Dvorak, Colemak and Mechanical Keyboards: The Honest Math

Two questions come up constantly among people trying to type faster, and both are usually answered with enthusiasm rather than arithmetic. Should you switch to Dvorak or Colemak? And does a mechanical keyboard actually help?

Both questions have a genuine answer. It is just that the answer requires doing a calculation almost nobody does — comparing what the change costs against what it plausibly returns, and then asking how confident we can be that the return exists at all.

This article does that calculation for both. It also examines where the widely quoted numbers on each side come from, which turns out to be the most interesting part: in both cases the confident figures circulating online rest on considerably less than their confidence implies.

The payback question nobody asks

Any change to how you type has the same structure. You pay an upfront cost in lost productivity while you adapt, and you receive a per-day return afterwards. Whether the change is worth making is a break-even problem, and it can be written down.

The asymmetry is what makes this decision hard. The cost is large, certain and immediate. The gain is small, uncertain and spread across years. Even generous assumptions about the benefit often fail to justify the change, and that is before asking whether the benefit is real.

One input deserves emphasis, because it is where most estimates go wrong: the return applies only to time spent actually producing characters, which in most jobs is a small fraction of the day. The same point undermines a lot of typing advice, and it is worked through in what actually saves time at a keyboard.

What a layout switch actually costs

The transition is the part people underestimate, partly because the early progress is encouraging and the later progress is not. Basic function returns quickly; your previous speed takes considerably longer.

Colemak is the gentler of the two, and deliberately so. It leaves a number of letters in their QWERTY positions and preserves most common keyboard shortcuts, which reduces both the relearning burden and the daily friction of using software built around QWERTY conventions. The design rationale is set out on the Colemak project's own FAQ.

There is also a cost that no timeline captures. During the transition your typing occupies conscious attention again, which is precisely the resource you were trying to free up. For anyone whose work involves thinking while writing, that is a larger tax than the raw speed loss suggests.

Does Dvorak actually make you faster?

This is where the story gets genuinely interesting, because the popular account and the evidentiary record diverge sharply.

The widely repeated claim is that Dvorak is substantially faster, and that QWERTY survives only through historical accident. The primary support for the first half comes from studies conducted by August Dvorak himself, who held the patent — a conflict of interest that would disqualify the work in most fields.

A 1956 study by the US General Services Administration, led by Earle Strong, compared retraining typists on Dvorak against simply retraining them on QWERTY, and found no advantage to the switch. Economists Stan Liebowitz and Stephen Margolis later reviewed the whole literature in The Fable of the Keys, concluding that the evidence for Dvorak superiority was far weaker than its reputation, and that the famous wartime Navy results were poorly documented.

In fairness, the critique has its own critics: some supporters of Dvorak argue Strong was hostile to alternative layouts before he began. The honest summary is not that Dvorak has been disproven but that it has never been convincingly demonstrated, which is a different and more uncomfortable position. Nearly seventy years on, there is still no large, well-controlled, independent study showing a clear speed advantage.

The related claim that QWERTY was designed to slow typists down is also a simplification. The arrangement addressed mechanical jamming by separating letters that were frequently typed in sequence, which is a different objective from making the typist slower. Background on the layout's development is summarized on the Dvorak layout's overview.

Why less finger travel does not mean more speed

Here is the part that resolves the apparent contradiction, and it is the single most useful idea in this article. The efficiency claims for alternative layouts are largely true. They simply do not imply what people assume.

Colemak genuinely keeps far more typing on the home row than QWERTY, and genuinely reduces the frequency of awkward same-finger sequences. Those are measurable properties of the layout, computable from any body of English text, and they are not in dispute.

But finger travel is not what typing time is made of. A skilled typist's hands are not moving sequentially from key to key waiting for each journey to finish — multiple fingers are in motion at once, the next movement begins before the current one lands, and the limiting factor is the coordination of that overlapping sequence rather than the distance any single finger covers. Reducing total travel by some percentage does not reduce total time by that percentage, because the travel was substantially overlapped in the first place.

This also explains an observation that otherwise looks strange: the fastest typists in the world are overwhelmingly on QWERTY. If the layout were a meaningful constraint, the top of the distribution would have migrated long ago. It has not, which tells you the binding constraint is the typist rather than the arrangement of the letters.

The break-even table

Putting numbers to the formula makes the decision concrete. The scenario below assumes a 60 WPM typist who produces one hour of actual keystrokes a day — already a high estimate for most office work — and who loses about 40 percent of their output during an eight-week transition.

Read down the verdict column and the shape of the decision is clear. Only the bottom two rows justify the switch on speed grounds, and those are exactly the rows the evidence does not support. Under the assumptions that survive scrutiny, the change does not pay for itself within any horizon worth planning around.

That is not an argument against switching. It is an argument against switching for speed, which matters because speed is the reason most people give. If your reason is comfort, the calculation above is simply the wrong one to be running.

Mechanical versus membrane: what is actually different

The hardware question has the same structure and the same evidence problem. Start with what is genuinely, physically different, because that part is not controversial.

The genuinely important row is the actuation point. On a good mechanical switch the key registers partway through its travel, which means a practiced typist can release before bottoming out and avoid driving each keystroke into the base. On a typical rubber dome the registration happens near the bottom, so every keystroke ends in an impact. Over thousands of keystrokes that is a real difference in cumulative load — and it is a comfort argument, not a speed one.

