SLA · Spatial Layer Architecture
Benchmark ArenaOpen ChallengeEnter the Arena →
P vs NP · 1971 → 2026

In 1971, Stephen Cook
asked the wrong question.

Cook built the P vs NP problem inside a single unexamined assumption — that computation is sequential, memory is a flat line, and problems are solved by walking that line one step at a time.

He never questioned it. In 1971, no one did. Every computer ever built worked that way.

So the question became: can a sequential machine solve NP-hard problems in polynomial time?

Fifty-five years later — the answer is probably no.

But that was never the right question.

See the Results →Enter the Challenge
The Flaw

Cook's framework assumed
the only kind of computer that existed.

What Cook assumed
✗Memory is a flat sequential list
✗Problems are solved by walking the list
✗Distance = counting steps
✗Computation moves in one direction
✗The address space has no shape
What SLA proposes
→Memory is a three-dimensional coordinate space
→Problems are navigated geometrically
→Distance = sightline proximity
→Computation reads spatial relationships
→The address space has elevation, peaks, a floor
Cook's P vs NP question was asked about a machine that processes the world sequentially.
SLA describes a machine that inhabits the world spatially.
Those are not the same machine. And P vs NP was only ever asked about the first one.
The Evidence

Real datasets. Verified benchmarks.
The signal gets stronger at scale.

Every standard algorithm gets relatively worse as the problem grows. That's the law of NP-hard problems. SLA goes the other direction. The geometry reads the structure more clearly when there's more structure to read.

Berlin5252 cities
+0.031% from optimal
KroA100100 cities
+10.27% from optimal
↓ Gap shrinks at 1,577 cities — signal gets stronger
FL15771,577 cities
+3.395% from optimal
Time complexity
O(n¹·⁸⁸)
Sub-quadratic. Polynomial.
FL1577 accuracy
96.6% of perfect
1,577-city real dataset
Berlin52
2.37 units
From known optimal. Essentially solved.
The Architecture

What SLA actually is.

🏗️
Not a flat line
The address space is a three-dimensional irregular polygon boardwalk. Peaks push outward. Valleys pull inward. Every address has elevation, angle, and depth.
👁️
Sightline distance
Two addresses that are numerically far apart may be geometrically adjacent — visible to each other across the interior void. Distance is proximity, not counting.
🔬
MRI geometry
Large problems are sliced along their principal axis like a CT scan. Each slice ordered by SLA elevation. Macro structure from the scan. Micro structure from the geometry.
📈
Scales up, not down
Standard algorithms degrade at scale. SLA improves. At 52 cities: 0.031% from optimal. At 1,577 cities: 3.395% from optimal. The geometry reads what's already there.
SLA was not designed in a laboratory. It was not derived from existing computer science literature. It was encoded from one person's first-person cognitive experience of space and number — a perception that has operated continuously since childhood. The geometry existed in a human mind for decades before it was translated into computational terms in May 2026.
The Open Question

Can a geometric machine
do what a sequential one cannot?

We're not claiming to have solved P vs NP. We're claiming that the question was built on an assumption that deserves to be challenged — and that the geometry we've produced is the strongest empirical argument for challenging it that has ever been published.

The benchmark arena is open. The data is public. The challenge is live.

Beat us.

Enter the Arena →View the Challenge
© 2026 UnityFlag LLC · Spatial Layer Architecture · USPTO 2-01026094 (May 26) · Continuation 2-01029681 (May 27)
Benchmark ArenaOpen Challenge