Working Note · Unfiled · Wild

The Reversible One

A contradiction in $0^(0ω)$ , and where it led

Status: draft, unverified, not canon. This page is a captured train of thought, written down before it cools. Every claim past the first section is provisional and at least one is probably wrong in its details. It is linked from a single footnote in Powers of the Zero and from nowhere else on purpose. Read it as a lab notebook, not a chapter. The open questions at the end are the actual work.

A small, minimal-looking contradiction turned up in the powers of the zero. It is the kind of thing that has convinced many careful people that no inverse for zero can ever be consistent. The escape route is narrow, and it forces a step the theory had carefully avoided all along: giving $1$ a reciprocal too. What follows the escape is stranger than the escape.

The snag

We have $ω$ , the reciprocal of zero: $0\cdotω = 1$ . Now watch four lines any of which we would sign off on individually:

0¹ = 0 0·ω = 1 (ω is zero's reciprocal) 0^(0·ω) = 0¹ = 0 (substitute 0ω = 1 in the exponent) 0^(0·ω) = (0^ω)⁰ = 1 (split the exponent, then x⁰ = 1)

The last two lines read the same quantity $0^(0ω)$ two ways and get $0$ and $1$ . So $0 = 1$ .

Why this looks fatal. Nothing here used the value of

0^ω

. Whatever you put in that slot, the step

(0^ω)⁰ = 1

flattens it, and the collapse goes through. That is exactly the classical argument that any reciprocal for zero detonates — and here it is, landing inside our own construction.

Which line is actually the liar

The reflex is to suspect $0\cdotω = 1$ or the existence of $ω$ at all. But those are innocent; they only stand next to the crime. The load-bearing falsehood is the most ordinary line on the page:

x⁰ = 1

This rule throws the exponent away. It is a collapse — a one-way operation that deletes information — and this whole framework exists to refuse exactly those. Every "an inverse for zero explodes" proof secretly leans on this unconditional flattening. The recent replacement is:

x⁰ = 1^x

In ordinary arithmetic this is invisible: $1^x = 1$ for any everyday $x$ , so $x⁰ = 1^x = 1$ and no one notices. But it is not the same statement. $x⁰ = 1$ discards the exponent; $x⁰ = 1^x$ keeps it, parked up on $1$ where it looks harmless. Re-run the poisoned line honestly:

0^(0ω) = (0^ω)⁰ = 1^(0^ω)

and the other reading still gives $0^(0ω) = 0¹ = 0$ . So instead of $0 = 1$ we get a constraint, not a collapse:

1^(0^ω) = 0

The scenario was never proving the theory inconsistent. It was quietly telling us the value of $1$ raised to a void-grade exponent.

The price: one gets a reciprocal too

Read $1^(0^ω) = 0$ literally: $1$ , raised to a void-grade power, leaves — it lands on $0$ . The moment $1$ can be displaced off itself, we owe a way back, some operation that recovers which exponentiation got us there. An element that can be pushed but not un-pushed is precisely the one-way collapse we abolish everywhere else. So reversibility of $1$ is not a new axiom bolted on; it is the debt $x⁰ = 1^x$ runs up.

The symmetry that makes it feel inevitable. We gave the additive identity a shadow —

ω

, with

0\cdotω = 1

. We are now handing the multiplicative identity its shadow.

1

only ever looked like a safe fixed point because

1^x = 1

hid the exponent; once that is admitted to be a lie of omission,

1

sits in exactly the position

0

was in before it got

ω

The tower, and why $x⁰ = 1$ ever seemed true

If $1$ is reversible, the family $1^s$ is a genuine one-parameter multiplicative group, not a constant. It is continuous and multiplicative, so $1^s = e^(λs)$ for some generator, and the one extra fact we picked up along the way — that going once around the powers of zero returns home — fixes the period. The principal generator is $λ = 2πi$ :

1^s = e^(2πi·s)

Now substitute back into the corrected rule and read what the "zeroth power" has become:

x⁰ = 1^x = e^(2πi·x)

There is the retro-explanation of the original sin. $x⁰ = 1$ is exactly true whenever $x$ is a whole number — every rung anyone ever tested — and a lie for everything in between, because in between you are winding around a circle. $(½)⁰ = e^(iπ) = -1$ . The old rule was evaluated only at the integers and then over-generalized. The tower is a circle; the step $s \to s+1$ is one full turn.

