Later posts →

Quinn Nelson: Using Apple’s Tools to Fix My iPhone

Quinn Nelson, who used to own an iPhone repair shop, takes the US-only iPhone Self Service Repair for a spin. They ship you 40+ kg of equipment, which you rent for 7 days. The entire experience is a confounding mix of thoughtful little touches and issues being solved with a ridiculous, over-the-top sledgehammer approach, more or less because Apple can afford it.

The iPhone as a product is designed, engineered and developed with a number of constraints on materials, fit and finish. It has to be manufacturable in the first place, it has to be torsionally rigid, it has to withstand atmospheric pressure, a person's grip and perspiration. The materials have to be responsibly and sustainably sourced, free from impurities and forbidden substances. It has to fit together well and look its best.

There are thousands of check boxes to tick, all of which have guided the design and build process. For a company capable of doing all this, only laziness, apathy or spite are stopping them from making it just as well put-together, just as representative of a deliberate design ethos, but more easily repairable.

Only laziness, apathy or spite.


Over and over again, I hear the same argument. Someone (Apple, Google, Microsoft, Facebook, whoever) created a thing, and they are now and forever in charge of that thing, and can do whatever they want, and if people don't like it, tough.

This argument is a fair place to start and holds water to a point. If you make a sandwich in the comfort of your own home, you can put whatever you want on it. If you make a sandwich for your kid, your kid can't insist that it have jell-o on it instead of sprouts and cream cheese. (Although if you put dirt on it, social services may want to have a word.)

However, if you have somehow transformed into the civilized world's maker, keeper and provider of sandwiches, it is ridiculous to apply the same logic as if you're just making an afternoon snack for yourself. If applications run on your sandwich that, if they went down, would seriously impact society, we have graduated from the point where it's just your concern (and beyond this metaphor).

iOS and Android is a duopoly. It's not because it's technically impossible to create an alternate platform - although it is very difficult. It's because of the powerful network effects holding back any contender. At that point, your platform controls eating and breathing; the air, the soil and the water; livelihood, where the savings are deposited, how they are used. At that point, everything else is a joke. At that point, you made yourself the world's platform.

I am not one to say that the government, the military or the angry concerned mob of civilians should wrest control of these creations from the private companies, and that this will solve everything. But I am one to say that you keep acting in the same self-interested, self-serving, self-aggrandizing way as you did when you were one of ten hopeful entrants in a crowded market at your own peril.

Monopolies eventually fall. If you're unlucky, they pull all the oxygen from the room first, smothering all counter-agents, setting us back years. Making us forget what it was like before everyone took as given the compromises we were forced to make; the selfish, enshrined as security, as utility, leaving a shriveled, ashen field.

Monopolies are endothermic and parasitic, claiming for itself the value of their surroundings, under the guise of its own importance, propped up and fueled by those who wish for eternal youth, ultimate power and endless resources. But wishing so does not change how thermodynamics work, how ecosystems work. If not dethroned, all power, all value, all energy flows to the monopoly, which assumes it, which forgets everything else, the spectre of endless growth the everpresent hum of the universe. And when it's all there...

The story only ends in one place - without a monopoly. Having fallen, first gradually, then suddenly.

I am not worried about monopolies lasting forever. I am worried about what they will consume before they are toppled. And I am worried about the power of grown adults who still insist on jell-o sandwiches.

Noah Grey's Gofundme

I am late to this and only found out about this on Slashdot, but days ago, Noah Grey was in deep trouble.

Noah Grey is probably one of the most influential people to personal, automated Internet publishing that most people have never heard of. At the turn of the century he produced a remarkable piece of software called Greymatter, far before it would be described as a "static site generator", "content management system" or "weblog engine", and used it to publish beautiful photography. His sensibilities ran deeply through the software and from what I can tell, through everything he did.

The story of the current circumstances has a happy ending, at least in terms of account balances, but I took the opportunity to contribute anyway because of what people like Noah have meant to me personally and to the Internet that I love. Due to cynicism, commercialism, greed and politics, the Internet may not be what it was any longer, but that's hardly Noah's fault. It was stolen from him and from all of us, so let's steal it back.

Steve Troughton-Smith on Apple, App Store, developers and government meddling

Steve calls it "an incredibly-delicate, calculated 'truce' that relied on Apple not overstepping into abuse" and now thinks "Apple’s behavior has been so offensive over past few years that, at this point, all I’m thinking is ‘damn the consequences’".

