19 min read

Changing Tunes? Linear keys for typing?

Fast typists have preferred the tactile, sometimes clicky, keyboard for a long time, but I've been thinking that maybe there's a more optimal formula out there.

I like to type fast. I'm a fast enough typist that the choice of equipment makes a real difference in my overall potential, and for the most part I can actually tell some of the differences between the various types of keyboards on a measurable and qualitative level. This means that I also have had some pretty firm opinions about what I like and don't like in a keyboard. But things might be changing.

The vast majority of people find a tactile key switch (with or without a clicky sound to it) more pleasurable to type on and usually faster to type with. Additionally, a great deal of people have found that full travel keyswitches give them a better experience than the low profile ones, despite what some of the laptop manufacturers might try to claim. But there is a lot of innovation going on with different types of keys and the various choices of different combinations of behaviors. With all of the different approaches, I felt that it was time to examine things with a fresh view, and see if some of the old standby rules of thumb still applied.

The reasons that the tactile and clicky full travel keyboards have traditionally been favored over other keyboard options for typists has to do with what it takes to be a good typist. So, rather than going with the rules of thumb, I decided to take a step back and think a little bit about what makes for a good typing experience for me.

I was brought into this self-examination over a period of time doing some FPS gaming on Overwatch as well as doing a lot of typing for my thesis. I've long held my typing record with a Truly Ergonomic keyboard with Brown MX style switches, and I usually find myself typing with a Topre switch keyboard on most days. However, in playing with Overwatch, I began to notice something. My ability to control when keys were being struck, particularly my ability to control the relative timing to two simultaneous key presses next to one another was remarkably poor. I came to this conclusion after getting some strange results using Zenyatta's abilities and getting some weird failures to trigger actions when I thought that they should trigger (in other words, not actuating the keys at the time when I thought that they should be activated) as well as trying to use Mercy's kit, particularly super jumping and the guardian angel ability.

If all of that sounds like something that shouldn't matter to a fast typist, then bear with me. The key to fast typing is about developing a consistent rhythm that allows you to maximally leverage your fingers, hands, and arms to actuate keys at the most rapid pace you can for the longest time at the most consistent pace. It's not just about distance, or travel, or the like. It's about the holistic combination of factors that leads up to you being able to move your fingers consistently at a rapid enough rate and with enough force at the right places to actuate the keys that you want in the right order.

A huge part of this challenge is your ability to predict when and how a key is going to actuate. But that is only part of it. It's also about your recovery from a keystroke. Anyone can slam down a bunch of keys a bunch of times, but if you use too much effort in the wrong directions, your ability to transition to new keys can be hampered, thus affecting your overall rhythm. This is one of the reasons that I find myself a Qwerty typist rather than Dvorak, Workman, or Colemak, all of which I have used before at very high rates (120+ WPM). Qwerty is one of the best layouts that I have found for fast transitions (though not the most efficient in terms of raw travel distance).

So how does the keyswitch play into this, and why does gaming matter? Well, many gamer's prefer tactile keyboards because they like the bump that gives them feedback and a "wall" that they can ride (sort of like prepping the trigger when firing a gun) to get ready to trigger an ability. This helps give them some physical feedback, but the gaming arena is one of those places where the linear switches (switches that have a smooth travel without any tactile bump or audible clicks) have made a big splash, because the linear nature is smooth and allows for players to trigger and spam a key without worrying as much about a bump interfering with they're recovery from the key stroke and hitting the next key stroke.

But linear switches have traditionally been viewed as terrible for good typing practice, because the long travel keys (for example, Cherry MX Reds) without feedback meant that for many people it was hard to accurately predict when they would get the keystroke to actuate. This inconsistency and unpredictability meant that the typist would normally have to overcompensate in the moment of speed typing to ensure that a key was actuated, leading to excessive motion and wasted effort to maintain the accuracy that was desired.

However, in doing some thinking and examining my keyboard situation, I started to deconstruct a bit of these relationships and put them together in a different way.

When it comes to my typing, what I need to achieve maximum performance is the ability to very precisely "trigger" a key or a set of keys in a way that allows me to very confidently know what keys will be hit when, and to do so precisely, with the minimum amount of continued effort after the actuation. In other words, I need to be able to actuate a key as atomically and instantly as possible and then immediately have my finger working on transitioning to the next key that it should be pressing.

