QWERTY vs Dvorak vs Colemak: should you switch layouts?

If you type on QWERTY and you keep reading that Dvorak or Colemak would make you faster, the short answer is that nobody has ever demonstrated it convincingly, and the people who tried had reasons to want a particular result.

That is not an argument against switching. There is a real case for the alternatives, it is just a different case from the one usually made, and it rests on something you can check rather than something you have to take on faith.

What the three layouts actually change#

All three put the same 26 letters on the same physical keys. What changes is which letter sits where, and the row that matters is the middle one, where your fingers rest between keystrokes.

QWERTY came off a typewriter and was never arranged around English. August Dvorak and William Dealey patented theirs in 1936 and rebuilt the letter map from scratch. Colemak arrived in 2006 with a narrower brief: move as few keys as possible and still fix the worst of what QWERTY does to your hands.

That leaves you three options, and what separates them is how much you would have to relearn:

• QWERTY. Nothing to relearn, nothing to install. Whatever else it is, it is already everywhere.
• Colemak. 16 of the 26 letters move. Z, X, C and V stay put, so cut, copy and paste still work under one hand.
• Dvorak. 24 of the 26 letters move: vowels to the left hand, the common consonants to the right, and the shortcut cluster scatters.

That last column is the part worth holding on to. On QWERTY, 32.5% of the letters you type are already under a resting finger. On Dvorak it is 70.4%, and on Colemak 73.9%. Those are not marketing numbers. They come from multiplying each layout's home row against the letter frequencies Peter Norvig published from 3.56 trillion letters of Google Books text, and you can reproduce them with a spreadsheet in ten minutes.

So the alternatives really do move your hands less. Roughly half your typing shifts from a reach to a resting position. The open question has always been what that buys you.

The other number you can check: same-finger pairs#

Home row placement is half of what the alternative layouts are designed around. The other half is avoiding pairs of letters that land on the same finger, because your hand has to lift and return before it can type the second one. That is the slowest thing a keyboard can ask of you.

We ran the same kind of calculation. Take the fifty most common two-letter sequences in English, again from Norvig's corpus, and check which of them ask one finger to type both letters under standard touch-typing fingering.

All three of QWERTY's offenders sit on the left middle finger. Type the word “edited” and you can feel it: one finger doing three separate journeys while seven others wait.

Read that table for what it is. Fifty bigrams cover 55.6% of English two-letter sequences, so this is a floor rather than a full count, and the alternatives are not actually at zero across the whole language. What it does show is that the two redesigned layouts clear the most common cases completely, and QWERTY does not clear them at all. If you want the mechanics of why this pattern costs you time, we go into it in same-finger digraphs.

Why the speed evidence is so weak#

The claim that Dvorak is faster traces back to Dvorak himself. During the Second World War he ran Navy experiments which he said showed typists could be retrained in ten days. He had also run earlier studies whose results were never published. A researcher testing his own patented invention, and keeping the run that worked, is not evidence anybody should build a decision on.

The counter-study has the same problem in the other direction. In 1956 the US General Services Administration ran a comparison under Earle Strong and concluded Dvorak was no more efficient, and that retraining would never repay its cost. That finding kept Dvorak out of government offices for decades. It has also been criticised as biased toward the QWERTY group, and when other researchers asked to inspect the raw data, it had been destroyed.

Both of those studies, and the rest of the record, were examined in “The Fable of the Keys”, which Stan Liebowitz and Stephen Margolis published in the Journal of Law and Economics in 1990. They took up QWERTY as the standard example of a market locking in an inferior technology, went back through the experiments, and concluded that the best-documented ones, along with the ergonomic studies available then, showed little or no advantage for Dvorak.

Two things follow. The first is that the confident percentages you will find on other comparison pages, the 5% here and the 15% there, are not traceable to a study anyone can name. The second is that a genuinely fair, large, modern trial of layout against layout has still not been run. “Unproven” is the honest word, and it points in both directions.

