few weeks I force a reset. Close the research notes. Open a blank page. Write the rule in one breath. Pick a number I have not romanticized. Walk it until 1 or until I have learned something about my patience. If I cannot do that without reaching for a private trademarked framework, I have drifted from Collatz into shrine maintenance.

On 2025-01-10 the reset was the point of the day. I was early enough in the project that drift was already possible. Names arrive fast when you are hungry. Plain language is how you weigh them. If a name cannot survive a dinner table, maybe it should not lead your private notes either.

This is also why I keep returning to 7 even though 27 is the famous long climb. Seven is short enough to do from memory and long enough to include both lifts and drains. It is a tuning fork. When my explanations get ornate, I strike 7 and listen for whether I still hear the original pitch.

The cycle 4, 2, 1 is not controversial as an object. It is there. Apply the rule inside it and you stay inside it. The conjecture says this is the unique attractor for all positive integers under the map. In the wider Collatz universe people also study other cycles in variants, negative numbers, different multipliers, and related maps. For the classic positive-integer claim, the fear is less about a second cute small cycle we somehow missed in the single digits, and more about either a divergent trajectory or a monstrous cycle hiding far away.

I am not going to pretend I can sketch a professional impossibility proof for small alternate cycles in this plain entry. Others have done extensive checking. My point is narrower: when I say home, I mean the known cycle, not a vague spiritual rest. Arithmetic home is specific. Specificity is what makes the conjecture bite.

If a divergent trajectory exists, homecoming language fails in the math lane and becomes, at best, a story about hope under incomplete knowledge. If every trajectory returns, homecoming language stays aligned without becoming a proof. Either fork demands honesty. Plain language keeps the forks visible.

Watching a dozen orbits reach 1 is not understanding. Repeating that computers checked huge ranges is not understanding. Quoting Erdős without feeling the warning is not understanding. Quoting Tao without respecting the gap between almost and all is not understanding. Understanding, at my level, looks more like being able to explain the claim, the open status, the best public landmarks, and the failure modes of the heuristics I am tempted to overbelieve.

I also refuse to count vibes as understanding. The sequence looks alive. The sequence looks intentional. The sequence looks like a parable. Those vibes may fuel attention. They do not constrain integers. If I ever write a sentence where a vibe does logical work, I want a future reader to strike it with a red pen. Including me.

Faith belongs in my life. Faith does not get to finish Collatz. I can thank God for curiosity and still owe the literature a clean citation. I can pray for patience and still need a lemma. Plain language is one of the ways I keep those categories from performing identity theft on each other.

If you continue in the notebook you will meet look-ahead, energy scores, ugly windows, Mersenne odd starts, obstruction language. None of that is required to know what Collatz is. I am foreshadowing lightly only so you do not think those tools are secret ingredients inside the original conjecture. They are attempts. Some die. Some narrow. Some teach.

Mersenne numbers show up later as stress tests in my technical notes because numbers of the form 2^p - 1 can produce stubborn patterns of shallow collapse under certain analyses. You do not need that today. Today you need even and odd. If the later pages intimidate you, come back here. If the later pages intoxicate you, come back here faster.

Density language says that the proportion of integers with a property can approach 1 even while infinitely many exceptions might exist, depending on the property and the notion of density. That is why almost-all theorems can be both gigantic and incomplete for a universal claim. Imagine a room that fills with green balls until the red ones are rare beyond ordinary sight, and yet the red ones never finish disappearing. Universal quantification cares about every red ball. Density can celebrate the green ocean.

I use that crude picture knowing professionals would refine it. Refinement is good. Crude pictures are how non-specialists avoid a worse error: hearing almost all as all. If this entry does nothing else, it should make that mistake harder to commit while reading Going Home.

When I say the gap is the mountain, I mean the logical distance from density-one success to success for every positive integer. Mountains are not insults to the climbers who reached high camps. Tao's high camp is extraordinary. Base camp for the full summit is still not the summit. I live most days at a desk far below both, which is exactly why I need plain speech about altitudes.