Let's break that down. There are two phases. There's the phase where my finger is being pushed by my brain through the nerves to type a given key. This is the "press" phase. It's the effort required for me to get a guaranteed actuation from a given key. After the press completes, I enter the "recover" phase. This is the phase where the finger is transitioning to a new point where it can then enter the "press" phase for another key.

The recover phase needs to be as soft and as relaxed as possible, because this is a time when I am not spending any effort to press a key, and I need to move my fingers quickly through space that has no resistance (or opposite pushing resistance from the spring of a key). Here I need to be as relaxed as possible and moving my fingers as fast as possible.

The press phase ideally is a sharp, fast, active pressing motion that gets activated quickly and lasts only the bare minimum amount of time necessary to activate a key. Any time that is spent in the "press" phase after the key has been successfully actuated is a wasted motion, and slows me down and affects the rhythm and efficiency of the typing, since the press phase is the one that requires the most energy and has the greatest potentially to reduce the overall dexterity of the fingers.

So how can we improve our ability on these two phases? Well, it all comes down to how short and fast we can make the press phase. Some people talk about this by saying something like, "don't bottom out," or other sorts of references to using a full travel keyboard and not fully depressing the key. The idea there is that bottoming out supposedly requires more effort and therefore is less efficient and wastes effort, as well as making the typing process less comfortable.

While the idea of not bottoming out seems to make sense to some people, after spending a lot of time speed typing and a lot of time watching other people speed type, I have come to the conclusion that not bottoming out isn't really the right perspective on the keypress. Instead, what we're really trying to do is clean up the amount of mental energy (transmitted via the nerves) required to create a reliable press. The more distinct feedback loops that we require in a given press in order to be reliable, the slower we will be. Ideally, we want a single neural signal to be sent out with a single "twitch" reaction to occur with the minimal amount of effort that reliably actuates the key and completes the press. In this way, that little burst then gets turned into a burst pattern over a group of letters, and so forth, until you end up typing whole phrases as a single burst in your mind.

But almost no one really works off of a single burst for a given press, at least not traditionally. What I found out when I looked at my fastest typing sessions as well as those of others, working with various keyboards, was that things like bottoming out was the norm. The fastest typists were bottoming out all the time. I was bottoming out all the time. Trying to chase the idea of not bottoming out wasn't actually going to make me any faster, and it wasn't a good metric to use when thinking about or evaluating a keyswitch.

How does this all relate to tactility? Well, the tactile bump and the audible clicks have traditionally been viewed as a way to send a signal back to your brain that the key has been successfully pressed. Non-linear keypresses work by having a force curve that isn't linear or smooth. Instead, they have a "break" in them which signals a shift from the force requirements to press the key further downwards stop going up and start to decline. This has come from the famed IBM Buckling Spring model, where the key press was linear up to a point where the internal spring "buckled" and then the force disappeared, the key actuated, and you were done with the typing. Typists like this model because of the amount of feedback that they get during the press to give them additional information about when they have pressed far enough down the key to actuate the key.

But it was gaming with Topre switches that made me look at things a little differently. See, Topre switches break fairly high in the travel (at least qualitatively) and then actuate somewhere in the middle of a full travel. They have little hysteresis because they are a capacitive switch, but the tactile break on a Topre switch is triggered by the dome on them, so they don't have a clear connection between the break and the actuation. A lot of people talk about how comfortable Topre switches are because they are so smooth, but others also complain about bottoming out too easily with them. The problem I encountered when trying to game with them was that I found it extremely difficult to accurately trigger simultaneous actuations of two keys at the same time as necessary to trigger the Mercy Superjump.

Topre switches really reveal a critical aspect about the relationship between tactility and actuation precision. With the Topre switch, when typing fast, my strategy  is to fire the press with enough force to overcome the initial wall, and that initial force and surprise will generally be enough to create momentum to actuate the key through the rest of the press. In other words, I am sort of launching my fingers out at a fast rate, and then relying on momentum to do the actual actuation. And if you look at other tactile approaches, that's really what's going on there. You create a feedback loop where you increase your force to press far enough to break through the tactile bump, and because you have an excessive amount of force for the amount of resistance after the bump, your finger flies through the rest of the press at the right speed and actuates the key. A key like the Cherry MX Browns or Blues have the bump and the actuation a little bit closer together, and the Buckling spring have them right at the same spot, but the Topres break early and actuate further away from the break. In all of them, though, the key insight for me was that the bump introduced a type of "stored momentum" that was let loose once the wall was broken, and that this momentum is what was being used, primarily to do the actuation, in all but the buckling spring case.

