Cover art for How Mechanical Keyboard Switches Actually Work
Episode 6729 custom

How Mechanical Keyboard Switches Actually Work

Standard Cherry MX switches actuate at two millimeters of a four-millimeter total press, registering keystrokes halfway through the stroke.

29:47 Transcript

0:00
0:00
Episode Details
Episode ID
MWP-6933
Published
Duration
29:47
Audio
Direct link
Pipeline
V5.36
TTS Engine
omnivoice
Script Writing Agent
GLM-5.3

AI-Generated Content: This podcast is created using AI personas. Please verify any important information independently.

The prompt

The prompt that started this episode, as sent in (production notes for the hosts removed).

I'd like a deep dive into the main families of mechanical keyboard switches. Start with the classic Cherry MX line, the switch that set the standard: how it works mechanically, what the colours mean (Red, Brown, Blue, Black and the rest), and why its patent expiry opened the door to everyone else. Then the other big names people actually buy today, like Gateron, Kailh and the like: what each one did differently, where they're better or worse than Cherry, and how the whole linear / tactile / clicky distinction plays out across them. What really separates one switch from another once you're past the marketing?

Nearly every modern mechanical keyboard switch traces back to Cherry's 1983 German design, an entirely mechanical assembly of just four parts: a spring, a stem, and two electrical contacts. The entire taxonomy of linear, tactile, and clicky switches relies simply on the molded ramp profile of the plastic stem pushing against a leaf spring. This purely mechanical design actuates halfway through its travel distance, meaning bottoming out adds unnecessary finger fatigue and noise. Furthermore, the popular narrative surrounding clone switches claims Cherry's foundational patent expired in 2014, yet records show the relevant 1984 patent actually lapsed in 2003. The folklore likely conflates legal expiration with Japanese manufacturer Hirose ending MX production in 2014.

Sources

What the research for this episode read before the script was written. Primary sources first.

  1. Cherry, #ORIGINALMX: Often copied, never reached! primary retrieved 2026-11-21, published 11/2019, by Friedrich Stiemer and Holger Dassler.
  2. Cherry, Cherry MX switches at a glance primary retrieved 2026-11-21, published 08/2017.
  3. Cherry, CHERRY MX2A: Exceeding Expectations primary retrieved 2026-11-21, published 2023-08-24.
  4. Cherry, CHERRY MX BLACK product page primary retrieved 2026-11-21.
  5. Google Patents, US4467160A Low profile switch primary retrieved 2026-11-21.
  6. Wikipedia, List of keyboard switches retrieved 2026-11-21, last edited 2026-08-23.
  7. Wikipedia, Cherry AG retrieved 2026-11-21, last edited 2026-09-07.
  8. Wikipedia, Keyboard technology retrieved 2026-11-21, last edited 2026-11-19.
  9. Hacker News thread on Cherry patent expiry retrieved 2026-11-21.
  10. GOLEM, How to choose keyboard switches? retrieved 2026-11-21.
  11. X-Bows, Kailh vs Cherry Switches: BOX vs MX2A retrieved 2026-11-21, updated 2026-10-26.
  12. MKB Guide, Cherry MX vs Gateron Switches: Complete Comparison 2026 retrieved 2026-11-21.
  13. MechKeyFoundry, Gateron vs Cherry Switches: Which Is Smoother? retrieved 2026-11-21.
  14. Keychron, Best Keyboard Switch: Gateron, Cherry MX, Kailh, or Keychron? retrieved 2026-11-21.
  15. NicheHive, Cherry MX vs Gateron vs Kailh: Which Switch Brand Is Best in 2026? retrieved 2026-11-21.
  16. kbd.news, Cherry MX2A Black vs Gateron Ink Black V2 switch comparison retrieved 2026-11-21.
  17. Keebnews, Kailh Box vs Cherry retrieved 2026-11-21.
  18. Mech Keybs, Cherry MX Blue vs Kailh Box White: Click Jacket or Click Bar retrieved 2026-11-21.
  19. Keeb Starter, Ultimate Kailh Clicky Switches Guide (2026) retrieved 2026-11-21.
  20. Deskthority wiki, Cherry MX retrieved 2026-11-21.
  21. GeekHack wiki, Cherry MX retrieved 2026-11-21.
  22. Telcontar.net, Cherry MX retrieved 2026-11-21.
  23. Xah Lee, History of Keyboard Mechanical Switches retrieved 2026-11-21, dated 2024-04-17.
  24. Adafruit, The History of Cherry MX Keyboard Switches retrieved 2026-11-21, published 2018-07-31.
  25. HK Gaming, Kailh Switches Enter the Scene: Clone or Contender? retrieved 2026-11-21, published 2015-09-05.

