The hyperoptimiser's mechanical keyboard trap
Mechanical keyboards give you the power to be maximally opinionated about your keyboard configuration. You can choose keycaps, switches, layout, firmware-level settings, and even materials to affect the overall touch and sound. All contribute to an individualised setup that may only work for you—and that is part of the charm.
I have been using a split mechanical keyboard for ~2.5 years already. It is the Iris from Keebio. Over time I have refined everything to be as I want. My preferred keycaps are the Cherry profile, specifically the Keychron Dolch Red set that I have had for ~2 years now. The switches I use are Cherry Silent Black since ~1.5 years ago. And the firmware has the exact mappings I like, which I settled on fairly early on. Details here: https://github.com/protesilaos/qmk/tree/master/keyboards/keebio/iris/keymaps/prot.
My personalised keyboard is the distillation of experimentation that unfolded over the course of several weeks. I tried a tall keycap profile, namely, MT3, as well as a uniform MOA set, only to realise that I prefer the Cherry profile. I further learnt that I am more comfortable with the cylindrical top surface of the Cherry keycaps as opposed to the spherical tops of the other options I tried. With switches, I used linears (Keychron Silver, Gateron Yellow) and a tactile variant (Keychron Banana), to ultimately understand my exact needs, which led me to the Cherry Silent Blacks. I want my keyboard to be quiet and firm to the touch. I had also figured out that I dislike low profile switches/caps, so the MX platform is what I am interested in.
What I have is not a “perfect” setup. If I start thinking about variables with margins for optimisation, I can produce the following non-exhaustive list of items:
- reduce the switch stem wobble;
- lower the sound profile even more;
- have switches that are about 10g lower on the force curve;
- have slightly taller keycaps at the topmost row, to make them even easier to reach;
- change the case of the keyboard to something more elegant that also affects the acoustics;
- use my own colour scheme for the keycaps and even the whole board.
This, however, is a slippery slope. To improve the stem wobble, I either have to modify all my switches by adding film to them or I have to buy new switches. Same idea for the springs to control the overall weight. My first choice would be to do market research in the hope of finding an existing solution. And if I do it myself, I might not like the results, so I will be back to searching for a commercial solution. But would, say, the extra 50+ EUR I would need to pay truly improve my experience? Or would I be buying something that might open up horizons for new optimality gains and thus a renewed commitment to experimentation?
Suppose I eventually lower the bottom out event to 70gf from its current 83gf level. Ceteris paribus, this brings down the actuation point by 13gf, meaning that I must now get used to the slightly lighter touch. I must also hold myself back a little bit so as to not bottom out hard. I can adapt, though it takes time. Imagine this takes me a couple of weeks or maybe even a month of being relatively sloppy. At the end of that exercise, I am left with a slightly lower force curve which, in theory, is less tiresome to type on. So I count that as a longer-term win.
Is it though? What about the restraint I am applying to not bottom out? If I am bottoming out hard, then I am wasting the extra force regardless. Can I reliably measure that marginal change to my setup to have an exact reading of the delta to my efficiency? How exactly to factor in the adjustment time? And if the prevention of injury is my primary concern, why not simply take more breaks here and there? A couple of minutes each time contribute to long-term wellness. Plus, if I do not even have any problem with the existing force curve, I am merely hoping that there are improvements to be made and then act with the wishful thought that I will somehow solve my unknown problems preemptively.
The same kind of thinking applies to every other factor. One tweak addresses a given consideration while creating a new one. Such is the nature of systems. The reason is that the process of optimisation is about finding the set of trade-offs you are willing to live with, given the constraints you have determined or must take for granted. Optimisation does not lead to perfection but only to a notionally imperfect state that meets your requirements. In other words, “good enough” is the closest to perfect you shall ever get.
The hyperoptimiser’s trap consists in believing that they can reach a point of perfection through continuous tinkering. With mechanical keyboards this practically translates into a money-intensive and time-consuming endeavour. Now, there is nothing inherently wrong with buying stuff you like (what may be problematic is ruining your finances and other externalities, but that is a different discussion). Though understand that you are doing it because you find it fun, not because you will actually reach perfection.
This discussion extends to the number of keys a keyboard should have. I think the sweetspot is the 50% to 65% range, depending on priorities. Though this too is the kind of mire that you can avoid by accepting the “good enough” state of things.
In principle, I can work with fewer keys as I am adept at switching between layers. Though I can also be equally productive with a full-sized keyboard. This is because what I fundamentally produce on the computer is not governed by the input device: it is in my head and it involves clarity of thought as well as precision of statement. The fancy keyboard is a convenience that contributes, together with the overall desk setup, to a sustainable office arrangement. And, yes, it can be fun!
When I talk to people who are already into mechanical keyboards and who ask me about my opinion on the new custom board they plan to buy, my answer is always the same: admit that you are making a purchase for the thrill of the experience, as what you have is already good enough.
Whether you reduce the number of keys or move things around in your layers, you are not directly improving your productivity. You might add to your sense of comfort, which can indirectly affect your productivity. Though this is hard to measure reliably once we start talking about marginal gains and rigorous scientific methods of understanding them. Again, I am focused on marginality, not the big kind of changes whose effect is easier to notice.
Another thing to keep in mind is that the most efficient implementation is the least adaptable. This is because efficiency can only be realised within constraints, so the most efficient necessarily is that which has best adapted to its constraints. As such, even a tiny tweak to the constraints—constraints qua parameters—can have far-reaching implications, which undo the once present level of efficiency.
To be concrete, if you settle on, say, 32 keys on the assumption that you will be using Vim long-term, this will work well for you for as long as you are on Vim, but may give you trouble if you decide to switch to Emacs-style keybindings because those make heavier use of modifier keys. And if you have home-row modifiers set up, now you might want to reconsider their placement, tweak their timings, potentially rework your layers, et cetera. One small change to the system always becomes more than “one small change”. What once was the most efficient, then, will have to be remade anew (and so, did you actually gain productivity from all those roundabouts?).
In this regard, a keyboard with more keys does not need any further changes. Plus, you do not have to use all the keys, anyway.
This is not to argue for or against a specific choice. There is no clear right or wrong here, as what works for one is unsuitable for another. My point is that there always are trade-offs, diminishing returns, and the illusion of approximating perfection. The sooner you accept this fact, the easier everything will be.
As for buying new hardware, once you admit that you are doing it at least in part because you find excitement in the activity, then you are framing the issue in a way that gives you clarity. It is a consumer choice at heart, rather than the pursuit of some elusive productivity boost. Once it is impressed in the conscience as yet another purchase, you start asking financial questions. Do you really want to spend more money on this keyboard business to ultimately get nothing substantial out of it? What if you saved that money to, say, buy new clothes, or some furniture, or go on a trip abroad, or build a house?
My thinking is consistent for newcomers: you will have to play around with at least some of the variables to figure out what your preferences are. Be mindful of the trap though. Practice thoroughly to give whatever setup a fair chance. You want to get used to the keyboard the way it is so as to diminish the influence your prior expectations have on your present decision-making.
No decision need be final, after all. You might have different requirements in the future, so you can adapt accordingly. Do not make the mistake of trying to anticipate those requirements. Robust future-proofing is the kind of overthinking that can never yield reliable results as you are fundamentally trying to optimise without knowing the constraints. This is why there is no pressure to optimise your initial choices beyond the realm of the immediately knowable.
The appreciation of “good enough” will help you long-term.