With a little analysis, you can see why this works so well for typists, but why gamers will sometimes not prefer it, and why I started to have trouble with superjump and Topre switches. For typists, the fact that you have created that excess amount of force means that you are going to get a very reliable actuation and your chance of "under pressing" is almost none. If you have enough force to make it through the break, then you almost surely have enough force to also be guaranteed to make it through the actuation distance and trigger the key. The feedback loop created by the tactile bump means that you also get the mental clarity to ensure that you know that the key was pressed, which allows you to track a "bad press" because the difference between not hitting a key hard enough and hitting it properly is pretty visceral.

But typing is often not limited by your raw ability to quickly spam a limited number of keys. There's a much wider window in when the key can actuate and still be in the right order than with some gaming commands. As long as the key presses actuate in the right relative order, everything is fine, even if two of them actuate sort of close together, and two them actuate further apart in time, as long as the relative ordering is correct, then you're fine. But in gaming, sometimes, you need to correlate a keypress precisely with something else happening, such as a second key press or something else happening in the game. And this is where those linear keys start to come into play. The linear keys are easier to "spam", and they are easier to correlate.

The key insight here is that using a tactile key switch means that your key press isn't actually tied strictly to the actuation, but it is tied to the "bump" or break in the wall. You're not actuating a key precisely, you're just triggering a set of motions that will eventually result in the actuation within a time frame that is acceptable to you. This lack of actuation precision isn't a problem with typing, because you are in control of the pacing, but when doing the superjump like I was trying to do, my inability to precisely control the exact moments when two keys actuated meant that I was more likely to miss an ability than I would have been otherwise.

This is all well and good for gaming, but what does this have to do with fast typing? After all, didn't I just say that the tactility didn't seem to matter much to typing because you could control your own ordering of keys? Yes! But I also talked about effort. And here's where I've started to change my tune a little bit.

When I began to think about it a bit, I started realizing something. When typing on a tactile key with a distinct wall, in order to overcome that wall, you're introducing excessive force that is almost guaranteed to result in a press phase that is longer than necessary that involves more momentum and longer travel than necessary. This requires more recovery and also means that there will be more, unnecessary time spent in the press phase. Remember that we want to make the press phase as short as possible, and to enable as good a recovery as possible. Tactile keys require more effort and trigger more involved feedback loops. The tactile key is fundamentally creating more neural feedback that we are relying on to just whether a key has been pressed or not. In other words, it's a bit of overkill, and it is using up a little too much brain power and muscle juice, while increasing the required effort of typing on a neural and muscular level.

What I came to realize is that what you really want is a way to create an instant neural burst that is tied directly with "actuation" and not with breaking a wall. Something that requires no feedback to know that it has occured. You simply "trigger" the idea and don't rely on much feedback to know that it has happened because it requires so little effort. There needs to be enough forgiveness in the model to account for the level of precision capable by human fingers, but you don't want to have much more than that. In other words, I want to be able to think about twitching a finger and know that a letter is going to actuate because of that. I want to make it possible to do this with the minimal or very minimal amount of muscle effort, and get into the recovery phase and on to the next press as fast as possible without having to worry about extra effort or time spent in the press phase.

The problem with the traditional set of keys is that linear switches didn't do this. You had full travel keys, and then you had some low travel keys, such as scissor switches. With the full travel switches, the balance between force and travel meant that the linear Reds and others like were unreliable to actuate without excessive force to overcome the 2mm travel required to make the key actuate (4mm total travel). The tactile keys helped to provide the necessary feedback to overcome this by ensuring that you always got enough force, at the cost of excessive follow through. This is why you always ended up bottoming out more with tactile keys when you were typing fast, because you needed to be able to hit things and move on as fast as you could, so you were going to overcompensate in order to get reliable actuation.

There were designs with less travel that actuated sooner, but the problem with those was that they were not reliable and they didn't provide consistent feel. The scissor switch chiclet keyboards were where this idea first started to come into being for me, because I realized that there was something about them that was contributing to them being faster than a membrane keyboard would normally be with full travel, but they still degraded too quickly in feel and the actuation accuracy suffered because of that. They started requiring excessive force too fast and didn't actually result in a reduced effort on the keys.