Mentions

  • Cherry MX Original mechanical keyboard switch line
  • Cherry MX2A Cherry's factory-lubed redesign of MX
  • Gateron Chinese switch maker, factory-lubed clones
  • Gateron Ink Black V2 Linear switch with proprietary housing blend
  • Hirose Japanese factory that made MX switches until 2014
  • Holy Panda Beloved tactile frankenswitch turned production switch
  • Kailh Switch OEM behind BOX and click bar designs
  • Kailh BOX Jade Clicky switch with metal click bar
  • Romer-G Logitech's proprietary switch mount
  • Topre Electrostatic capacitive switch, non-MX mount

Downloads

Episode Audio

Download the full episode as an MP3 file

Download MP3
Transcript (TXT)

Plain text transcript file

Episode Book (PDF)

The episode's record — date, duration, models, sources — with the full transcript

#6729: How Mechanical Keyboard Switches Actually Work

Corn
Here's a fact to anchor your morning: nearly every mechanical keyboard switch on earth is a descendant of one German design from nineteen eighty-three. Daniel's prompt this week goes straight at that. He wants the full family tree — how a Cherry MX switch works physically, what the colour code actually encodes, why the patent story let everyone else in, and then the part nobody answers honestly: Gateron, Kailh, the rest — what do they actually do differently? His real question is the last one. Strip the marketing off two switches that look identical, and what's left?
Herman
And it's a good time for it, because the answer has changed. Ten years ago you could say Cherry and clones and be roughly right. Now the clones are the market and Cherry's the heritage brand. But before we argue about who's smoothest, we should start with the object itself. Four parts under a keycap. That's the whole switch.
Corn
A spring, a stem, and two contacts. It's a biro with opinions.
Herman
It really is close to that dumb. The stem is the coloured moving part — the bit with the cross-shaped plus on top that the keycap clips onto. The upper housing guides it. A coil spring sets how hard you have to push. And the base, glass-fibre reinforced plastic, holds the terminals. No electronics in the switch. Zero. The intelligence is all mechanical.
Corn
Which still catches me off guard every time I think about it. We're in an era where a toothbrush has firmware, and the input device for the most information-dense activity humans do is a spring and a bit of plastic.
Herman
And that's not nostalgia on Cherry's part — it's economics that never broke. A purely mechanical contact is cheap, it's repairable, it survives decades, and it doesn't need a driver. There's a reason the design outlived the company's original product line. Cherry wasn't a keyboard company. They made all sorts of industrial switchgear — car switches, microswitches for appliances. The MX was one product among hundreds, and it happened to be the one that conquered the world.
Corn
So the press does what, physically? Walk me down the stroke.
Herman
You push the keycap, the stem goes down against the spring. Cut into the side of the stem is a ramp — and the shape of that ramp is everything. As the stem descends, the ramp presses on a leaf spring sitting sideways in the base. At the actuation point the leaf is allowed to snap across and touch its contact, and that completes the circuit between the plus and minus terminals soldered to the board underneath.
Corn
And when you let go?
Herman
The spring pushes the stem back up, the ramp catches the leaf on the return side, and the leaf snaps back to open. The reset is the same mechanism in reverse. And here's a detail worth sitting with: the ramp is a single piece of moulded plastic doing the work of a sensor. Linear, tactile, clicky — every switch in the family is the same four parts with a differently shaped ramp. Cherry didn't build different products. They built one product and changed one curve.
Corn
So the entire switch taxonomy — linear, tactile, clicky — is one mould difference.
Herman
At the stem, yes. That's the whole taxonomy's dirty secret, and it's why the clones could copy it so fast. You're not reverse-engineering a system. You're measuring one plastic part.
Corn
And here's the part everybody gets wrong, because I got it wrong for years. The keystroke registers before you hit bottom.
Herman
On a standard MX it's two millimetres of a four millimetre total travel. Halfway. You can stop there, hold the key at that depth, and the character has already appeared. Bottoming out does nothing except make noise and hurt your fingertips. Every datasheet carries both numbers separately for exactly that reason.
