Why Did I Put Padding Inside a Keyboard?

I didn’t find keyboards. I heard one.

It was a YouTube Short. An ASMR video — somebody’s hands, a board I couldn’t identify, no face and no explanation and nothing being sold. I hadn’t searched for it. It arrived because something decided it should.

I watched it, and then I went to Amazon and started looking for whatever I had just heard. That took about a week, which I would have described at the time as doing my research. At the end of it I bought a Redragon Draconic Pro K530, mostly because it was affordable.

That is a strange way to buy an object, and it took me years to notice how strange. I have never bought a monitor because I heard one. Nobody has ever been talked into a mouse by the noise it makes. A keyboard is close to the only thing I have owned that reached me as a sound — not an advertisement, not a spec sheet, not a review I went looking for. A stranger typing, recorded close, with the microphone doing most of the persuading.

I want to be careful here, because this is the part I got most wrong. I thought I was buying my way back to something.

The golden age I was restoring did not exist

The story I had absorbed, without ever once checking it, went like this: keyboards used to be built properly, then they got cheap, and buying a mechanical keyboard was a way of undoing that. A restoration. Me, personally, correcting an industry’s mistake with my own money.

The first half of that is true, and it’s the half everyone repeats. The IBM Model M went into production in 1985, ran buckling-spring switches, and weighed between two and two and a half kilos, most of it a sheet of steel. Mechanical keyboards were not a luxury then. They were what a keyboard was — the Model M came bundled with the computer. There was no premium tier, because there was no other tier to be premium against.

I had a theory about why they felt so good, and I should say plainly that it was wrong. I assumed the tactility was a side effect — that the bump was just what you got from the cheapest reliable way to make a key go down and come back up, and that feeling good was a lucky accident of engineering for durability.

It wasn’t an accident. IBM’s buckling-spring patent, filed in 1977 by Richard Hunter Harris, puts it in the summary of what the mechanism is for: a “non-teasible, snap action, tactile feedback, key mechanism of extreme mechanical simplicity and high reliability.” Feedback sits in that sentence beside simplicity and reliability, not behind them.

And it goes further than feel. The patent describes the snap as producing “a tactile feedback to a human operator due to the sudden decrease in force,” and then, in the same breath, notes that it “also produces an audible feedback since the sudden pivoting of the rocker member produces a clicking noise.”

The click was engineered. Not tolerated, not a by-product somebody failed to suppress — written into a patent as feedback, on purpose. Somebody in 1977 sat down and decided what a key should sound like, and IBM shipped that decision to offices by the million.

Which makes the rest of the story worse, not better.

Because the industry didn’t lose that by accident either. As PCs went from office equipment to household objects, manufacturers replaced dozens of individual mechanical switches with one moulded sheet of rubber domes. Same letters on the screen, cheaper to make, at a scale where cheaper to make is the whole business. Nobody misplaced the knowledge of what good felt like. They had it, it was written into a patent, and it was cheaper not to use it.

And restoration is still the wrong word for what I did, just not for the reason I thought. Nobody in 1985 chose a Model M. It wasn’t taste. It was a cost of doing business — a corporate purchase, in a corporate beige, bought in bulk for people whose job was typing. You cannot restore something that was never yours to begin with. It was a workplace standard, and I was buying it as a personality.

Consumers didn’t get bored of mechanical keyboards. Accounting did.

The home buyer went along with it, and had reasons. A Model M in a shared living room is loud in a way that is genuinely antisocial. It’s heavy. It costs more. Membrane boards were quiet, light, cheap, and survived a knocked drink.

I should be fair to them here, because the enthusiast telling of this story never is. The best technical reference I could find on rubber domes describes their feel as softer and more comfortable than a metal spring, with cushioning at the end of the stroke that removes the ringing a spring leaves behind. In its own terms that isn’t a compromise, it’s a different set of choices. What rubber domes do badly is age — the rubber stiffens, the domes eventually split, and the board you liked in its first year is not the board you’re using in its sixth.

So the thing I thought I was restoring was never mine in the first place. It was a workplace default that got optimised out of existence for perfectly sound reasons, and then, forty years later, sold back to me as a personality. I hadn’t recovered a golden age. I’d bought the reissue.

1985 to 2024

What a keyboard shipped as mech, then dome

Apple laptop keyboards butterfly 15–19

Butterfly repair programme 2018–24

1985201520192024

Then I opened it up

The K530 was a 60% board — 61 keys, no number row, no arrow keys, no function row, everything above the letters reached by holding a modifier. White. It came with whatever switches it came with, and I could not tell you now what those were. Redragon sold that board with linear reds in some markets and tactile browns in others, and the Indian listing for the one I bought has since gone dead. I never checked at the time, which tells you how much I understood about what I’d bought.

What I remember is the sound. Sharp, hollow, a bit cheap. I called it clicky, and I decided I wanted it gone.

I’m fairly sure now that I was wrong about where it was coming from, and that mistake is the whole rest of this post. I thought it was the switches. It was probably the box they were sitting in.

But I didn’t know that then, so I started spending evenings on it.

First the switches. I swapped the stock ones for browns — tactile rather than clicky, a small bump partway down that tells your finger the key has registered so you don’t have to drive it into the base. That worked. My hands liked it immediately.