But the Mercy superjump fiasco has caused me to think about this problem again, and I went searching. I knew what I wanted. I wanted a key with very low profile and small total travel. This would ensure that if I overcompensated, I bottomed out fast enough that I could recover with less space to move. I wanted a key that didn't require excessive pre-force leading up to the actuation. The idea here is that I wanted to be sure that merely thinking about actuating the keys and then twitching my fingers were guaranteed to get results. This meant that I either needed a very, very subtle tactile key, or a linear switch. And then it hit me that if I combined a low profile total travel with a linear switch that had a very, very short actuation travel, this would overcome the traditional issues of typing on linear switches.

As it turns out, Corsair has just such a key from Cherry. The Cherry  MX RGB Low Profile Speed switch in the Corsair K70 Rapidfire Low Profile MK.2 keyboard was amazingly close to what I wanted. Ideally I would have had it in a tenkeyless model and I would have liked even less travel than what the Corsair provides, but it meets the main requirements of having a linear switch with a very short actuation travel and a short total travel.

This switch is what I'm now experimenting with. The idea here is that having a 1mm actuation instead of a 2mm actuation on the usual 45g key actuation force means that my "quick twitch" signal that is sent from my fingers is enough to guarantee an actuation of the key under nearly all circumstances, essentially eliminating "under pressing" the key and not getting a full actuation. On other linear switches with the full actuation travel of 2mm, I would very often find myself not actuating the key because I didn't press the key far enough. I do not have this problem with the 1mm actuation. Furthermore, the low profile total travel means that I will bottom out faster, and that leads to faster recovery.

The whole combination is pretty amazing so far. It's extremely comfortable, and it allows me to simply trigger my fingers without over-typing. The end result is that any bottoming out I might do is pretty light, which was always hard to do when I was using other keys, and this is perhaps the first keyboard where I can type fast and also not bottom out at the same time without having to consciously work on one or the other. In most other designs, the result would be either bottoming out to maintain max speed or a much slower speed to make sure that I wasn't bottoming out too much. It's very good for simply making letters show up on the screen, and in the days that I've owned the keyboard, I've already been able to break my old records. I'm not able to get consistent 140WPM results with this keyboard on some texts and even get regular 150+ WPM results.

Those who have followed my other keyboard thoughts will know that my previous records were 153WPM on the Truly Ergonomic keyboard with Browns. I broke that in a day with this keyboard at 154WPM. But what was more surprising was my ability to do this even while tired, and to consistently get in the 145+. On one text in 15 second bursts I was able to consistently get 155+ results. That's with a different piece of software, so I am not sure that the results are exactly equivalent, but at the very least it shows that this keyboard really is fast.

I can also report that it did fix my issue with superjumping with Mercy. :-)

So, let's recap. By rethinking things a little bit, I optimized my keyboard approach by focusing on my ability to create precise key actuations with minimal effort and not rely on an explicit neural feedback loop because I could be confident about the keys actuating without it. This helped me to reduce the time and energy spent in the press phase and improve my recovery phase results. I did this by going with a low profile linear keyswitch that has a 1mm actuation with a 3.2mm travel, versus the standard 2mm and 4mm travel. Going with a linear switch in this instead of doing something tactile, combined with the short actuation is what allowed me to reduce the effort required to type, because I don't have to overcome the hump and I can actuate reliably and precisely with a 45g key without over-typing at speed. It also ensures that I can get much more accurate actuations and really push the limits of how rapidly I can chain keys together and still maintain a desired ordering.

This may all sound like a great thing, but I also want to put out some warnings to people. Your mileage will vary from mine. I've been typing a long time, and I have spent a lot of time with a lot of different keyboards and I know what I want. If you are not already regularly typing above 120WPM then I doubt that you are going to find any measurable difference here. We're talking extremely diminishing returns versus just getting a good, solid basic keyboard. It's also true that there are a lot of people typing better than me with other keyboards. Keep in mind that hardware and software all matter, and it's not just about the right keyboard, but also recognizing your own limitations and adapting to your own needs.