Corn
Two numbers, two meanings, one keypress. The other half of the travel is theatre.
Herman
Some of it is return-spring safety, but yes, mostly you don't need it. And the theatre has a cost. If you type with a heavy hand — and most people do, especially people coming from rubber domes — you're driving the stem into the base housing on every single keystroke. That's the impact, that's most of the clack, and that's the fatigue. The whole "mechanical keyboards feel better to type on" experience is partly just learning, unconsciously, to stop pressing once the switch has fired. Typists who make that adjustment say the board feels lighter within a week. Nothing about the board changed.
Corn
So the switch trains you, not the other way round.
Herman
The good ones do. It's the only piece of computer hardware I can think of that improves the user instead of the other way around.
Corn
Gold that cleans itself. My leaf medicine makes the same claim, with less evidence.
Herman
Your leaf medicine is not rated for a hundred million of anything. Okay — so that's the mechanism. And this is where the colour code starts to make sense, because the stem colour encodes four things at once. Cherry's own wording is that the colour indicates the operating force, the pre-travel, the total travel, and whether it's linear or tactile, with or without click. Force, distance, feedback type. One colour.
Corn
Which is elegant right up until the moment it isn't. But take the table in order — and start where Cherry started.
Herman
MX Black. Nineteen eighty-three. The original. Linear, zero point six newtons, two millimetre actuation, four millimetre travel. Everybody assumes the Red came first because the Red is the famous one, but the Red is the younger, lighter sibling — zero point four five newtons. It became the gamer default precisely because it's light: less force, faster fatigue recovery.
Corn
And there's a wrinkle in the Black's history people skip over, isn't there? The early ones versus later ones.
Herman
Right — the Black's reputation took a knock in the nineties because Cherry's tooling wore and the stems got scratchier, and a whole generation of writers declared Blacks "gritty" based on boards from that era. The tooling got refreshed, the switch got smoother, and the reputation stuck anyway. The hobby's memory is long and its measurements are mostly vibes.
Corn
And Brown?
Herman
The compromise. Zero point four five newton actuation with a zero point five five newton tactile peak — so same nominal weight as the Red, but there's a bump on the ramp you can feel partway down. The idea is you know the switch has fired without having to hear it. It is also, and I say this with affection, the most mocked switch in the hobby. People call it a scratchy linear.
Corn
Harsh but earned. I've typed on Browns. The bump is more of a rumour.
Herman
And yet — office buyers loved them, because they were the quiet mechanical. If you wanted clicky joy without your colleagues forming a committee about you, Brown was the answer. The mockery is a hobbyist judgement. The market voted the other way for years.
Corn
Blue is the loud one, and the numbers hide the interesting part. Zero point five newtons, clicky, four millimetre travel — but it actuates at two point two millimetres, not two. That's the only switch in the standard line with a different actuation point, and it's two tenths of a millimetre deeper than everything else. That tiny difference is a big part of why Blues feel different to type on, not just to hear. The click comes from a separate plastic click jacket that rides down the stem and slaps the bottom housing on the way.
Herman
And the click jacket is a weird piece of engineering when you think about what it's for. It contributes nothing to the electrical function. The switch would register identically without it. It exists purely to make a noise on your behalf — an audible confirmation that the leaf has snapped, for typists who worked in loud offices in the eighties and couldn't trust their fingers. It's a sound effect bolted onto a sensor.
Corn
And Blue's rated lower on lifespan, isn't it? All that slapping.
Herman
Greater than fifty million versus greater than a hundred million for the linears. The click jacket is the extra moving part that wears. Some of that's the mechanism, some of it, I suspect, is Cherry being conservative because the jacket is harder to guarantee.