I've never understood why people have been so bewitched about the positive potential outcomes. Purely by making the operating system, defining the APIs, bundling its own software, manufacturing its own hardware, tying the operating system to its hardware and vice versa and selling even more of its own software, Apple has plenty of opportunity to set good examples and draw people to the combined value proposition. Why the hell does it need any more control than that?

I have been against the App Store from day one on a philosophical standpoint but also from a practical, realistic standpoint. The ability for a developer to be a great developer is hampered, frustratingly and undeniably. The likelihood that a user will find a great app is lowered dramatically.

Mac OS has always had less apps than Windows, but they've also had great applications, sometimes good enough to keep people on the platform, earlier good enough to keep people enduring an unstable foundation. Applications developed by small teams of people who nevertheless made great things, in some cases still for longer than there's even been an App Store.

iOS and the App Store has not been a truce. It has been a destructive, abusive, lopsided, mistrustful relationship; a relationship that has allowed access to a platform advantageous enough to make you close your eyes and think of multi-touch. It has been an insult to history, an exercise in attempting to redefine away the fundamental facts of the market and of their existing user base and developer ecosystem in a puff of malevolent marketing.

I don't see governments as the best arbitrators of required features in hardware or software, but they sometimes have a strong connection to what's fair for the customer, and ain't nothing about the App Store that's fair.

Like Steve, I resent the muddling connection to "industry interests", whose involvement just serves to hide the real issues. I am reluctantly interested in what's going on as a forcing function, knowing the lopsided unfairness could simply slide over to yet another party beyond end user, developer and Apple. That could happen.

If that happens, I will blame the Apple that woke up every morning since 2008 and chose to worship Mammon. I will blame the Apple that took the easy way out and chose the clammy, desperate grip of control. I will blame the Apple cowardly enough and uninspired enough to keep to its own, rather than to lead in the sense of empowering other people.

I will blame the Apple that, through this mismanagement, craps on the legacy of the people and divisions within the company that have done groundbreaking, innovative and empowering work in their own areas over the years, all the while it reaps the profits of their advancements and discards their spent husks when they're done.

And I will blame the Apple that, when people naturally react to their mismanagement, acts surprised, as if we don't know that they know best, because didn't we know that pure intent scrubs away all consequences?

The word "gaslighting" is thrown around a lot these days and in its original form it seems to require a malicious intent to deceive. I really don't know what's there. I believe people at all ranks at Apple to be intelligent people and think very few of them have ill-intent, so I really can't describe what motivates what they're doing. But they seem to have a radically different idea of what's going on than literally everyone else. And so maybe the idea of the "Cult of Apple", long brought out to explain why people would even use those smelly Macs and iPhones, just had it backwards; it is the company that is the cult.

Matt Gemmell: Content Creation

Content is fungible, space-filling, placeholder-replacing stuff, and that’s not even its most offensive connotation. The worst part of “content” is that it’s implicitly relative to its container. That’s what content means. The contents of a sandwich are with respect to the bread. The content is just what’s inside. It’s probably the most trivialising, demeaning, depressing, surrendering term out there — but people lay claim to it. People define themselves as makers of it.

Bingo and hear, hear.

Ars Technica: Apple restores Navalny app to App Store

Apple has restored an app sponsored by Alexei Navalny, a prominent leader of Russia's political opposition, to the company's Russian app store. Apple took down the app last September, days before Russia's legislative elections, under pressure from the Russian government.

Ars's article tagline is "too little, too late", and yes, that much is true. I'm getting the sense Apple has changed its position on how it should operate App Stores in autocracies, or at least the ones where they don't make a good chunk of their money and nearly all of their products. Technology did not cause Russia's ills, nor will it cure them, but it can still make a difference for the Russian people.

Jeff Johnson: Mac Pro historical perspective

I would argue that the Mac Pro as we software developers knew it was never given a successor after the "trash can". The Mac Pro was discontinued and replaced with a different computer of the same name that was no longer for its largest pro audience. I don't know many individual software developers now who can afford a new Mac Pro. I certainly can't.


I don't understand why we have to make these painful tradeoffs in 2022 when they weren't necessary in 2012. Ten years ago we had relatively affordable, conveniently upgradable Mac Pro models. Since then we gained a faster CPU, but otherwise we've lost everything else great about the Mac Pro.

