monkeystyping

Give a few million monkeys a real keyboard. Pick a phrase. Find out exactly how long you would be waiting — and whether QWERTY was ever a good idea.

Loading the simulator…

Expected time for the whole troop

—

—
—
—
—
The troop
The keyboard
Run
Simulated time
—
Characters typed
—
Key strikes
—
Actual speed
—

The troop

One cell per monkey — every one of them is really being simulated. Dark means nothing doing. The brighter a cell, the more of your phrase that monkey has just typed by accident. The bright flickers are near misses.

Best anyone has managed

—

The front row

Eight of them up close, live. Drop the speed back to 1× if you want to read along.

Where that number comes from

This phrase

Length
—
Probability at any one spot
—
Expected characters, one monkey
—
Expected key strikes
—
Overlap set
—
Overlap penalty
—
Median time
—
90% certain by
—

Where the difficulty is

Each character's probability on this keyboard. Taller is rarer.

What a bigger troop buys you

The keyboard they are bashing

Strikes that emit a character
—
Strikes that hit Shift
—
Strikes on a dead key
—
Strikes that miss entirely
—

So is QWERTY any good?

The monkeys are the excuse. This is the actual question. Same keyboard, same key sizes, same everything — just the letters moved around. Each arrangement gets judged twice: how easily a monkey blunders into your phrase, and how a person would fare typing English on it. Green is the winner in each column.

LayoutMonkey difficultyHome row TravelAlternationSame fingerEffort

The paper

The whole thing written up properly: how the model works, the proofs, two independent checks that the maths is right, and a straight account of what it gets wrong. It starts gently and gets progressively more serious — read as far as you like. Rendered live from paper/monkeys-typing.md, so there is no second copy to drift.

Loading the paper…