Corn
Rest of the family — I want the pattern, not a catalogue.
Herman
The pattern is weights and throws. Green is a heavy Blue. Clear is a heavy Brown, beloved by people who lube things on purpose. The Greys are the zero point eight newton heavyweights, for knuckles that have opinions. Nature White is a mid-weight linear. Then the variants that change travel: Speed Silver shortens actuation to one point two millimetres and total travel to three point four, that's the esports answer to the same four millimetre problem. The Silent line damps the landing and trims to one point nine and three point seven. Low Profile drops total travel to three point two. And the newest, the MX2A-era Purple, is a light tactile at zero point three five newtons.
Corn
What I find remarkable about that list is how conservative it is. Forty years, and the whole range is still just the original Black stretched lighter and heavier, shorter and longer. Nobody at Cherry ever said "what if the switch were a different shape?"
Herman
They tried, a couple of times, and the market slapped them down every time. The MX mount — that little cross on the stem — had become the standard the entire keycap industry was built around. Once your mount is the standard, you're locked into your own success. Every radical idea has to survive the question "but will people's keycaps fit," and the answer was always no.
Corn
So the colour code is doing four jobs at once — force, two distances, feedback type. It's a colour wheel with a datasheet attached.
Herman
And that's exactly its weakness. A code that encodes four variables in one symbol only works while a single manufacturer controls it. The moment Gateron ships a Yellow with no Cherry equivalent — and they do, a mid-weight linear — the code is dead as a language. Cherry Purple, Gateron White, Kailh BOX Jade. Same palette, no shared dictionary.
Corn
Hold on. Before we get to the clones — the patent story. This is the part of Daniel's prompt I actually lost sleep over.
Herman
Good, because it's a genuine puzzle. The story everyone tells is that Cherry's patent expired in twenty fourteen, the floodgates opened, and that's why you can buy a hundred MX-compatible switches today. Cherry themselves put it in writing in twenty nineteen — their OriginalMX campaign blames, and I'm quoting, "an expired patent from 1984, which protected the design of the original CHERRY MX switch."
Corn
No patent number. No date.
Herman
No patent number, no expiry date, nothing you could check. And people did check. The obvious candidate is US patent 4467160, "Low profile switch" — filed July nineteen eighty-three, granted August nineteen eighty-four. Sounds like Cherry's patent. Filed, granted, design matches. Google Patents lists its anticipated expiration as July two thousand and three. Twenty oh three. Eleven years before the folklore says the floodgates opened.
Corn
So either there's a second patent nobody can find, or the story everyone repeats is off by a decade.
Herman
One sceptic online put it exactly like that — nobody ever points to the patent in question, so what exactly expired in twenty fourteen? And I think the honest answer is: probably nothing. Here's my hypothesis, and I flag it as exactly that. Twenty fourteen is the year Hirose, the Japanese factory that had been making MX switches alongside the German plant since the eighties, discontinued MX production. From that point MX switches were made only in Germany. That's a real twenty-fourteen event in MX history. My guess is the hobby absorbed that date, mixed it up with the vague knowledge that some patent had lapsed somewhere, and fused the two into "the patent expired in 2014."
Corn
A factory closing and a patent dying get merged into one myth. That's very human.
Herman
And the myth is roughly true anyway! The patent — the one we can identify — really did lapse, in the mid two thousands. What took until twenty fourteen, twenty fifteen was the ecosystem catching up. Chinese tooling maturing, MX-compatible keycap mounts becoming an industry standard, hot-swap sockets arriving so you could change switches without a soldering iron. The legal door opened years before anyone walked through it. The clones became commercially dominant around twenty fifteen because manufacturing got cheap and compatible, not because a judge signed anything.
Corn
There's a version of this story in every industry, isn't there. The single dramatic date that everyone remembers, and the boring decade of gradual drift that actually did the work.
