Like many, I have the feeling that this year’s WWDC is going to be a particularly meaty one. But frankly, of all the things Apple’s doing now, what concerns me the most is the rumoured Mac transition from Intel to ARM chips.

Looking back on my very blog, I’ve realised that it’s been at least two years since the tech world has been musing on it. But in 2018 it was mostly a ‘what if’ scenario. Today we know that something is really about to happen, one way or another.

Since the publication of Mark Gurman’s article, I’ve been reading a fair amount of commentary about this next transition. Most of the more balanced takes seem to share some optimism about how Apple will handle the transition. Since the company has already handled two chip architecture transitions in the past (Motorola 68K CISC → PowerPC RISC in 1994, and PowerPC → Intel in 2005–2006) — and things went rather smoothly in both cases — they guess that this third transition will play out just as smoothly. 

I don’t entirely share this confidence. The past two transitions were handled by two very different Apples, and this next one will be handled by yet another different Apple. Same company, but different times, different people, different leadership, different priorities.

Back in 2018, in Speculation and dread for the next transition, I wrote:

All these major transitions [In the article I also included the major operating system software transition, from Mac OS 9 to Mac OS X, that took place in 2001–2002] have common characteristics:

• They were all rather user-friendly and customer-friendly.
• They weren’t particularly rushed: there was both preparation and confidence on Apple’s part, and they unfolded over a long period of time and at an acceptable pace. Users had to update eventually, but they were given plenty of time to do so.
• All these transitions were for the better. […]

Like others, I’m sure that the ARM architecture, coupled with Apple-designed custom chips, will certainly benefit the Mac from a performance/consumption standpoint. But what still makes me apprehensive about the whole thing is how Apple — this Apple — will handle the software transition.

The previous transition, from PowerPC to Intel chips, gave users an inordinately long grace period when it comes to software and backward compatibility. Shortly after Jobs announced Apple’s plans at WWDC 2005, many developers started converting their PowerPC-only apps into Universal Binaries (apps that could run on both architectures). And for all those older PowerPC apps that were not converted into Universal Binaries or recompiled to run exclusively on Intel Macs, a dynamic binary translator called Rosetta — included in the Intel versions of Mac OS X Tiger and Leopard, and optionally available under Snow Leopard — allowed people to run PowerPC apps at almost native speeds. If you consider that the last minor release of Snow Leopard (10.6.8) was released in June 2011, this means that you could still run a PowerPC app on an Intel Mac as late as five years after the transition was complete, hardware-wise.

For this next transition, I really don’t expect such a generous grace period. When it comes to Mac OS, in recent years Apple’s attitude has been like, Let’s try and get rid of whatever we don’t have the time or the manpower to fix. To be fair, Apple’s willingness to drop as much baggage as possible whenever possible has always been one of its most characteristic traits for the past 20 years, but Cook’s Apple seems particularly interested in doing so considering just how many platforms they’ve chosen to juggle.

From what I understood, on a technical level, there’s this:

• Recompiling an Intel app to work under the ARM architecture should be a relatively easier task (with exceptions) compared to what was recompiling a PowerPC app to work under the Intel architecture 15 years ago. Given the appropriate tools, of course.
• Emulating an Intel app so that it runs in an ARM-based Mac, while possible, doesn’t offer adequate or usable performance. One should not expect Rosetta-like results, in other words.

Some have speculated that Apple, to ease the transition, could initially ship Macs that contain both an ARM processor and an Intel processor, the latter taking care of running x86 code at native speeds. Sort of a hardware Rosetta, if you like.

The scenario I’m fearing, however, sees a more pragmatic stance on Apple’s part, and it would play out like this:

• Apple announces their transition plans at WWDC 2020, previewing the tools the company will make available to developers so that they can recompile/rework their apps to run under the new architecture.
• Apple will offer new ARM-based Macs gradually, so that if you still need to run Intel-based apps that are either not updated anymore or whose redevelopment needs time, you can keep using your Intel Macs to run them; and if you still need such apps after the hardware transition is complete, well, you better keep that old Intel Mac close to your chest, because ARM-based Macs will only run ARM apps.

In other words I think that Apple, once the wheels of this next transition are set in motion, will do the bare minimum to make this transition smooth for developers or end users. The music will be: Developers, here’s what’s new. Get to work, the sooner the better.

If this happens, it’ll certainly result in even more pruning of all the software that, for a reason or another, won’t be ported to the ARM architecture. And this, after the already bitter pill of Mac OS Catalina dropping support of 32-bit apps (a vast catalogue of perfectly good software), will be another hard pill to swallow. 

The excuse will be that ARM Macs are going to be more efficient, more powerful, more secure, and with a fresh catalogue of optimised apps to run under the ARM architecture. Apps notarised by Apple, with the company’s seal of approval. Everyone wins! Well, everyone except those who would like for the Mac to keep being a versatile and ‘open’ platform, where you can install apps developed by anyone if you want; where you can use Boot Camp to reboot your Mac into Windows if you want (or — gasp — need to); where you can maybe enjoy all those nice vintage 32-bit Intel games every now and then.

Apple can put in place a lot of solutions for developers and users to ease the hassle of this new transition, and I really really hope they’ll be gracious enough to do so. But when we start looking at what can possibly be the incentive to do so, I can only think of “To avoid further alienating Mac developers and users”, and I can’t help thinking that Apple is quite ready to take such risk. 

This is, in a nutshell, the main reason of my current apprehension. I’m waiting for Monday with trepidation, hoping to be wrong — or to have been too pessimistic — about this.