Category: Input Devices

This is not a controversial position to take, I think among the computer programming set. However, I think it deserves a little more of a centralized, rigorous defense, instead of the general appeal to "common sense" and preferences that usually happens. Especially in response to some critiques of this preference which I'm going to respond below.

I prefer keyboard-driven interfaces to mouse-driven interfaces for three predominant reasons.

If your primary job, hobby, or both involves typing, you need an ergonomic keyboard. Period.

Traditional keyboards were not designed with humane ergonomics in mind. They were designed around the technical constraints of 19th century typewriters, and then that layout was ported forward and forward until we've now ended up with a miniature version of it on our damn smartphones. There are many inherent design flaws in the traditional keyboard, but here are a few:

• keys are all arranged into one single central group and designed to be typed on in a vertical fashion, instead of the keyboard being split into two halves and rotated outward to meet the user's arms naturally, leading to typists' wrists being crammed together and oriented straight up and down, which causes ulnar deviation;
• keys are also row-staggered, forcing your fingers to travel along strange diagonal paths alien to how your fingers naturally extend when curling and uncurling them, instead of keys being arranged in columns so that your fingers only need to move in straight lines, or even columns that naturally splay out from where your palms are, in keeping with the most relaxed angle of finger extension;
• keys along the left and right edges are extended to make them slightly easier to reach, but otherwise the keyboard makes absolutely no accommodations for the natural variations in length and dexterity of human fingers, such as making keys at the top of the keyboard that may be hard to reach taller, or using an (entirely non-existent in traditional keyboards) columnar layout to stagger the columns according to relative finger length;
• instead of using the thumb, which is the strongest finger on the human hand, and the one that is used to holding things down for long periods of time, for things like control, alt, and shift, that burden is instead placed on the weakest finger on the human hand, and the one that is most likely to cause nerve issues (due to its connection with the carpal tunnel), the pinky;
• most modern keyboards require you to press keys all the way, slamming your fingers into the backboard over and over, instead of registering about halfway through the key's travel.

If you spend hours a day, almost every day, furiously using a device that was fundamentally not designed for ergonomic human interaction, with your most complex and fragile appendages, it is only a matter of when, not if, you will run into physical problems. A lot of us writers, programmers, and others that spend all their time in cyberspace or imaginary space all day tend to forget this, but we're physical human beings too (unfortunately). We have bodies and, even if we have it easy compared to, say, construction workers, those bodies are subject to physical wear and tear of our day to day jobs (or hobbies), just like everyone else. In this case, that wear and tear is predominantly centered around our human-computer interface system, broadly conceived: the mouse, keyboard, chair, desk, monitor. Desk and monitor ergonomics can do a lot for your back and neck, and mouse ergonomics can help a little with one wrist, but paying attention to all of that and completely ignoring one of your main human-computer interface devices, the keyboard, is like leaving a giant sinkhole in the middle of your otherwise sparkling-clean living room.

At first the problems you eventually do run into with your wrists and fingers as a result of keyboarding all day will seem minor – sure you've got a little pain, aching, numbness, or tingling but the pain quickly goes away and you can get back to work, right? But this doesn't last. Maybe it will for a year, five years, even ten, but don't you want to be able to do the work or hobbies you love much longer than that, maybe even your whole life? You can't get away with ignoring your body forever, and that damage, that strain, that wear and tear, those warning signs – they add up over time.

If you don't treat the root cause of the symptoms, the carpal tunnel and tendonitis will begin to become chronic, an ever tightening vicious cycle of irritation and pain that leads to more irritation and eventually outright damage, and eventually you'll be faced with full-on repetitive strain injury, which does not heal. Once you have RSI, short of getting a surgery that costs tens of thousands of dollars and isn't guaranteed to do anything for you at all, your body can't heal it. The damage is too minute and fine-grained for it to deal with. So you'll just have to accept that crippling pain and numbness and weakness for the rest of your life at that point.

This means that you really can't approach this problem with a cavalier "I'll cross that bridge when I come to it" attitude. Sure you might be one of the, say, 30% of people who are lucky, whose bodies are just somehow so resilient and flexible, or whose psychology just naturally makes it easy (or even effortless) to regularly take breaks, but once you've gotten to the bridge, it's far too late to actually deal with it then, so why take risks? Why not use a little prophylaxis?

Preserve your body now, so that you can enjoy having it in your old age.

How do you go about preserving your body, though? Breaks that are only long enough to treat the symptoms – stopping for a second to take a breather, stretching your wrists, cracking your knuckles, and so on – don't actually deal with the underlying wear and tear that your body is warning you about. The only way to actually deal with those issues properly is to consistently take pretty long breaks, something on the order of thirty minutes to an hour, regularly, and that's difficult to keep up with any consistency: obviously, it's easy to forget to do it, even in response to pain symptoms, or constantly give excuses to never do it, and timer based approaches to solving that won't work well for many jobs that involve a lot of keyboarding, such as programming or writing, which tend to also involve a flow state that's extremely important to maintain. Moreover it's well known at this point that willpower-based approaches to habits and routines and things you need to do simply don't work very well.