It also barely touched the sound. Which was not what I expected, and is the single most useful thing I learned from any of this.

The sound is not in the switch.

A keystroke makes noise in several places at once, and the switch is only one of them. There’s the keycap hitting the top of the housing on the way down. The stem snapping back up on release. And then the part that dominates all of it: the keyboard is a hollow plastic box, and every one of those small impacts is happening inside it. The box rings. That’s the high, hollow, slightly cheap ping that a bad keyboard has and that no switch swap will fix, because the switch isn’t making it — the empty space underneath is.

“Thock” is just what’s left when you take the ringing away. Damp the cavity, let the mounting plate flex a little instead of transmitting every hit straight into the shell, use a keycap thick enough that it doesn’t sing on impact, and the same switch that sounded like a biro sounds like a book closing.

Which is why people put padding inside keyboards. It isn’t a superstition and it isn’t cosmetic. It’s the correct fix for the actual source of the noise.

I want to give myself that one, because I got there by hand: two mods, two different targets. The browns changed how it felt. The padding changed how it sounded. Most of the way this hobby gets talked about collapses those into one word — “switches” — and they’re not the same problem at all.

But here’s what I couldn’t see from inside it.

If the sound lived in the switch, there would be a finite number of switches. You could work through them, find yours, and be finished. Because the sound lives in the assembly — the mount, the plate, the case material, the foam, the keycap plastic, the desk it all sits on — every one of those is a variable, and they interact, so changing one changes what all the others do.

There is no configuration you can arrive at and know you’re done. Not because people chasing it lack discipline. Because done is not defined. The thing has no terminal state, and I was treating it as though it had one that I just hadn’t reached yet.

I bought a second board eventually, an Aula F75 Pro. I never opened it. I’ve only realised writing this that the chase quietly stopped there, and at the time I didn’t notice it stopping.

The enemy had already been dealt with

The other thing I had wrong was bigger, and I’d held it longer: that thin keyboards were the problem.

There was good evidence for it, and it all came from one place. Through the 2000s and 2010s the entire industry optimised for thinness, because thinness is what a laptop is judged on. Scissor mechanisms and flat chiclet caps were how you kept a shallow key stable. Then Apple went further, and in March 2015 shipped the butterfly in the 12-inch MacBook — a mechanism chasing almost no travel at all, around 0.7 mm of it, against roughly a millimetre for the scissor keyboard that eventually replaced it. It was a disaster. It jammed on debris small enough to be invisible. It bottomed out like tapping a countertop.

The scale of it is easiest to read in the repair programme. Apple opened one in June 2018, widened it in May 2019 to cover every MacBook, MacBook Air and MacBook Pro with those keys, and only closed it in November 2024 — six and a half years after it started, and five years after the keyboard itself had been discontinued. Apple was still buying back that decision half a decade after it stopped making it.

I took that, and I generalised it. Thin keyboards are bad. Then I spent money on the conclusion.

How far a key actually moves, drawn to scale

Butterfly
12-inch MacBook, 2015 0.7 mm

Scissor
the Magic Keyboard I use now about 1 mm

Cherry MX
full travel 4 mm

I’m typing this on a 2025 MacBook Pro. It has a scissor keyboard — the same basic mechanism I’d written off — and it is fine. Not fine as in tolerable, fine as in I sat down to write two thousand words on keyboards and never once thought about the one under my hands.

The butterfly was the failure. Scissor was fine before it and fine after it, and the company that made the mistake went back to it — the 16-inch MacBook Pro in November 2019, the rest of the line by the following May. Which means the villain in my story had already been beaten before I bought my first mechanical keyboard. I’d taken one company’s five bad years and turned it into a law about an entire category, and then bought a cure for a disease that was already in remission.

And then it stopped

I don’t have a desk setup any more. The boards went with it. I type on the laptop now, and that’s all of it — there was no decision, no moment of seeing through anything. Circumstances changed and the habit didn’t survive them.

For a while I missed it. That’s the honest part and I’d rather not smooth it over: there was a stretch where typing felt like nothing, where the absence of the sound was a small flat thing I noticed several times a day.

And then I didn’t miss it. I can’t tell you when that happened, which is itself the finding.

What the padding was for

I’m not going to retract the good part. The evenings spent with a keyboard open on the table were genuinely good evenings, and taking something apart to change how it behaves is one of the few honest pleasures left in a stack of technology that is mostly sealed. That stays.

But I notice what happened when the sound stopped being in front of me. I stopped wanting it. Not reasoned out of it — I didn’t argue myself down. It just went, the way a want goes when nothing is topping it up.

So: why did I put padding inside a keyboard? Not because keyboards used to be better. They were, and it turned out not to matter. Not because thin keyboards are bad. They aren’t, and the one I’m using proves it. I heard something I liked, in a recording of somebody else’s hands, that I had not gone looking for. There was no point at which the thing I was chasing could be reached, so I kept going until circumstance stopped me for me.

The sound was real. The wanting was real too. It just didn’t start with me — it started with a video I was shown.

What I checked, and how sure I am

Every claim in this post that could be wrong, with where it came from and how much weight it will take. One of them didn’t survive checking, and it was the one holding up my whole argument about the eighties.