5. Fitting the Keyboard to Mara's Long Typing

Watching Mara type before changing the keyboard

Mara's pointer trial is written down. Her mouse sits beside the keys, acceleration is off, and one button does undo. Now we turn to the hands that type. Mara is still our illustrative example. Her numbers show the method. They are not your numbers.

The question for this chapter is simple. What does Mara's typing need from a keyboard at both of her desk heights? And how much of that can she get before she spends anything?

Right now she types on her laptop's own keyboard. That keyboard is built into the laptop, so it goes wherever the laptop goes. In chapter one, that was the trouble. When she raised the laptop to bring the screen up, the keys came up too, and her hands went high above her elbows. Since then, a lot has changed. She has a sit-stand desk with two saved heights. The seated preset is twenty-eight and a half inches, and the standing preset is forty-one. Those match her adjusted seated elbow and her standing elbow. Her monitor sits on its own stand, and one stand setting works at both heights. So the laptop screen no longer has to set her gaze. The monitor does that now. The laptop is free to go wherever the keys work best.

Before moving anything, we watch her work. We are looking at two kinds of typing, because her day has two.

The first is long typing. She writes reports and long emails, sometimes for an hour or more without a break. In those stretches she uses letters, the space bar, backspace and the arrow keys. Errors show up as backspacing. Fatigue shows up as a change in posture.

The second is image work. Here she barely types words at all. Her right hand stays on the mouse. Her left hand presses shortcuts: undo, redo, zoom in, zoom out, the keys that change brush size, and holding a key down to pan around the image. Many of these shortcuts need a modifier, such as Control on Windows or Command on a Mac, held at the same time as a letter. So her left hand spends a lot of time stretched between a modifier in the corner and a letter near the middle.

We watch four things in both kinds of work.

Reach is the first. How far do her fingers stretch to the keys she uses most? Does her hand have to leave its home position to find a key?

Wrist angle is the second. Looking from the side, does the wrist bend up or down? Looking from above, does it bend sideways toward the little finger? That sideways bend has a name. It is called ulnar deviation. It happens when the hands are closer together than the shoulders, so the forearms angle in and the wrists bend back out to line up with the keys.

Shoulder width is the third. Are her hands roughly as far apart as her shoulders, or are they pulled in toward the middle of her chest?

Errors are the fourth. Which keys does she miss? When does she miss them? Are they missed early in the hour, or only late?

Then we check all four at both presets. Sitting at twenty-eight and a half inches, and standing at forty-one.

Here is what Mara's watching session finds, as an illustrative account. The laptop sits flat on the desk, with the mouse beside it. Sitting, the laptop's keys are a little above her elbows. A laptop's keyboard deck is raised off the desk by the laptop's own thickness, so the keys sit perhaps three-quarters of an inch above the desk. That is a small gap, and it mostly works. Standing gives the same result, because the desk preset matches her standing elbow the same way. So height is close to right at both presets. That part of the problem was solved in chapter two, when the desk heights were matched to her elbows.

What is not solved is width. A laptop keyboard is narrow. Mara's home keys are only a few inches apart, and her shoulders are much wider than that. So her forearms angle inward, and after forty minutes of report writing her wrists are bent outward. Her shoulders also creep forward a little. In image work, the reach problem shows instead. Her left little finger lives on the Control key in the far bottom corner. Her ring or middle finger stretches up toward letters like Z for undo. She also has to find the bracket keys for brush size, which sit on the right side of the keyboard, while her right hand is on the mouse. So she either takes her hand off the mouse or makes a long reach across.

Now we know what to fix. Width during long typing. Shortcut reach during image work. Height is mostly fine.

Two free changes

The series has one habit that has paid off every time. Try the change that costs nothing first, on real work. Then let the result decide whether there is even a product question to ask.

Mara has two free changes available.

The first is the one she is already using, done carefully. The laptop lies flat on the desk surface, which already sits at elbow height. Nothing props it up. The monitor handles her view, so the laptop screen can sit low and simply be ignored, or used as a second screen for reference. We test it at both presets, using the same report and the same image task each time. The result is that the height is close and comfortable. The pointer is right beside the keys, as chapter four arranged. But this change cannot make a laptop keyboard wider. The forearms still angle in. It also does nothing for shortcut reach. So the first free change confirms the height and leaves width and reach open.

The second free change is a borrowed external keyboard. Mara has a friend's ordinary full-size board, the kind with a number pad on the right. An external keyboard does one important thing. It separates the keys from the laptop completely. Now the laptop can go anywhere, even off to the side, and the keys can go exactly where her hands want them.