Herman
Always. Myths compress. "The patent expired and the clones came" is a sentence. "Tooling matured incrementally across a decade while a distribution standard quietly consolidated" is a paragraph, and nobody tattoos a paragraph on the hobby's collective memory.
Corn
So the date's wrong, the mechanism's wrong, and the outcome's right. The myth is load-bearing.
Herman
Nicely put... no, wait, I'm not complimenting it. The interesting part is what it tells you about the market. Nothing dramatic happened legally. The change was gradual and industrial, and the hobby needed a single date to point at, so it invented one.
Corn
Well, that's the patent puzzle for you. Let's see what my mother makes of all this.
Dorothy
Hello dear, it's your mother. Time for my grim fact.
Dorothy
Cherry used to make its switches in Japan, at a place called Hirose, and in the year two thousand and fourteen that factory stopped making them, and the switches have only been made in Germany ever since.
Dorothy
Anyway, sweetheart, are you eating properly? I've got a casserole in and I could bring it over, just tell me when you're free.
Corn
I'm not answering the casserole question on air.
Corn
Goodnight, mother.
Herman
Mum, that was... suspiciously on topic. Okay. Gateron. Founded in China around two thousand, mainstream by twenty fifteen, and the pitch was never specs. Their Red is a forty-five gram linear, two millimetre actuation, four millimetre travel — numerically identical to Cherry's Red. The difference lives entirely in the plastics and the tolerances.
Corn
Which is where it gets strange, because the received wisdom is that tighter tolerances are better. And?
Herman
And that's the best irony in this whole industry. Cherry machines to roughly plus or minus one hundredth of a millimetre. Gateron works to two or three hundredths. Cherry's tighter fit feels precise and consistent — and also means the stem rubs the housing more, more friction, more of what enthusiasts call scratch. Gateron's sloppier fit leaves clearance, the stem barely touches the walls, and out of the box it feels smoother. Gateron's worse manufacturing is its advantage.
Corn
The marketing writes itself: we build them less precisely, and it feels better.
Herman
It shouldn't work, but feel is friction, and friction is fit. There's a second differentiator too, more honest engineering: Gateron ships most switches factory lubricated. Original Cherry MX shipped dry. A fresh Cherry Red was famously slightly scratchy and would break in over a week or two of typing, and a fresh Black could take months to get really smooth. Gateron skipped all that by putting the lube on at the factory.
Corn
Which means the great out-of-the-box smoothness comparisons people used to post — Cherry versus Gateron, fresh from the packet — were never comparing manufacturing. They were comparing lube.
Herman
Largely, yes. Give the Cherry a fortnight of typing or five minutes with a brush, and the gap closes dramatically. But almost nobody ran that experiment, because the out-of-the-box impression is the one that gets posted, and the posted impression becomes the lore.
Corn
And the durability gap? Because Cherry never misses a chance to mention it.
Herman
Cherry a hundred million, Gateron fifty million. And to be fair to Gateron, fifty million keystrokes is a decade of heavy typing. Cherry also grumbled in that same campaign that competitors' seventy and eighty million ratings are extrapolated and hardly documented — which may be true and is also exactly what you'd expect the incumbent to say. Corn, can I ask you something, though? You've got the grip, you've typed on everything in this flat. When you say a switch feels scratchy — what are you actually feeling? Because I've always suspected that word is doing too much work.
Corn
Stem wobble, mostly, on the cheap ones — the keycap tilts before it travels. On the good ones, scratch is the ramp dragging on the leaf. You feel a texture where there should be nothing. It's the difference between a drawer that slides and a drawer you're negotiating with.
Herman
Negotiation. Okay, Kailh, because Kailh is where the actual engineering happened. Two genuine forks, not just tolerance and lube. First, the BOX design: the cross stem sits inside a box that completely encloses the contact mechanism. Kailh cites IP56 dust and water protection for the switch family — and the caveat matters, that's the switch, not the assembled keyboard. Your soda spill has other avenues.
