PhysWallThe Physical-Logic Gateway

An AI gave you a number. Can your data carry it?

A physics engine, and a verification tool enforced on top of it. PhysWall inverts a closed, published, non-linear physical law at a single measurement point — and refuses when the inverse is not unique. Seven laws, one engine, and the same refusal in all of them.

home the extension the tools about which tool do I need? support

PhysWall

Try it — free

Paste the answer and what it was based on. We tell you how far it holds — and what would take it further.

The model said

“Your conversion rate improved by 30% after the change.”

PhysWall

The arithmetic doesn’t hold here.

At a 4% rate and 40 visitors, fewer than two conversions are expected. Every significance test rests on a normal approximation, and that needs about five. Nothing can be concluded at this size — not even “no change”.

Get to 4,783 per group and the question becomes answerable.

Not “wrong”. Not yet. And now you know what “yet” costs.

Four answers. All of them point forward.

Three of the four are refusals — and that is the product, not the caveat. A tool that always returns a number is a tool that cannot tell you when there isn’t one.

It holds.

→ Build on it. One value fits, and only one.

Not yet. NOT UNIQUE

→ One more measurement — and we say which.

Not at that target. NOT ACHIEVABLE

→ Move the target — and we name the law that sets the ceiling.

Something upstream is off. IMPLAUSIBLE INPUT

→ Check the reading. The answer is outside what is possible.

Twelve tools. Same four answers.

Whether you are checking a figure from a chatbot or a calibration lab facing an audit — the same four come back.

Start here · any number
Can your data carry the answer?
Not sure which · ask in plain words
Describe it, and we point you
Finance
What growth a share price already assumes
Streaks
Hot hand, or sample size?
RF & antennas
Seven published limits, run backwards
River flow
Four rating curves, 3.4 ft apart
Competing explanations
Which ones the evidence leaves
Waves
Where both textbook shortcuts fail
Climbing
Average power, from a stopwatch
Light time
Distance from round-trip alone
Network flow
Counters against sampled traffic
Your own law
Bring a formula, we check the inverse

Nothing here is ours to be wrong about.

Every law was published by somebody else and is named on the page that uses it. Bode 1945 · Fano 1950 · Airy 1845 · Manning 1889 · Landauer 1961 · Heuer 1999 · Miller & Sanjurjo 2018.

No measurements of our own. Zero. One tool is held back because it has never seen a real reading, and its page says so instead of showing a number.

And where each law stops: Bode-Fano bounds passive networks and does not apply to an active or tunable match. Manning is an empirical fit, not a derivation. Every page states the limit of the law it uses.

Bring us a number.

Yours, or one a model gave you. Free, open in a browser, nothing to install.

Try it now → How it was built →
(function(){ var page = 'start'; function send(o){ try{ o.page = o.page || page; var b = JSON.stringify(o); if(navigator.sendBeacon){ navigator.sendBeacon('/api/count', new Blob([b],{type:'application/json'})); } else { var x = new XMLHttpRequest(); x.open('POST','/api/count',true); x.setRequestHeader('Content-Type','application/json'); x.send(b); } }catch(e){} } send({kind:'view', ref: document.referrer || ''}); var t0 = Date.now(); document.addEventListener('click', function(e){ var a = e.target && e.target.closest ? e.target.closest('a') : null; if(!a) return; var h = a.getAttribute('href') || ''; if(h.indexOf('/app/')===0 || h.indexOf('/tools')===0){ send({kind:'cross'}); } }); window.addEventListener('pagehide', function(){ send({kind:'dwell', s: (Date.now()-t0)/1000}); }); })();