She sets the board flat on the desk in front of the monitor and repeats the tests at both heights. Two things improve. A full-size board has slightly more room between the keys than her laptop, and the keys travel deeper when pressed, so long typing feels a little more relaxed. It also has a full set of keys, including separate navigation keys like Home, End, Page Up and Page Down. That helps her report editing.

Two things get worse. The first is the number pad. It sits to the right of the letters, so it pushes the mouse further right, by about the width of the number pad. That is several inches. Her right arm now reaches out sideways to find the mouse. The careful placement from chapter four just got undone. The second thing is width. A full-size board still puts both hands next to each other in the middle. Her home keys are a bit further apart than on the laptop, but not by much. The inward angle of her forearms barely changes.

There is a small fix she can try within the borrowed-board test. She can slide the board left, so the letter keys, not the whole board, sit centered on her body. That brings the mouse back a little closer. It helps, but the number pad is still in the way.

So we record what each change fixed and what it left open. Laying the laptop flat confirmed that the desk presets put her keys at the right height, sitting and standing. The external board proved that separating the keys from the laptop works. It improved long typing a little and added useful navigation keys, but it pushed the mouse out. Neither change fixed the width of her hands. Neither change fixed the shortcut reach during image work.

That leaves two open problems. One is about the shape of the keyboard. The other is about which keys do which jobs. Those are different problems with different answers. The shape problem might need new hardware. The key-job problem might be solvable with software she already has. We look at both, because the cheapest answer to one might not touch the other.

The record so far

Mara's record now has a keyboard line. Laptop flat on the desk at both presets: height acceptable. Borrowed full-size board: separates input, adds navigation keys, pushes mouse right by the number pad's width. Open: hands narrower than shoulders in long typing; left-hand stretch to modifiers and right-side shortcuts in image work. Installed cost of the whole build is still about three hundred and ten dollars, all of it the desk. Nothing has been bought for typing.

That cost line matters. The desk was bought because a measured requirement said so. A keyboard should be bought the same way, or not at all. So the next step is not to shop. It is to understand what different keyboard shapes actually change, so we can tell whether any of them answers Mara's open problems, and at what cost in learning.

Choosing and configuring a keyboard for Mara's typing

Ergonomic keyboards come in shapes that look strange the first time you see them. Each odd feature changes one specific thing about reach or angle. If you know what each feature changes, you can match it to a problem you actually measured. If you don't, you end up buying a look.

Let's take them one at a time, starting from what Mara has now.

A split keyboard is cut into two halves, one for each hand. Some are joined by a hinge. Many are two fully separate pieces linked by a cable or by wireless. The point of the split is simple. You can move the halves apart until each half sits in front of its own shoulder. Then the forearms point straight ahead instead of angling in, and the wrists have no reason to bend outward. This is exactly Mara's width problem. A split does not change how far any single finger reaches to any single key. The letters are where they always were on each half. That is why a split is usually the easiest change to learn. The main surprise comes for people who type certain middle keys, like B or Y, with the "wrong" hand. On a split, those keys are on one half only, so that habit has to change.

Splay is the angle of each half. With the halves apart, you can also rotate them so the front edges point slightly outward, like the two sides of a shallow V. That lines each half up with the forearm coming toward it. Splay fine-tunes what the split started. It costs nothing to adjust on a board with separate halves.

Tenting means raising the inner edge of each half, so the halves tilt like the two sides of a tent roof. Your forearms naturally rest a little rotated. Laying the palms perfectly flat on a desk twists the forearm further than its resting position. That twist is called pronation. Tenting lets the hands sit partway between flat and the handshake position. The keys stay the same distance from each other. Only the angle of the hand changes. People report that steep tenting, past about fifteen or twenty degrees, feels unfamiliar at first, even though finger-to-key distances have not moved. Tenting also raises the board a little, which matters for typing height. Mara would need to recheck her elbow-to-key gap at both presets after tenting.

Row stagger versus column stagger is about how the keys line up. On an ordinary keyboard, each row is shifted sideways from the row above it. That is row stagger. It is left over from typewriters, where the mechanical arms under each key needed room. Because of that shift, many fingers reach diagonally to the row above or below. Column stagger lines the keys up in straight vertical columns, one column per finger. Then each column is shifted up or down to match finger length. The middle finger's column sits higher because that finger is longer. The little finger's column sits lower. Each finger moves straight up and down instead of diagonally. This mainly helps people who type long stretches by touch. It is also harder to learn than a split. Reports put the slowdown at about one to three weeks, and most of the trouble is on the bottom row, where Z, X, C, V and B used to sit at a slant.