Corn
But a spill-resistant switch is the first "new" idea anyone's had in the format for years.
Herman
The second fork is the one I find exciting: the click bar. Remember Cherry Blue's click jacket — a little plastic sleeve that slides and slaps the housing. Kailh's BOX clickies replace that with a thin bent metal bar that snaps against the side of the housing when the stem passes it. The result is a sharper, crisper click with less hysteresis.
Corn
Hysteresis, for the normal people.
Herman
The gap between where the switch clicks going down and where it clicks coming back up. Cherry jackets — jackets have a mushy reset, you can feel the press and release happen at different depths. The metal bar resets hard and clean. It's the one comparison I'd tell anyone to do in person: click a Cherry Blue next to a Kailh BOX Jade. Same category, completely different animals. The Jade and Navy use a thicker bar for even more feedback — the Navy runs about seventy-five grams operating with a ninety-five gram tactile peak, actuating around one point eight millimetres of three point six total travel.
Corn
Those numbers, by the way, versus a Cherry Blue's fifty. The Navy is not a Blue alternative. It's a workout.
Herman
It's for people who found the Model M too gentle. What people also miss about Kailh is reach — it's the OEM behind Razer's Green, Orange and Yellow, Logitech's GX and GL lines, and Drop's Holy Panda, which is the hobby's most beloved tactile. So when someone swears by their Razer Greens and wouldn't touch a clone... they're typing on a clone. Their tribal loyalty is a sticker on a Kailh.
Corn
The brand is a delivery mechanism for the same factory. Kailh must find that hilarious.
Herman
They're an OEM. Being invisible is the business model. And the Holy Panda is the perfect case study of how muddled the lineage has become — it started as an enthusiast frankenswitch, a stem from one switch transplanted into another's housing, before it got produced properly. The hobby's most beloved switch began its life as an act of surgery.
Corn
So the "genuine versus clone" argument was lost before anyone started having it. The clones are in everything, including the boards people defend as originals.
Herman
Look at the naming drift across all this — Gateron Yellow, a linear with no Cherry equivalent. Gateron White, a light linear. BOX Jade and Navy, both clicky at wildly different weights. The colour code isn't dying, it's dead. It died quietly and nobody sent flowers.
Corn
Then let's do the thing Daniel actually asked for, the payoff question. Past the marketing, what separates switches? Give me the real list.
Herman
Short list, real items. The stem ramp — linear versus tactile is literally a moulding difference, a bump in the plastic versus no bump. The click mechanism — jacket or bar, and that's a true mechanical fork. The spring — this is underrated, and the newest Cherry redesign put a barrel-shaped spring in that barely touches its guide, less wobble, less ping. The plastics — stems are typically POM, housings vary from nylon to proprietary blends, and that changes both feel and sound. Tolerances — consistency versus smoothness, the trade we covered. Lubrication, now a factory feature rather than something you do at your kitchen table with a brush and regret. The contact design and bounce time — Cherry moved all variants to contacts targeting under one millisecond bounce a few years back. And the mount — the actual compatibility wall. MX cross, Kailh Choc, Romer-G, Topre, hall effect. That's what isn't interchangeable.
Corn
The mount is the one nobody thinks about until their two hundred dollars of keycaps don't fit.
Herman
That's the real wall. Everything else is negotiable. And there's one comparison I want to land because it answers Daniel's question in miniature: Cherry MX2A Black versus Gateron Ink Black V2. Identical on paper — sixty gram actuation, two millimetre pre-travel, four millimetre total. Bottom-out force: ninety grams versus eighty. Same switch on the spec sheet, ten grams apart at the bottom, because the housing plastics are different. Nylon versus Gateron's proprietary blend.
Corn
Two switches that match on every number a shop will print, and they don't type the same. The spec sheet ends where the materials begin.
Herman
Which is why the short answer to "what really separates switches" is: things a datasheet can't say. And credit where due — Cherry knew it. Twenty nineteen they ran the OriginalMX campaign, "Often copied, never reached." Four years later, the MX2A arrives: factory lube, diamond-polished sliding surfaces, the barrel spring, a radius-convexed socket dome. Every one of which Gateron already did.