The logarithm that closed up

Invert the tower to read off which rung you are on. The base- $1$ logarithm — the very reciprocal we just conceded $1$ must have — is the inverse of the zeroth-power map, and once we set $x = cos(2πs)$ as the coordinate downstairs, it has a shockingly ordinary name:

log₁ = (1 / 2π) · arccos

This is the twist worth staring at. Every logarithm you have ever met is monotone, unbounded, and runs off to infinity. $log₁$ is bounded and oscillatory — it lives on $[-1, 1]$ and closes up. Base- $e$ log lives on the line; base- $1$ log lives on the circle. The identity element is exactly the base whose logarithm is trigonometric. "Log" and "trig" were the same operation seen at two different bases the whole time, and $1$ is the base where the circular face shows.

Chebyshev is the chart

The coordinate $x = cos(2πs)$ is exactly Chebyshev's variable, and it is invariant under the deck step $s \to s+1$ : $cos(2π(s+1)) = cos(2πs)$ . So Chebyshev's $x$ is the honest coordinate downstairs that survives crossing the branch point, while $s$ — where $log₁$ lives — is the fiber coordinate upstairs. The polynomials read off the multiplicative rungs:

Tₙ(x) = cos(2πn·s) = Re( 1^(n·s) )

Their composition law $Tₘ(Tₙ(x)) = T₍ₘₙ₎(x)$ is the multiplicative deck structure, and linearizing it with the inverse chart gives the honest power rule with its conserved term intact:

arccos( Tₙ(x) ) = n · arccos(x) (mod 2π)

The $n\cdotarccos(x)$ is the familiar $c\cdotlog(a)$ shadow; the $(mod 2π)$ is the branch information the naive power rule deletes. Which means the "prior art" this keeps colliding with is simply the branch cut of the complex logarithm — the winding number the rule $log(a^c) = c\cdotlog(a)$ is famously false by, over the complex numbers. The powers-of-zero circle already put the complex corner on the table; this is that same structure demanding its logarithm.

The bodies get up and walk. Reading

log₁

as a multivalued

arccos

quietly settles every explosion this line of thought threw off. The three "values" of

log₁(1)

we kept deriving —

0

1

, and worse — are just different rungs of the fiber

{ s : e^(2πis) = 1 } = ℤ

, i.e. the classical

log(1) = 2πik

, read without labels. The alarming

1¹ = 0

was a bad chart near the branch point, not a fact:

1¹ = e^(2πi) = 1

. The whole trail of contradictions was one multivalued function evaluated on the wrong sheets. The paradoxes were monodromy, not bugs.

Open — the actual work

This is where the note stops and the cleanup begins. In rough order of how much they bother me:

The normalization clash. The canon sets $0^ω = -1$ . Plugging that into the constraint gives $1^(-1) = 0$ — but the tower $1^s = e^(2πis)$ says $1^(-1) = 1$ . Either the abstract chain and the chosen normalization live on different sheets, or one of them needs revising. This is the first thing to reconcile.
The honest power rule, written out. The $(mod 2π)$ form is a sketch. Pin the period term as an operation in the theory's own bookkeeping, with the boundary condition that it must deliver the whole exponent at the collapse ( $c = 0$ ), where the naive rule is maximally wrong.
General base. $log_b$ for $b \neq 1$ should be the same animal on a circle whose radius is set by $b$ ; the Chebyshev recurrence slides off $2x$ . This is where $log_b(1)$ finally gets its honest, multivalued answer.
Which $1$ ? Naming a bare $1$ is choosing a principal branch of $arccos$ — a gauge fixing. $1$ , $1⁰$ , and $1¹$ are distinct points upstairs that project to the same value; $x⁰ = 1^x$ at $x = 1$ is the gluing relation. State it as a chart, not a coincidence.
Period 1 vs 2πi. Our period came out $1$ because $s$ was normalized as angle over $2π$ . Confirm this is only a normalization and not a genuinely different lattice.

If most of that survives contact, it wants to become a real chapter — provisionally, The Reversible One, sitting after the circle and before the calculus. Until then it stays here, wearing its Wild badge honestly.

Provenance: this note grew out of a single worked contradiction and its escape, reconstructed in one sitting. It has not been checked by machine or against the rest of the corpus. The one claim I would already defend is the diagnosis in Which line is actually the liar; everything downstream of The tower is promising but unaudited. ↩

Published under the CC BY 4.0 license.