A thumb cluster is a small group of keys under each thumb. On an ordinary keyboard, both thumbs share one long space bar, and that is nearly all they do. The thumbs are strong and close to the keys, but they sit mostly idle while the little fingers stretch to the corners for Control, Shift and the rest. A thumb cluster moves some of that work to the thumbs. This touches Mara's shortcut problem directly. If Control sat under her left thumb, she would not need to stretch her little finger to the corner every time she pressed undo. A thumb cluster needs a layout that puts useful keys there, which we'll come to.

A concave keywell sets the keys into a curved bowl rather than on a flat plate. Each finger closes in an arc, and a keywell places the keys along that arc, so fingers travel less to reach the far rows. Of all the shapes, this has the steepest reported learning curve, about two to four weeks. It also works best for strict touch typists, because there is no flat area to wander across. It also asks you to get your arm and desk height right, because the bowls only fit if your hands arrive at the right angle.

None of these is the best shape for everyone. Each one answers a particular reach or angle. Two contrasting cases make that clear.

Imagine a broad-shouldered person, much wider across the shoulders than Mara. On a laptop keyboard, their forearms angle in even more sharply, and the outward wrist bend is larger. For them a split is likely to be the change that matters most, with the halves set far apart. A keywell might add little if their fingers already reach the rows comfortably.

Now imagine someone who types very little, but who works all day through shortcuts in a video editor or a spreadsheet. Width barely matters to them, because they are never in a long typing stretch. What matters is which keys sit under their fingers and thumbs. For that person, remapping keys on the keyboard they already own might solve most of the problem, and a split might solve almost none of it.

Mara sits between those two. She has the long typing, where a split addresses width. She also has the shortcut-heavy image work, where the question is which key does what. That second question can be worked on today, on the borrowed board or her laptop, for free.

Building a layered layout

Two terms first. Key count is how many keys a board has. Many split boards have far fewer keys than a full-size board, sometimes thirty-four to forty-two keys instead of over a hundred. That only works because of layers. A layer is a second, or third, set of meanings for the same keys. Hold one key down, and the keys under your fingers change jobs. Let go, and they go back. It works like the Shift key, which already gives every letter a capital version. A layer key just does the same thing for whole groups of keys.

Let's map Mara's work onto a simple layout. We start with what she uses, not with what a board has.

For letters, she keeps her ordinary layout. Changing letter positions is a huge learning cost, and nothing in her observation points to it. Letters stay on the base layer, the one active when nothing is held.

For modifiers, the problem is the left corner. The simplest fix on any keyboard is to give a closer key the job of Control. On her current board, Caps Lock sits right next to her little finger's resting place. Remapping Caps Lock to Control is a common first change. Her little finger moves a short way sideways instead of curling down to the corner. On a board with a thumb cluster, Control could go under the thumb instead.

For navigation, she holds a layer key, and her right hand's home keys become arrows, with Home and End beside them. Her hands never leave home position to edit a report.

For image work, this is where layers earn their keep. While her right hand is on the mouse, her left hand holds one layer key. On that layer, the keys under her left fingers become undo, redo, zoom in, zoom out, and brush size up and down. The brush-size keys no longer live on the far right side of the board. They are under her left hand, a short press away.

That is a working layout, and it fits on paper. Base layer: ordinary letters, Caps Lock as Control. One navigation layer for the right hand. One shortcut layer for the left hand during image work.

Now, where does that layout live? There are two places, and they behave very differently.

The first is the operating system. The computer catches each keypress and changes it before any program sees it. That works on any keyboard, including her laptop's own and the borrowed board, with nothing to buy. But the layout lives only on that computer. Plug the board into another machine and it acts like a plain keyboard again.

On Windows, the free tool is Microsoft PowerToys, and inside it the part called Keyboard Manager. You open PowerToys Settings, go to Keyboard Manager, and switch it on. Under the keys section you choose to remap a key, press the plus button, press Caps Lock as the physical key, and choose Left Control as what it maps to. Then you press OK to save. There is also a section for remapping shortcuts, which can apply everywhere or only in one program. That is useful for Mara's image editor. One limit matters a lot here. Keyboard Manager does not do layers. It can swap one key for another and change one shortcut into another, but it cannot make a held key turn a group of keys into something else. So on Windows, Mara could get the Control fix and some shortcut changes, but not the shortcut layer.