What we do know about what makes people fast#

The largest picture of real typing comes from a 2018 study at Aalto University and the University of Cambridge, which analysed 136 million keystrokes from 168,000 people. It did not compare layouts. It did establish what separates fast typists from slow ones, and the answer was rollover: fast typists start the next key before finishing the last, so their keystrokes overlap.

That matters here because rollover is a coordination skill, not a geography problem. Moving the letters closer together shortens the distance your fingers travel. It does not, by itself, teach your hands to overlap. Somebody who types 45 WPM on QWERTY and hits every key as a separate deliberate event has no particular reason to expect a different number on Colemak once they have relearned it.

What those numbers are actually worth#

Less travel and fewer same-finger pairs make a comfort argument before they make a speed one, and it is the argument the alternatives can actually support. Fewer long reaches means fewer awkward stretches for the little fingers, and less of the row-jumping that makes a hand rock back and forth.

Be careful how far you take it. There is no clinical trial showing that switching layouts prevents or resolves repetitive strain injury. The layout communities are full of people who report their hands feel better, and that testimony is worth something, but it is testimony rather than evidence, and people who have just invested a month in a change are not neutral witnesses to whether it worked. If you have pain, the first stop is a clinician and your desk setup, not a keymap.

What switching actually costs#

You lose your typing for a while. Every established typist reports the same shape: speed collapses immediately, comes back over weeks, and the last stretch back to your old number is the slow part. How long depends on how fast you were, since a 100 WPM typist has more to rebuild than a 50 WPM one. Anyone quoting you a precise number of weeks is quoting a feeling.

There are smaller costs that people underestimate:

• Every keyboard that is not yours stays QWERTY. Shared machines, a colleague's laptop, a login screen before your layout loads.
• Shortcuts move on Dvorak. Every cut, copy and paste goes through a hand that has to think about it again.
• Maintaining both is a skill of its own. Some people keep QWERTY and Colemak in their hands at once. Others find the second layout erodes the first.

So which layout should you choose?#

Switch if typing is most of your working day, your technique is already sound, and reaching for the far keys is what tires you out. In that position you are paying the relearning cost once against years of use, and the comfort case is the one that holds up.

Stay where you are if you still look down to find keys, or if corrections take up a real share of every run. Your technique is what is limiting you at that stage, and a second layout on top of an unfinished first one means learning both at once. Get to the point where you can type without looking, then ask the question again; the argument in full is in touch typing against hunt and peck.

Past that, the choice comes down to what you are optimising for:

• Borrowing machines all day, or locked out of installing anything on them: QWERTY.
• Decided to switch, and want the shortest route back to your old speed: Colemak. Fewer keys change position, so the muscle memory you keep is muscle memory you do not have to rebuild.
• Want the rearrangement done thoroughly, and can live without the shortcut cluster: Dvorak.
• Already touch typing on QWERTY with no complaint about your hands: stay put. Nothing in the record says you are leaving speed on the table.

Colemak's own site claims your fingers travel 2.2 times further on QWERTY. That is the designer's figure rather than an independent one, but it points the same way the home-row calculation above does.

The thing to check before you decide#

Most people asking this question have diagnosed the wrong bottleneck. Before you commit a month, find out what is actually slowing you down.

Take a timed typing test and read three numbers rather than one. Speed tells you where you are. First-attempt accuracy tells you whether you are losing time to corrections. Consistency compares your speed word by word, so a low score means bursts with hesitations between them. It is not a rollover measurement, which would take timing between individual keystrokes, but stopping to find a key is what missing rollover looks like at the scale a test can show you.

Then look at which keys are actually costing you. If your errors cluster on a handful of letters, or on pairs typed with the same finger, that is a technique problem, and it is the problem the alternative layouts are designed to solve by moving the letters. Fixing it on the layout you already know is cheaper than relearning one, and it is the same underlying skill either way.

Keep the two claims apart when you decide. The alternatives do move your hands less, and that part is measured rather than argued. Whether they make an experienced typist faster is unproven, and nobody has run the trial that would settle it. Switch for the comfort and you are on firm ground. Switch for the speed and you are spending a month of practice on a claim no one has demonstrated.