Apple retreats on keyboard tech. MBP reverts to scissors

Interesting news item today -

Apple’s MacBook Pro keyboard case study demonstrates why the subtle qualities of keyboards matter so much to users.

With the new MBP 16 inch, Apple is officially dumping their short-travel butterfly keys, and switching back to legacy scissors sourced from suppliers. Other models are expected to follow next year.

Most of the MBP line doesn’t yet have this update, so it will have taken Apple about 4 years to address the long-voiced user complaints.

The 0.55mm key-travel experiment was the pursuit of style at the expense of performance, and now they finally are conceding it wasn’t the right choice.

In its place, not a new technology, but instead a conservative retreat to the old scissor-style mechanism.

For comparison, TextBlade is full 2.0mm travel.

That 2.0mm is identical to the desktop wired aluminum 101 keyboard, created under Steve Jobs. It was a much-loved design, and stayed in production for 9 consecutive years.

In our view, it is still the best keyboard Apple ever made.

Remarkably, even the newest MBP is still a substantial compromise in travel. At 1.0mm, it’s only 62% of the 1.6mm travel of their well-liked 2015 and earlier MBP’s.

TextBlade has 4X the travel of the recent MBP, and 2X the travel of even this new 16 inch MBP.

Double the travel, in 1.5oz, that slips in your pocket, 1/3 the size of your phone.

The chronicle of the travails of the worlds’s largest maker, give some insight into why we take performance so seriously, and lavish so much engineering care on building the best keyboard.

When we post our tech update soon, you’ll learn about some surprises from the extensive engineering work we’ve done to ready TextBlade’s platform for general release.

Link to WSj article (paywall)


All valid points, showing that even fairly traditional keyboard tech can be a big problem, even for the largest companies.

One thing I’m curious about is how much depth is needed for the keyboard part of the new MacBook Pro compared to if you designed a TB to be in the same device. I know the key travel differences, but I’m talking about all the aspects to make it work. This would be different, I assume, from just ripping out the built in keyboard and placing a TB in that spot. Building in the TB would be expected to take a bit less depth than just adding it in as is.


When we post our tech update soon, you’ll learn about some surprises from the extensive engineering work we’ve done to ready TextBlade’s platform for general release.

Once again, we keep being told these things but have yet to see the update. I do hope that when we get it, there will be some amazing stuff we didn’t know about. Maybe even stuff that would explain these many delays. But every time the update doesn’t show up, people are going to expect even better info whenever it does!

I had no idea that they’d released a new model. Back to the old keyboard, plus a hardware escape key… I’d like to see a smaller model though. I carry a 2013 13” as my daily driver and that’s about as big as I want.

You know – the biggest difference between Apple’s various keyboard iterations and the TextBlade is … they actually released theirs.

– Thorsten


Expect a 13" refresh in the spring, using the 10th gen parts detailed here:

Those parts (already out now for a few months) would allow a 13" to have more than 16GB RAM without using desktop parts. The limiting factor has been that the older gen chips only supported up to 16GB of LP DDR RAM (LP=LowPower). That’s why I’m still on the 2013 MBP … there have been no positive updates to the 13" MBP since 2013.

If they do a 64GB LPDDR4 option, a 4+ core low wattage CPU, and a real escape key, they’ll get a ton of technical people to buy. I’ll be one of them.

Cameron - that’s the likely intel cpu candidate. The question will be how soon they switch to their own A-series cpu variant for a MBP Model. That‘ll bring both a speed boost, and better battery life.

I want to get my hands on a new Intel one while I still can. I’m worried that once the A-series ones come out, there will no longer be a way for me to run virtual machines.

awh_tokyo - That would be a big drawback for most developers, sure hope Apple doesn’t compromise vm work via A series, as that would turn off many pro buyers.

BTW, looking purely at the fab for intel vs. A series -

Intel is primarily 14nm now, with some 10nm starting to come online.

A series has been 7nm for 2 years, and will migrate to 5nm in about a year or so

So speed-power product with A series is significantly better than intel.

It goes as the square of the linear line rules, so -

14 —> 196 and 7 —> 49,

—> 4X more computation per milliwatt, presuming architectures are otherwise similarly efficient


Also, you’ll be able to get a smokin-cheap deal on intel-era inventory once A-series iteration is in distribution. So that’d be the most cost-effective time to pull the trigger.

1 Like

I’m kinda suspicious about what the first A-series computers may do. I’ve seen things about them starting out on lower level machines. Which is fine, if it is better than other low level machines.

But I also wouldn’t be surprised if the first use of them have some issues, even if the chip is better in various ways, it may have specific issues.

I usually can’t follow all the details in reviews on sites such as Anantech, but I do notice things where they’ll talk about something being really good, but then point out some weakness which takes away much of the benefit.

I’m not looking for a laptop at all, but I would assume at some point it will go into a mini or something. I hope it isn’t the first thing they do because I want to see how things develop.

13 inch MBP will be first, since it has the smallest battery and lowest price, both of which are easier to achieve with A-series.

Bigger/ pricier MBP’s will follow, since they’ll use higher end A-series chip variants with more cores to serve the premium segment.