The mechanical keyboard speed claims

Search for whether mechanical keyboards make you faster and you will find specific, confident figures: seven words per minute, twelve percent improvements, fifteen percent error reductions, each attributed to a study.

Try to locate those studies and the trail goes cold. The claims appear on retail pages, affiliate reviews and content-marketing blogs, cite institutions without linking anything, and reproduce each other's numbers. This is the same pattern as the productivity statistics examined in the shortcuts article — a plausible figure repeated until it acquires the texture of fact.

There is real research on keyboards, mostly in ergonomics rather than performance, measuring muscle activity and load rather than words per minute. That literature is more cautious and considerably less quotable, which is presumably why the marketing figures travel further.

• Treat any specific WPM gain attributed to hardware as unsupported until you can find the actual study.
• Be especially suspicious of round percentages — 10, 12, 15 percent are the signature of estimates presented as measurements.
• Note who benefits from the claim. Most of these figures appear on pages selling keyboards.
• The absence of good evidence for a speed gain is not evidence that keyboards do not matter — they matter for other things.

Switch types, and which suits typing

If you do buy a mechanical keyboard, the switch choice matters more to your experience than the brand does. There are three broad families and the differences are real, though preference dominates any general recommendation.

Most people who type all day and try all three settle on tactile, because the bump indicates registration without requiring you to bottom out or listen for a click. But this is genuinely a preference, and anyone telling you there is a correct answer is describing their own hands.

The ordering is the point. The two factors at the top are free and available to everyone, and both are addressed in the complete ten-finger method and a practice routine that actually moves the number. The two at the bottom are the ones people spend money and months on.

Form factor: laptops, low profile and split boards

Switch type gets most of the attention, but the shape of the keyboard affects a working day more than what is underneath the keycaps. Three form factor questions come up repeatedly and each has a clearer answer than the switch debate does.

Laptop keyboards use scissor mechanisms with very short travel, typically a fraction of what a desktop board offers. Most people type slightly slower on them, though less because of the switches than because of the cramped arrangement and, more importantly, the posture a laptop forces. The screen and keyboard are welded into one fixed relationship, which means a comfortable neck angle and a comfortable wrist angle cannot both be achieved. If you work on a laptop for hours, raising the screen and adding an external keyboard changes comfort substantially and speed only slightly.

Low-profile mechanical keyboards attempt to combine short travel with switch consistency. They suit people who have spent years on laptops and find full-travel boards tiring, and they are a reasonable middle ground. The shorter travel does not confer speed, but it does reduce the distance every finger covers thousands of times a day.

Split and ergonomic keyboards are the one category where the comfort case is strong and widely reported. Separating the halves lets the wrists sit straight rather than angled outward and lets the shoulders open instead of rolling forward. The cost is a genuine adaptation period — expect a few weeks of reduced speed, in a smaller version of the layout transition described earlier. That is a sensible trade if discomfort is the problem and a poor one if speed is.

• Laptop keyboards cost more in posture than in speed. Fix the screen height before blaming the keys.
• Low profile suits long-time laptop users moving to a desk setup, on familiarity grounds.
• Split boards have the strongest comfort case here and a real adaptation cost of a few weeks.
• Any form factor change temporarily lowers your speed. That is adaptation, not a verdict on the hardware.

Who should actually switch

None of this means alternative layouts and good keyboards are pointless. It means the usual justification is the wrong one. There are people for whom switching is a reasonable decision, and their reasons look different.

• You have persistent hand or finger discomfort and have already addressed posture, wrist angle, tension and breaks without resolving it.
• You type for many hours a day, so reduced cumulative load compounds into something meaningful over years.
• You find the process genuinely interesting and are not doing it as an investment. This is a perfectly good reason.
• You are learning to touch type from scratch anyway, in which case the relearning cost largely disappears.

That last case is worth emphasizing. Almost the entire cost of switching is unlearning, so someone who does not yet touch type pays very little of it. If you are about to learn properly and are drawn to Colemak, the arithmetic is genuinely different for you than for a fluent QWERTY typist.

• You are chasing a plateau — that is a practice problem, and equipment will not touch it.
• You regularly type on machines you do not control, where your layout will not be available.
• You are hoping for a specific speed number, since the evidence does not support one.
• You are avoiding the harder work of changing how you practice, which is the most common real motive.

That final point is the uncomfortable one. Equipment decisions are appealing precisely because they are decisions rather than habits — they can be made once, they feel productive, and they require none of the discomfort that actual improvement demands. Why that discomfort is unavoidable is set out in why automaticity stops progress.

What to do instead

If the goal is to type faster, the ordering implied by the evidence is unglamorous and has nothing to do with purchases.

Establish whether you touch type. If you still locate keys visually, that is your constraint and nothing else comes close — the ceiling of sight typing sits in the mid forties regardless of what you type on. If you already touch type but have been flat for months, the constraint is that your practice has become repetition, and the fix is a change in kind rather than amount.

If neither applies and you type competently, the remaining lever is not speed at all but volume: the keystrokes you never needed to produce. And if discomfort rather than speed is what actually prompted this question, then posture, breaks and tension are the first things to change — the habits involved are covered in the mistakes that quietly cost you accuracy.

Buy the keyboard you enjoy using, by all means. A tool you like using for eight hours a day justifies itself on comfort, and comfort is a real benefit that this article has no interest in talking anyone out of. Just buy it for that, and measure your speed on a one-minute test before and after if you want to see for yourself how little the number moves.