The Keyboard I Stopped Noticing, and the Three I Sent Back First

I bought four mechanical keyboards in eighteen months. Three went back or into a drawer within eight weeks, and one has been under my hands every working day for a year without my thinking about it once. That last sentence is the entire review, and the reason I now distrust every keyboard review written in the first fortnight.

Here is the pattern in the three failures, which is more useful than the specifications. Each one had a property I could describe in a sentence and that felt like a reason to buy. Very low actuation. A particular switch sound. An unusual layout that promised fewer hand movements. All three properties were real and none of them was wrong. What killed each one was something I had no vocabulary for at purchase time and could only find by using it.

The low-actuation board failed on typos. It registered on the way down at a depth my fingers rest at when I pause to think, so a moment of hesitation over a sentence produced a row of characters. I adapted within a fortnight by hovering, and hovering is a small continuous tension that I did not consciously notice for two months and that my forearms noticed immediately.

The one bought for its sound failed socially. It was genuinely lovely and I liked it, and I share a room with someone who did not, and a keyboard’s noise is a thing you cannot evaluate alone because the person it is loudest for is not you.

The unusual layout failed on everything else. I got fast on it in about three weeks, which is the part reviews cover, and then discovered that I was now slow on every other keyboard I touched — a laptop in a meeting, a colleague’s machine. The cost was not learning the layout; it was that I had made all other keyboards worse, permanently, and I do not spend my entire working life at my own desk.

The one that survived is unremarkable and I am going to describe it in the boring way it deserves. Standard layout, tactile switches with a bump at roughly the middle of the travel, moderate weight, no software, plugs in and works. Nothing about it is interesting and I have not thought about it since about week four, which is the highest thing I can say about a tool.

What I would take from this into buying any peripheral: the properties that make a good review are the ones you can perceive immediately — the same reason a monitor arm beats a better panel and never gets reviewed — and the properties that determine whether you keep it are almost all delayed, which is the same thing I found with headphones worn for eight hours. Sound is immediate; the fatigue of hovering is not. Speed on a new layout is immediate; the cost to every other keyboard is not. So a first-impressions review is not a bad review, it is a review of a different question than the one you are asking.

The practical version is a rule I now apply to hardware generally: buy the version that is standard in the dimension you have no experience of, and only be unusual in the dimension where you have already learned what you want — a trackball for a sore wrist is the exception that proves it, because the problem was specific enough to justify the odd shape. I knew I wanted a tactile bump; I had no idea what actuation depth suits how I type, so the standard one was the right bet, and it was.

I would also say plainly that the three failures cost me about £320 and were not wasted, because I could not have reasoned my way to any of those three findings. Some of buying is just paying to find out.