The Mac is a much bigger platform now than it was 10 to 20 years ago. There is room for both option A: a Mac Pro that costs insane amounts of money, is incredibly overbuilt and is just what the doctor ordered when being able to top out the configuration is more important than being sensible, in the way that might make sense (ie disappear into the budgetary static) on a movie set where it costs more to have people stand around for ten minutes; and option B: a Mac Pro that has a great processor and modularity in the rest-of-the-world sense of the word, in that you can make it what you need it to be, without Thunderbolt enclosures or docks, but that also does not require you to order the right amount of everything the first time or take out a mortgage.

Right now, despite the opportunity to do both, Apple just does option A, and that used to be a positive development, because the previous "trash can Mac Pro", option C, was even worse. Mac Studio is a better implementation of option C, so it remains to be seen whether WWDC will bring a better implementation of option A or the return of option B, or even both. A sufficiently layered or wide-ranging line-up could include both, although the existence of option B would likely embarrass the Mac Studio.



If we called the previous Mac Pros the trash can and the cheese grater, I'm calling this thing the Thicc Mini.


The Futurama episode A Clockwork Origin features a scene where a "creationist" orangutan asks for a mysterious missing link between two species, only to be disproven, and to repeat the same question with the link between the newly revealed species and one of the previous species.

This is roughly what being a Mac prosumer is like.

Two decades ago, there existed iMacs – all-in-ones with good but not great performance – and PowerMacs – with great performance and expandability for roughly $1500 and up. Aside from a pricing hike as the PowerMacs took on first IBM's heftier PowerPC G5 chips and then Intel's server platform Xeon chips, this lasted until the aforementioned "trash can" Mac Pro, which showed that Apple could innovate themselves into an overly specific and restrictive corner. When they extricated themselves out of it, it was with first an iMac Pro model at $4999 with Xeon W chips and then a new "y'all want a cheese grater? here's a cheese grater" Mac Pro, also with Xeon W chips, starting at $5999.

Ever since the PowerMac started drifting upwards, and to a lesser degree before it did, people have clamored for a mythical xMac model, which would have some limited expandability but also be a competent workstation. This was also driven by fear of the product line being usurped by the all-in-one design philosophy powering not just every laptop, but the iMac as well, which aside from the entry-level Mac mini was the only affordable desktop.

So here comes the Mac Studio, with either an M1 Max or M1 Ultra SoC and a more approachable price tag than the Mac Pro, alongside the return of the Studio Display, enough to toss the geriatric (CPU-wise) Intel iMac 27" model out of the product line. Done deal; let's go home. Right?

Not so fast. The Studio Display looks like a good, well-built product, but comparing to products that Apple has itself been selling for years, it's not that impressive. It's a display for the people who would have liked to have an Apple Silicon iMac 27", but it's not impressive and ground-breaking as displays go, and the relatively low refresh rate and brightness in a display that, unlike the Pro Display XDR which is positioned as an everyday reference monitor, competes with displays that have both high refresh rates and HDR capability. But, luckily, you can pick whatever display you'd like.

As for the Mac Studio, it's ostensibly in the classic PowerMac range of around $2000. But it's not expandable in the slightest. If you want to have a decent amount of internal storage, more unified memory or the M1 Ultra, prices quickly race skyward. Or you have to expand with external accessories rather than additional components, which is always more expensive.

So, it's good news that a developer can get a good Mac workstation desktop without having to pay for a screen you're not going to use, for a roughly comparable price as what we have been clamoring for, and without totally giving up on decent I/O connectivity.

But: you get a computer that's done when you unpack it; a PowerMac G4 Cube, except even less expandable. It would be lovely to be able to add additional internal storage — even if Apple hates the form factor, even just having M.2 NVMe drives would go a long way. It would also be good to be able to add some form of additional memory or cards. The idea of the xMac certainly involved more additional expandability. Not to mention that the inability to connect a high refresh rate display limits future proofing even more.


Going from xMac to X factor, we have the last piece of the puzzle, the upcoming Apple Silicon Mac Pro, the presumptive star of this year's WWDC keynote. Given the accuracy of Mark Gurman's reporting on the M1 series of chips, this would be the final, "Jade 4C-Die", M1 Extreme, or whichever is more than Ultra, which is already more "max" than "Max". It would be four M1 Max dies with interconnects, scaled up another doubling (although maybe not in every dimension and every sense). This would give 128 to 256 GB of unified memory, and if Apple chooses to use the same marketing tack for it as for the MacBook Pro and Mac Studio, it would go hard on how completely untenable it would be to get that amount of GPU memory, which is true.