On a Mac, the free tool is Karabiner-Elements. It first asks for two permissions in System Settings, under Privacy and Security, called Input Monitoring and Accessibility. Then, in its Simple Modifications tab, you pick the keyboard, add an item, and set Caps Lock as the from key and Control as the to key. Layers live in a different tab called Complex Modifications. There you can add a ready-made rule, including rules imported from Karabiner's online collection, that turns a held key into a layer, and then enable it. So on a Mac, Mara's full layout, with both layers, is possible with software alone. Karabiner also comes with a small program called Karabiner-EventViewer, which shows what each keypress really sends. That is how you check that a remap took.

The second place a layout can live is the keyboard itself. Many ergonomic boards have a tiny computer inside, and the program on that computer is called firmware. Change the firmware's layout, and the board carries it to every computer it touches. No helper software has to run in the background. Two firmware families come up again and again. QMK is common on wired boards. ZMK is common on wireless ones.

You usually don't edit firmware by hand. There are friendly editors. For QMK boards set up to support it, VIA runs in a Chrome-type browser. You connect the board by cable, authorize it, and see a picture of the keyboard. You choose a layer, where zero is the base, click a key in the picture, and pick its new job from a tray below. That tray includes layer keys. One kind works only while held. One kind toggles a layer on and off. And one kind does two jobs, acting as a normal key when tapped and as a layer key when held. Changes are written to the board right away. Vial works much the same way and reads the layout straight from the board when you connect. ZMK Studio is the matching editor that some wireless ZMK boards ship ready to use. There is also the QMK Configurator, a website where you build a layout, and QMK Toolbox, a program that loads it onto the board. That route is more involved. It means putting the board into a special loading mode, called bootloader mode, before writing the new firmware. It is the path for boards without a live editor.

Why would Mara pick one place over the other? Operating-system remapping is free and works today on what she has. Firmware remapping only matters once she owns a board that supports it. Then it follows the board between computers. So she starts with the operating system.

Saving, backing up and reversing

A layout you can't restore is a layout you are afraid to change. So the save and the backup happen before the first trial day, not after.

In PowerToys, the Keyboard Manager keeps its rules in a settings file in your user folder, under the PowerToys folder. You can copy that file somewhere safe, or use PowerToys' own settings backup. In Karabiner-Elements, the whole setup is one settings file in your home folder's configuration area, and Karabiner also keeps its own automatic backups beside it. Copy the file. Name the copy with the date.

Reversing is just as easy. In PowerToys you delete a single rule with its trash icon, or switch Keyboard Manager off entirely, and every key goes back to normal. In Karabiner you remove a rule under Complex Modifications, or switch to an empty default profile.

Firmware editors have the same pair. In VIA, the Save and Load tab exports the layout as a file, and loading that file puts it back. There is also a special key you can assign that clears the board's changes and returns it to its original layout. In Vial, the File menu saves and loads layouts, and a reset option in its Security menu wipes changes back to the firmware's defaults. In the QMK Configurator, you export the keymap to a file, and a full reset means loading the maker's original firmware again. Each editor also has a key tester, so after any change you press every key on every layer and watch it register.

Adopting the layout gradually

Here is the adaptation plan. It has five steps, done in order.

1. Keep the old setup available, so the laptop's keyboard or the borrowed board can always go back to plain behavior with one switch.
2. Change one thing at a time, starting with Caps Lock as Control, and use it for a few working days before adding anything else.
3. Write the date of each change in her record, next to what changed.
4. During a set report task and a set image task, note backspaces and missed shortcuts, and roughly how long the task took, so a slowdown is measured rather than guessed.
5. Add the next layer only when the last change has stopped causing errors, and undo any change that keeps causing them.

That last step matters. A remap that still trips her up after a week or two is information. It is not a test she failed. She turns it off, writes down why, and tries something else.

For Mara, as an illustrative run, Caps Lock as Control settles in within a few days. The shortcut layer goes on her Mac with Karabiner the following week. Her undo errors drop, and her right hand stops leaving the mouse to change brush size. Her shortcut reach problem is now mostly solved with software she already had, for no money.

The width problem is not solved. Software cannot move her hands apart. That is the one open question that points at hardware.

Reading a complete purchase

If Mara ever buys a split board, the price on a listing is only part of what she is buying. A complete purchase has several parts. Each one changes what she has to do and maintain.

Assembled or kit comes first. A pre-assembled board arrives built. A kit arrives as parts, and many kits require soldering, which means joining metal parts with a hot iron. Some boards come both ways. The Cantor Remix and the Cheapino version two, for example, are kits with published parts lists, and both require soldering. The Kinesis Advantage 360 Pro and the Beekeeb Toucan 36, by contrast, can be bought built.