This is why, in my opinion, everyone that primarily works with a keyboard needs to spend some time picking out an ergonomic (preferably mechanical) keyboard that works for them. The properties you want to look for are:

1. Split, rotated layout, so that your wrists can be straight in line with your arms;
2. Column-staggered keys, preferably with splay;
3. Tenting capabilities if possible;
4. Mechanical keys, to avoid having to slam your fingers into the back of the keyboard.
5. Key sizes that have been adjusted to account for finger length and key position.

It is my strong opinion that arpeggiated key bindings – key bindings where you press multiple single, un-modified keys in a sequence, as you would in things like Emacs god-mode or Vim – are vastly superior, ergonomically, to key chords – like you would use in traditional GUI applications, or in vanilla Emacs. This is for a few reasons.

1. Having to hold a key down for an extended period of time is inherently more stressful and straining than just hitting it once, just like having to stand holding a heavy box for a long period of time is much more difficult than only having to pick it up briefly.
2. Having to hold a key down while hitting another key requires more effort even than holding down either by itself, since muscular strain is not additive but something like multiplicative.
3. Having to hold a key down and then hit another key (or hit two keys at the same time, which also brings in a timing and coordination difficulty) invariably leads to you having to stretch and contort your fingers and wrists – and thus the tendons inside them – to varying degrees in order to reach both keys at once. This gets more difficult nonlinearly with the amount of modifier keys required, while key arpeggios require constant effort no matter how complex the grammar of modifications becomes.

One might be tempted to bring out the stenographer's keyboard as a counterexample to my statements, but that's not so. The stenographer's keyboard reaches its level of efficiency and ergonomics not primarily from using the human body in a more ergonomic way, but by simply allowing you to get more across with a single key chord, therefore becoming significantly more efficient. Moreover, a proper stenographic keyboard is also laid out in a completely different way (as a piano-like double row of elongated keys) that makes doing chords not really require holding down keys at all!

When I realized that I needed a split ergonomic mechanical keyboard, I dove deep into the space, searching for something that could work for me. I quickly found the ZSA family of keyboards, but they were completely outside my budget. I also ran into the Kinesis Advantage and Advantage360, but not only were they also completely outside my budget, I also didn't like the straight up and down wrist angle of the Advantage, and the thick, tall design of the 360, nor the thumb cluster design that looked awkward to reach, and I didn't want to lock myself into proprietary firmware for programming the keyboard either, since I had wonderfully baroque ideas for what I wanted to do.

As my search deepened, I soon found myself immersed in the world of DIY ergonomic split keyboards. Generally, the ideas and design principles of this space were good, centered around rectifying very real and pertinent design flaws in traditional keyboards through such things as splitting, tenting, column-staggered key layouts, splayed key columns, low-profile housings, and mechanical key switches.

The space was not free from its share of blind trend-following and folk wisdom, however. One of the most baffling examples of this was the single-minded focus of the whole community at large on something which I can only call keyboard minimalism: the endless desire to seek the absolute minimum number of keys they could get away with on their keyboard designs, and the general dismissal of designs with more keys through the assumption that everyone, eventually, would ultimately move toward keyboards with a smaller number of keys over time, as those were just obviously superior.

You can see this thinking in the fact that nearly every relatively mainstream, well-known, often-recommended DIY split ergonomic keyboard design has very few keys. While a standard keyboard has anywhere from 90 to 104 keys, the average DIY split ergomech keyboard, like the popular accessible starter the Keebio Iris, or the even more popular and highly regarded Sofle design – which in turn was designed based on popular designs with similar key counts like the Lily58 and Helix – seems to have half that, around 58 keys, at a maximum. Many go even lower, too – the very popular and widely regarded as "endgame" (i.e. perfect, if you can adapt to it) Corne keyboard only has 42.

I even spent some time on Compare Split Keyboard looking for a design with enough keys for me – someone who prefers roughly the number of keys and keyboard size of a tenkeyless – and was shocked to find that the orthodox dogma in this community that fewer keys = better was so complete that, out of all of the many hundreds of varied split ergonomic mechanical keyboard designs that list documented, not a one had more than 80 keys. It seemed no one even thought to want this, let alone sit down and design it. Questions on related forums asking about keyboards with more than a mere handful of keys are met with usually one, maybe two, actual options, a suggestion or three to design your own keyboard, and several people simply telling the questioner that they're wrong for wanting more than about 60 keys or fewer.