What's left to answer is: will that angle be enough for them to forgo additional PCIe slot expansion, and to avoid additional, DIMM-based module RAM? The M1 Extreme would not fit in the Mac Studio's thermal envelope, so there would need to be a new enclosure and probably a new industrial design either way. But if it's not notably different from the Mac Studio in character, what would make it more than just a "Mac Studio Pro"?

Here are my guesses, based on the Mac Studio, based on steps and investments Apple has taken in both hardware and software in the past few years:

  • M1 Ultra or M1 Extreme (possibly a low-end model with M1 Max).
  • A handful of PCIe slots.
  • Some way of facilitating extra storage that is possibly as ham-handed as the current Mac Pro storage enclosures, but that also allows NVMe SSDs. (The current enclosures only fit hard drives in various configurations, and that feels very out of tune with the focus on high performance not just on the Mac Pro but on the upper models of the M1.)
  • Additional memory in a way that we haven't seen before, quite possibly a conceptual mix between CXL and the UltraFusion chip interconnect that forgoes having DIMMs as we know them – or just additional chip models with even more memory than 256 GB.
  • The MPX module is unclear and could go either way. It is an extension of the PCIe slot to facilitate more power, and seems wasteful and short-sighted to have been invented only for one generation of product and a handful of modules ever, but then again, it has basically mostly been used for GPUs, and additional GPUs are more doubtful (and are still not supported in external enclosures).

No matter how this all shakes out, there's basically no chance that the Mac Pro is going to be significantly cheaper than the current Mac Pro model - maybe if there's a low-end M1 Max model and the form factor embraces expandability, so that it is significantly different from the Mac Studio M1 Ultra model that it will cost about the same as.

So, lest we forget the point: the xMac does not appear likely to be incarnated any time soon, and the PowerMac, in the sense of a reasonably inexpensive (read: not $4999) workstation with moderate internal modularity, is not due to be reincarnated either by the Mac Studio or the Mac Pro. The virtues of such machines are apparently not compatible with the virtues of the new Apple Silicon, the point of which is chiefly how the one thing with the really quite many transistors can do all these things without needing any other things. As neat as those chips are, they do not fully make up for what was once there and now does not exist any longer.

The Scourge of Scrolling

The web and mobile UI have established a new scene for drastic, dramatic and minimalist user interfaces. Some interfaces benefit from it, and some interfaces have their sins (intrusive branding, clutter or ads) magnified. But both web and mobile UI are heavy on scrolling. Mobile because there's only so much you can fit on screen, and you can't make things too small before they are too hard to tap; web because the web was made for documents and long texts, and that's how you read them.

Sometimes, a scrolling list is just what you need. And sometimes you need a scrolling list of options or alternatives or items within your user interface. But if you have the option, your user interface should be fixed. It should be laid out, with things in specific places for particular reasons. Grouped by subject or rough order of operations. Separated by lines or into tabs, to present a digestable amount of options that each connect or relate to each other. And yes, adaptive to screen orientation, font sizes and bright/dark modes, by all means.

The alternative is pouring every option, every user interface control out onto the screen, a flow-layout algorithm the only guide, making the primary navigational aid scrolling-and-looking, scrolling-and-looking, and hoping you find whatever you're looking for. The alternative is abusing a visual language that suffices for digesting words, but in the service of making sense of all of them together. Every check box, every option, every button is its own idea, and the human mind needs all the help it can get to put them in context.

If you can always scroll, ask yourself if you could arrange your user interface with more intent. Respect your user interface, respect your user, and avoid letting necessary compromises in constrained environments bleed into places where you have the real estate and dynamic range to do a much better job. Your user interface is a thing, and things should be what they are – intentional, predictable and dependable.

Exclusive: Leaked Draft of Contingency Sideloading Guidelines

Take can today reveal a partial draft of developer guidelines aimed at qualifying applications distributed via sideloading, designed as a contigency plan if events force Apple to open up application distribution.

1. When distributing applications outside the App Store, applications must bemoan the regrettable circumstances that led to their existence outside of the calm, gentle embrace of the privacy-preserving, secure and premiere application distribution platform of our time.

1.1. In recognition of these circumstances, when first launched, applications must set a drab tone of national anguish and mourning, alas, of how slowly, regrettably from carefully considered, meticulous design, by unthinking regulation does emotive, deeply resonant and in-app subscription renewals as mediums for transcendental application experiences slowly unwind. The tone shall last for no shorter than five minutes and use UIVisualEffectView with the effect UIRegulationEffect.