Flashed firmware comes second. Does the board arrive with firmware already loaded, and does that firmware support a live editor? The Toucan 36 arrives with ZMK and with ZMK Studio ready. Kits like the Cheapino use QMK with Vial. You can also expect to load firmware yourself during the build.

Switches and hot-swap come third. Switches are the mechanisms under each key. Hot-swap sockets let you pull a switch out and push a new one in without soldering. That makes switches a replaceable part. You can change the feel, or replace a broken one in minutes. The Dygma Defy uses hot-swap sockets for standard full-height switches. The Kinesis Advantage 360 Pro does not, because its switches are soldered. The same goes for the MoErgo Glove80, whose tight keywell uses soldered low-profile switches. Low-profile switches are shorter and make a thinner board, which can matter for typing height. Full-height switches have the widest choice of replacements. Keycaps, the plastic caps you touch, are often sold separately on kits and must match the switch type.

Tenting hardware comes fourth. The Kinesis Advantage 360 Pro has tenting feet built in. The Toucan provides mounting holes, but tenting legs have to be bought separately. Some kits need 3D-printed wedges. If tenting is part of why you want the board, an untented board plus missing hardware is not the board you wanted.

Connection and battery come last. A wired split joins its halves with a cable. The Cantor uses a TRRS cable, which looks like a headphone cable, and the Cheapino uses an Ethernet-style cable. Then a USB-C cable runs to the computer. A wireless split has a battery in each half, so it has to be charged, and batteries wear out over the years. The Kinesis Advantage 360 Pro and the Glove80 both have rechargeable batteries in each half.

Here is the catch about variants. Some models come in more than one version under one name. The Dygma Defy's base model is wired, and wireless comes only with a higher tier. The Toucan's hot-swap support depends on the exact model. So before buying, match the exact variant against the maker's page, not against a listing that blends them together.

Cleaning belongs in the purchase too. Keycaps can be pulled off and washed. Hot-swap switches can be replaced. Soldered switches mean a broken key is a repair job, not a quick swap. And whatever the board, a backup of its layout file is part of owning it.

What OCDevel's keyboard research does and doesn't say

OCDevel keeps a comparison table of split and ergonomic keyboards at ocdevel dot com slash ergo slash keyboards. As of early October twenty twenty-six, it lists thirty-eight boards, and nine have a full research write-up behind them. OCDevel itself rates the keyboard table as thin. In its words, it is a catalogue, not a verdict. It names no winner and does not claim to have tested the list. So we won't rank anything from it either. A deeper research pass on keyboards has not happened yet.

What the table is useful for is its buying axes. In OCDevel's scoring, thumb ergonomics is the most critical factor. It treats two or three thumb keys per side as the comfortable range. It calls thirty-four to thirty-six total keys the sweet spot for minimal split boards, because layers carry the rest. And it treats a live layout editor like VIA, Vial or ZMK Studio as critical for ease of use. Those are OCDevel's positions. They are reasonable criteria to read a board against. They are not a test result showing that one board suits your hands.

For Mara, the table helps her narrow the question. If she ever buys, she'd want a split with a thumb cluster of two or three keys per side, an editor she can use live, and tenting hardware included or available.

Mara's decision

Here is what goes into her record. Shortcut reach is solved by remapping in Karabiner, backed up and dated. Width is still open. The candidate answer is a split board, with the halves set to shoulder width. Its expected learning cost is low if the keys keep ordinary row stagger. It runs higher, roughly one to three weeks of slower typing, if she picks column stagger, and more again for a keywell. She also has to recheck typing height at both presets if the board is tented. Nothing is bought. Installed cost stays at about three hundred and ten dollars.

What would justify the purchase? Her rule is that the outward wrist bend during long typing is still there after the remapped layout has settled for two weeks. Then she buys, to that requirement, with the exact variant checked.

One more line belongs in the record. A keyboard changes reach and angle. It is not a treatment. If aching, numbness or tingling continues in her hands or wrists, the next step is a qualified clinician, not a more expensive board.

One more note, before your own step. OCDevel's tables earn a commission on their links, and a commission never buys a pick.

Here is an optional exercise if you want it. Today, sit or stand at your usual typing height. Look down at your hands from above for one minute of normal typing. Notice whether your forearms angle inward toward the middle of the keyboard. Then pick the one key your little finger stretches to most often, and write down which key it is and what job you would move closer.