Additionally, I would caution you that I had two very specific goals in mind with pursuing this. I wanted to improve my typing speed together with my typing ease. I also wanted to improve my ability to play games. If I were just worried about one or the other, I might not have had the same set of design criteria. While I used the need for precision in gaming as a catalyst for improving my typing speed and comfort, it is still true that you don't have to have as much precision with your typing as you do with your gaming, because you are the master of your own rhythm when you are typing, and you don't have to hit two keys at the same time with such precision until you get fast enough that your SHIFT speeds and the like start to matter.

Because I was only focusing on raw performance and the like, I was willing to intentionally give up synaptic/neural feedback. Don't under estimate how strange that can be. One of the reasons that a lot of people like the tactile bump on their keyboards is because it gives them more confidence. It creates a better feeling for them and is more satisfying. I cut out that feedback loop with these linear switches. The only feedback you are going to get is the linear spring response from the key and the hard wall that you hit when you bottom out. In other words, for most people, these keys are going to feel mushy!

Keyboarding can be as much about emotional and sensory enjoyment as it can be about raw results. That's why some people gravitate towards manual transmissions even though they aren't faster than dual-clutch or automatics and also why Formula 1 doesn't use manual transmissions. I may be able to type faster with the linear switches more consistently than with other keys, and more effortlessly, but that comes at the cost of giving up the "clicky" typing experience, because the linear keys only give that mild, mushy spring feel. You might think of it like typing with trampolines for keys.

I am very confident with my key stroke and with signally my fingers to type what I want. I don't need the feedback to know that a key should go or not. I'm okay with not feeing it as long as I know that it is happening in the way that I want it to go. But for many others, that's not going to be the case, and even if they might be confident with where their fingers are going, they won't like not getting the feedback in typing from their keyboard.

For those of you who want to have a conversation with your keyboard and that loop of feeling the keys do things, maintaining that tenuous connection to the ancient art of mechanical devices, then I would recommend you stay away from the linear switches. But chances are, you are still rocking Cherry MX Blues or a Buckling spring anyways, and you wouldn't care about this.

In many ways, I write this article just to share my thought process that I often go through when working through a keyboard choice and working on improving my skills with keyboarding. I don't know that everyone should be going out and trying the Chery MX Low Profile Speed switches (exclusive to the Corsair K70), but the above should give you a clear model of why I decided to go with them, and why I was pretty confident I was going to like them even before purchasing the keyboard. I was familiar with how linear switches felt already, I knew what was wrong with them precisely, and I knew precisely what things were going wrong with my current keyboard setup that I wanted to fix. All of this meant that I could specify exactly what I wanted to see in a new keyboard and see what new switches were out there that met my needs the closest.

I don't think the Low Profile Speeds are exactly what I want, either. They are the closest I have been able to find, but I really wish I could get a low profile switch in the optical IR range, rather than relying on the traditional mechanical. But the IR switches that I have seen have a 3.5mm travel or thereabouts with a 1.5mm actuation and sometimes they are putting tactility into them. I'm not convinced that is enough of a change to warrant them. It really is the 1mm actuation (tolerances mean that it is not always exactly 1mm) and short travel that make the difference to me when combined with a linear switch.

If anyone is listening, I would be really curious to try out a 0.75mm actuation, 2mm total travel, linear switch with a 30 - 40g actuation force. I suspect that 30g might be a bit too low, but I think I would like the linear switches even more if they were in the 35 - 40g actuation force range. I think the 1mm actuation is probably fine, but I wonder what would happen with an even shorter one, and I would be curious to try out a 2mm travel to see if that's a little too short or not. I do find that the new near-zero travel Mac keyboards are interesting, but I think they have the problem that I can't seem to get my fingers to relax enough to use them efficiently yet. I still want a little bit of travel right now, but not much. I don't mind bottoming out, but I want to bottom out after my fingers are already done with the press and moving to recovery, not while they are still actively in the press phase. Bottoming out isn't a problem if you have already reached recovery phase, but if you bottom out while you are still in press phase then that's a bit of an issue. I think the 45g actuation force is high enough to prevent people who ride the keys from accidently actuating them, but I do think that it would be interesting to try them even lighter, because I don't ride the keys, so I wouldn't have a problem with accidental actuation, and I could have an even lighter flow over the keyboard that way. I'm curious to know whether 2mm would be too short to avoid a bottoming out during press phase or not.

TL;DR — when I say that I'm a typing geek, I mean, I'm a typing geek.