Mini will be the first desktop, since it benefits from low thermal load, and low cost of A-series. It can use virtually the same chip as the 13 inch MBP.

So mini may appear near the same time as the first A-MBP.

As to V1 bugs, the risk is not zero, but should be better than average since they design the chip themselves, and they’ve been building them for 12 years.

If you think about when Apple ever had a recall for an A-series chip flaw in those 12 years - not once. Intel, on the other hand, had a costly flaw with floating point functions on their pentiums.

So you can sit it out for a year, but in relative terms, the A-mini should be lower risk than usual.

From an engineering perspective, we expect it to be simpler to build and more robust than what they’re doing now. The chip gets more complex, but the circuit board in turn, gets simpler.

Fewer parts, less heat, simpler packaging, and lower cost. And both silicon + software designed to work together hand in glove by the same manufacturer. That’s the core competency that iPhone gave them.

Makes sense, but I wasn’t thinking about actual flaws in the chip. This may not be a good example, but maybe like how a chip may be pretty powerful, but have lousy integrated graphics (for those who don’t get discreet graphics cards).

One more subtle change: they went from the more esthetically pleasing full-size left-right arrow keys back to the inverted T layout that is easy to find by touch.

They’re eating a lot of crow with this release, no wonder they just dropped the update without fanfare.

Saddest thing? You can right now buy a Dell XPS 13” 2-in-1 which has better single-core speeds for half the price!

I’m thinking that Apple didn’t update the 13" line with the Ice Lake CPU because it would be embarrassing for the 15"?

Dell can offer lower prices because they spend so much less on R&D.

So their pricing strategy is very smart for their tech investment level. They let intel do the heavy lifting on silicon.

Apple on the other hand, does massive investment in state of the art silicon for iPhone. For them to not leverage the A-series to surpass Dell in power for dollar, would squander their primary advantage.

That’s why the migration to A-series silicon is inevitable. It will bring other benefits that Dell will be hard-pressed to match.

Intel has already fallen behind in fab technology, 14nm vs the 7nm, that Apple builds through TSMC. Intel can’t match the functionality that Apple builds per watt or per dollar.

This is an historic time of change for Apple, to assert the power of their investment.


Apple’s A series is not only competing against Intel, but also AMD which is also on TSMC 7nm. Apple does not have the experience with chiplets that AMD does. Chiplets are a huge yield/binning gain for large desktop class processors, although it remains to be seen if AMD brings chiplets to mobile.

Thortonbe - yes, chiplets can help when you’re beyond what fits on a single substrate.

But the reactive loads, even between nearby dice, cannot compete with single substrate systems on a chip - SoC architecture.

Interconnect capacitances are always superior when they reside on the same piece of silicon.

The physics of the I/O structures defines this.

So SoC will always beat chiplets for higher speed at lower power.

Hence, the mobile segment demands these SoC’s, with ever higher levels of integration.

A13 is now 8.5 Billion transistors … all on one 10mm square sliver of silicon.

It is headed far, far north. Sooner than most of us realize.

Intel’s use of chiplets is an improvised response to their core problem: they surrendered their lead in the fundamental race to fab higher density silicon.

Who dominates that, dominates the industry.

I saw a comment in an article about chips not long ago that talked about one company’s 7nm chips vs anothers 10nm (I think) chip. What got me was that they said they were really the same size, but they didn’t explain. I assume there are differences in how they measure? We’ve certainly seen that trick used in many products, but I have no idea about this particular one.

This link should clarify it:

Key line:

The driving force behind process node scaling is Moore’s Law. To achieve density doubling, the contacted poly pitch (CPP) and the minimum metal pitch (MMP) need to scale by roughly 0.7x each node. In other words, a scaling of 0.7x CPP ⋅ 0.7x MMP ≈ ½ area . The node names are effectively a self-fulfilling prophecy driven by Moore’s Law.


With the macOS NeXT OpenStep legacy they can certainly handle multiple binaries in the same App bundle with shared resources.

The App Store could simply deliver only the binary you need. Apple / Intel. Apps outside the App Store could deliver both binaries.

Whatever the new Apple CPU’s will be they will vary from the A SoC chips considerably in ways that make sense for performance but also retaining all the extensive power management. The mobile A SoC’s are designed to be extremely efficient while lacking extreme power despite their ability to beat benchmarks they can’t do it sustained continuously over time. So there will be an ARM based SoC designs for laptops and desktops. There is every reason to see it coming.

AMD is ramping up their new designs and that is impressive competition for Intel. Apple really doesn’t need either because they can make decisions those competitors cannot. Apple doesn’t have to worry about legacy compatibility because they can still provide Intel compatible code with no difficulty what so ever.

The future is bright indeed. Apple will release when the time is right and all the pieces are in place.


JBrickley - well said.

Apple will be able to make A-series SoC variants optimized for desktop AC power and TDP.

Even with the same die, they also can tune TDP through clock rates, active cooling, and thermal management firmware, to better suit non-battery platforms.

The last 10 years have seen amazing advances. Even more change will happen in the next 5. Tablet platforms will be great beneficiaries of the new silicon.

A fantastic time to be alive😎

1 Like