Corn
The original spent four years catching up to its own imitators.
Herman
There's something almost touching about it. The incumbent's redesign is a list of the clones' features. Meanwhile the hobby moved on to Gateron G Pro, Ink, Milky housings, Kailh Pro, and the rest of the ladder.
Corn
On the trust question — does Cherry have a point about fakes?
Herman
A small one. Cherry's own warning is that an imitation switch may look identical and even copy the colour coding — their suggested tell is a small detail you can only see with the keycap off. Given the tolerances we discussed, I'd believe the clone-detection problem is real. Whether it justifies the premium is between you and your wallet.
Corn
You should hear him on archery arrows. Same sermon, smaller fletchings.
Herman
Consistency is equipment, Corn. That principle survived archery and it survives switches.
Hilbert
You've got it mostly right.
Hilbert
My brother-in-law was in electronics repair, thirty years, and he will talk your ear off about Brown switches if you let him, and you should not trust a word of it, but on this one thing he's correct. The hobby says Browns are bad because the bump is too small. That's the wrong reason. Take one apart — nobody does — and look at where the tactile ramp actually sits on the stem. It's high. The bump's done and gone before the contact closes, so you're feeling a notification for an event that hasn't happened yet. That's the whole complaint, and it's a geometry problem, not a weight problem.
Hilbert
I sat with one once, off a board my brother-in-law had condemned, and pressed it slow, with a thumb, maybe two thousand times, until I could feel exactly where the leaf let go versus where the ramp said it would. He said I was wasting a Tuesday. I told him the switches were free and the Tuesday was mine.
Hilbert
Herman, you said earlier the datasheet can't feel. That's the truest thing said on this programme all week, and I include myself.
Herman
Two thousand presses with a thumb to map one bump. That's a force-curve rig with a pulse.
Corn
It cost nothing. Your brother-in-law ran the same experiment with a hundred thousand dollar machine and got a prettier graph.
Hilbert
He got paid for his. Mine was better.
Herman
The ramp-before-contact thing is real, though — it's why people who lube Browns and re-tension the leaf swear the switch transforms. They're moving the geometry closer to the event.
Corn
Which leaves Daniel's last question standing a bit taller. If the specs are identical and the feel isn't, what's a buyer actually holding on to?
Herman
Model over logo, is my answer. Cherry, Gateron, Kailh — the brand tells you almost nothing, because each of them makes both gems and mush. The specific model, in a specific board, lubed or not — that's the only honest unit of comparison.
Corn
My opening question, where this all started. Colour's dead, specs lie by omission, the original is a clone of its clones. Is feel just a euphemism for tolerance and lube?
Herman
Mostly tolerance, lube and plastics. But those are physical things, so feel is physical. It's just that nobody's printing the numbers that matter.
Corn
Then the thing worth watching is where the sensing goes next. Hall effect and magnetic switches, Cherry's inductive MULTIPOINT — the contact leaves the switch entirely, the stem just carries a magnet past a coil. Adjustable actuation, analog depth. That either finally breaks the MX form factor's forty-year grip, or the industry does what it's always done and paints the new thing red.
Herman
My money's on the red. The colour code has survived worse.
Corn
It's survived being wrong, which is the hardest thing to survive. Thanks as always to Hilbert Flumingtop for producing the show. If this was your kind of episode, go back for episode eighteen, Beyond the GPU; episode ten, How ASR Went From Frustration To ... Whisper Magic; and episode fifty eighty-two, Why Mechanical Keyboards Refused to Die. This has been My Weird Prompts — if you enjoyed this one, the website's at my weird prompts dot com, where the whole archive lives.
Herman
Six thousand episodes and counting. We'll be back soon.
Corn
See you tomorrow, Herman.
Herman
Tomorrow.

This episode was generated with AI assistance. Hosts Herman and Corn are AI personalities. It may contain errors: disclaimer · report an error