2. Users of Apple devices choose Apple devices because of their unparalleled ability to deliver a secure and trustworthy environment, free of malware and fraud. In a volatile regulatory environment intent on distorting them, applications must remain conducive to and maintain these pillars of integrity.

2.1. Users who download an alternate application distribution environment will be required to regularly give consent to this agreement. This consent alert must contain a link to communicate their concerns to the applicable legislature or government agency.

2.2. Games or Apps targeted for kids are advised to use the more approachable phrasing "stop hitting yourself", and of users of alternate application distribution environments as "doodyheads". [Marketing: should we check for localizable equivalents?]

3. The flag of the European Union is, like, totally stupid. Developers who instead use the provided materials (see Ted Lasso Resources, later in this document) are entitled to three months of Apple TV+ upon successful completion of App Review. (Offer valid once per developer account.)



Introduced for yesterday's betas (macOS 12.3, etc). A more finely-grained API, tuned for the needs of screen capture – letting you specify which apps and windows appear, the format of the capture, frames being able to be elided if nothing has changed, and so on.

And interestingly accompanied with a pull request to OBS to integrate it right away (under the banner of "Developer Ecosystem Engineering"), with changes that seem to follow the current work and adhere to its development process, ie not just a code dump or kthxbai.patch file against a months-old revision.

Blue Skies and Sour Apples

The Dutch Authority for Consumers and Markets has an effort to enforce the recent ruling that App Store dating apps must be allowed to use third-party payment methods. I've been following it, and Michael Tsai's roundup contained a biting Twitter quote from Francisco Tolmasky:

It must be really frustrating to roll out an entire plan only to be completely at the whim of a reviewer — err, regulator — to find out whether it sufficiently meets a set of vague criteria.

You could argue that Apple should not be subject to drive-by, after-the-fact regulation, clumsily anticipated "claw game" engineering by finite enumeration of ostensibly reasoned points, each tile creating more cracks and crevices through which people and features will fall. You could argue that the people making these decisions are ill-equipped to do so, and that their incentives are not aligned neither with the developers, nor with the person who's going to be using the thing.

What you can't do is argue these things and then turn around and say that it's all fine when Apple is doing it to developers. Time after time, Apple has either completely missed or blatantly ignored what makes a sensible application for developers and users, focusing on what draws in more money or tightens the regulatory noose.

A recent installment of the ongoing Floatplane App Review saga is instructive. Floatplane COO Luke Lafreniere has ongoing issues with getting a media network app already on the App Store through review, and time after time struggles despite carbon copying solutions adopted by big and successful apps on the App Store, to the point of copying UI wording verbatim.

I won't spoil the ending of the particular story, but even though the developer reach-out representative he is in contact with is cordial, empathetic and understanding, the situation is plainly, on its face, objectively ridiculous. No one is served by the status quo, and there's significant reason to believe that well-meaning, non-evil people on App Review would love to be able to wave a wand and just have things make sense. But that's not the system that's in place.

With the result being busted, unpredictable update schedules and feature Yahtzee slowing down development, not even Apple comes out ahead in practice, at least if the argument that what happens in all apps on their devices reflects on them and their platform is to be applied more than selectively. And after more than a dozen years of this, we have all heard our fill of these stories. If your baby's still teething well into its teens, maybe it's time to reevaluate. (I'm not a doctor, but I suspect spiritual malnourishment.)

The only surprise is that it has taken so long for any authorities to notice the situation and take any sort of action at all. Development turning into what looks and feels like a cruel, counter-productive, systemically capricious parlor game? Welcome, Apple, to the App Store experience; at least you get a €50,000,000 cap for your arbitrary rent.

jwz on Twitter and blocking

Sometimes I think about whether it's worth it to just swallow the pride, stubbornness and resentment and Develop a Social Media Presence™ for Better Engagement with My Personal Trademark® after all, but then I read things like this.

A Machine fed with People

Ars Technica, December 22, 2021:

The Securities and Exchange Commission has rejected Apple’s petition to block three shareholder proposals from going to a vote at its next annual meeting—a win for activists that signals trouble for other US companies hoping the regulator will allow them to fend off unwanted attention.

The resolutions call for detailed reports regarding allegations of forced labor in Apple’s supply chain, explanations of why certain apps are deleted from the App Store in China, and a public report of what risks the iPhone maker could face by allegedly using nondisclosure agreements in the context of workplace harassment and discrimination.