Apparently it's become received wisdom in these niche circles that using keyboard layers – where you tap or hold a special modifier button that changes the meanings of some or all of the keys on your keyboard, to emulate having more keys – to make up for missing navigation clusters, symbol keys, function keys, modifier keys, and often even number keys, is superior to simply having enough keys on your keyboard for all the basic things you need to do. The argument for this tends to be simple: for the sake of efficiency and minimizing effort/strain, why reach an extra row or column or two for a key, when you could instead hold down a key that's readily available under one of your fingers (perhaps the thumb or index finger) and have the key you want appear under you?

There are several problems with this logic. First of all, there's the problem of how to activate your layers: do you want to toggle the layer on and off with a key tap, in which case you have to remember what layer you're on, since most keyboards don't have a screen or way to communicate with the computer to display what layer you're on, unlike modal editors? Or do you want to treat layers as an additional modifier key, in which case you run into the problem of key arpeggios versus key chords, that being that key chords are inherently difficult and unergonomic? (Doubly so with keys that need to be modified in some other way, such as with the control or shift key, while also being activated by a layer). Moreover, with every layer you add, you're introducing more complexity to keep in mind. Control, Meta, Shift, and Super all introduce their own layers, and now you want to introduce one or two (sometimes up to six) additional layers into the mix that you have to have memorized?

More than that, the number of layers you'll need to use is going to scale with the smallness of your keyboard, but as the number of layers you want to access increases, so too does the number of keys needed to actually access those layers, meaning that as the number of available keys on your keyboard gets more and more cramped, the number of keys that have to be occupied with layer-switching increases. You can try to mitigate this via having keys do different things when you tap on them versus when you hold them, so that they can do double-duty, but now that's just more that you have to remember, and, depending on what keys you choose, inconsistent semantics between keys that repeat when held and keys that activate a layer when held, which, combined with layers changing the meaning of possibly all the other keys on your keyboard, is a recipe for turning your keyboard into a fucking minesweeper game.

More problematic than any of this, though, is the fact that by using layers to regain missing number, symbol, and navigation keys, in addition to the letter keys on the base layer that you have just enough room for, you're essentially converting entering different types of input into entirely separate modes, like on a mobile keyboard, instead of things that you can fluidly move back and forth between. In order to enter something like $variable4, you've now got to switch to your symbol layer for a single symbol, then back to your base layer for the main word, then to your number layer, and then back again so you can type space after it, and good luck typing something like mathematics.

Can this be overcome? Sure, and clearly many people do. But the point I'm making here is that juggling layers isn't cost-free. There's a huge learning curve and a lot of inherent cognitive overhead that is unlikely to be something you can easily encode into muscle memory and not have to worry about anymore, and there are a lot of inherent points of friction even if you do manage to actually turn all this detail into muscle memory.

This has the knock-on effect that you're going to want to try to cram all of your most used keys and symbols onto your base layer, which is a recipe for endless tinkering and adjustment trying to find the optimal layout for your keyboard, tetrising everything into the most awkward and inconsistent positions in an attempt to shove it all in there.

And all for what? Is having to move your finger one less row or column really a big gain compared to having to press (or especially hold) one extra key and keep all the rest of this stuff in your head while you do? In extreme cases, this might be the case (imagine a keyboard with no shift key, just duplicated keys for everything), but in my opinion the advantages of layering only hold when the alternative is often-used keys being shoved out of easy reach of your fingers without having to move your hands. In a situation where all the keys in question are within reach of your fingers anyway, saving your fingers some distance of movement just to add an extra key to hit or extra chord to hold down doesn't seem like a clear win to me.

Just take it from my girlfriend, who initially dismissed my strongly voiced concerns with her purchase of a $160 Keyboardio Atreus (which has only 44 keys), but who – as I'm writing this – I just found out has completely given up on learning to use the keyboard for precisely the reasons above:

The longer I used it, and the more different things I used it for, the more unavoidable the fact was that the limited number of keys was hampering my productivity, primarily due to the necessity to switch back and forth between different layers. My primary beef was with the lack of accessible number keys, which made navigation in my window manager difficult, made inputs that were mixed difficult, and overall the necessity in a lot of the software I used to switch between typing and numbers and game style WASD input makes the limited key set an absolute pain to use. It was bad enough that I came to feel anxious when I was going to have to use the keyboard, and found respite in my T420 laptop keyboard instead. Furthermore, the small number of keys makes placing symbols in quickly memorable positions impossible. So you wind up having to learn a lot more.

Why not avoid bringing all this trouble down on your head by just using a keyboard that has enough keys to put all the letters, symbols, numbers, and navigation keys you need right at your fingertips? I suspect the real answer isn't actually the ergonomic benefits, but the allure of small size, minimalism, and trying to be different. I hope, in a few years, this trend dies down and we see more approaches like that of the Glove80.

The split ergonomic mechanical keyboard that I eventually settled on, after trying out the Keebio Iris and hating it,