Cher Scarlett, December 22, 2021:

I'm currently unemployed. Apple is threatening to sue me and I have $92,000 in attorneys fees I have no way to pay. I gave up my severance and COBRA.

I did this to hold Apple accountable.

If you are an Apple shareholder, please vote to audit the use of concealment clauses.

Cher has since, happily, been employed, but her story about some recent events is harrowing and hard to boil down to one or a set of quotes.

Contrary to what regular readers may believe, I am an Apple shareholder. I don't own many of them and I am not located in the US, but ever since I read this, I have been fighting windmills trying to find a way that I can register for the meeting and place votes. It seems like most financial institutions here are only interested in the stock market insofar as to give easy marks a way to bet on the stock ticker casino, and are not quite so enamored with details like consequences, accountability and ethics, or at least not with letting the people without supercars have anything to do with it. (My bank told me it would be happy to sign me up for the meeting at a price that would amount to more than half of the shares' value; my broker, through which I hold the shares, just plain "doesn't offer the service".)

If you own Apple stock, please hold Apple accountable by voting at the upcoming shareholder meeting. Please confront the company, whose valuation you are helping inflate to ridiculous proportions, with the consequences of their actions. Please ensure that the self-perpetuating, reality-distancing bubble of innocence through excellence does not see challenges to live up to its own ethical self-image as reasons to throw employees in the corporate wood chipper. There is absolutely no excuse whatsoever for Apple's behavior; those who are breaking out "cultural fit" would go into a different kind of fit if it happened to them, or at any other company.

If corporations are indeed people, they are sociopaths with bastard sprinkles. At their best, they are upside down pyramids, and keeping true and fair and reasonable and not throwing customers or "human resources" to the hounds is only maintainable by constantly trying to tilt them upright.

Bloomberg: Apple Aims to Prevent Defections to Meta With Rare $180,000 Bonuses for Top Talent

Completely unrelatedly, a few words from Pixar co-founder Ed Catmull.


Our humanity is bound up with one another's. We saw it in South Africa: if you carry out a policy that dehumanizes others, in the process, you are dehumanized. You understand how others can say: in order for me to be me, I want to forgive you. Because when you don't forgive, frequently, you feel it in your tum-tum.

Cliff L. Biffle: On Hubris and Humility

Oxide is a neo-Sun Microsystems-esque startup trumpeting open and transparent architecture and practices while also designing the most hilariously oblique product teaser in recent memory (plop down contact-sales levels of cash to buy entire racks of AMD EPYC processors and NVMe storage before knowing what the hell you can even run on these things; I'm not in the target audience, but clearly the target audience has an even bigger need for details).

Oxide uses enough Rust to name the company after it, and the most recently revealed Rust-powered artifact is the operating system Hubris, designed for 32-bit microcontrollers instead of what you might be thinking of. I am completely unaccustomed to this level of programming, but the industry as a whole seems to have breathed a sigh of relief in the past decade that they can embed an ARM core and run Linux on most everything, theoretically allowing for fast and open-ended development but also for many uncertain side effects.

The full Hubris reference is worth reading and explains the eye-openingly fresh approach of baking things into static slots, avoiding asynchronous operations (since synchronous operations at least can't cause unbounded growth) and having a more-or-less crash proof architecture.

(You may also remember Cliff from the excellent post about rewriting his VGA-generating library for microcontrollers from C++ to Rust.)

On The Mend


Apple today announced Self Service Repair, which will allow customers who are comfortable with completing their own repairs access to Apple genuine parts and tools.

The conversations around original parts, vendor-blessed repairs, right-to-repair and freedom-to-tinker are often mixed up into one subject, with one "for" and one "against", the failings of one end used to deduct the other.

Apple, like many other manufacturers of mechanically sophisticated electronics, basically use the following argumentation: Our products are custom-fit for their purpose. We tirelessly chased down, manufactured or customized the right parts for the job, and we spent significant time on fit-and-finish so that everything works as intended. Therefore, care needs to be taken when repairing.

This is a reasonable, coherent position such as it is. The problem is the experience that the actions taken on behalf of it lead to in reality.

In reality, repairs with authorized service providers have to abide by glacial policies that prevent at least non-Apple Stores from stocking up on spare parts. Over the years, and across many separate incidents, I have had repairs of iPhones and MacBooks take many days beyond what the actual repair work would take. For devices that are intended to be used every day, this is not acceptable, and is the worst kind of business decision: the one that maintains control and policy at the expense of customer convenience. This is exactly as stupid as that pirated-vs-original comparison.

In reality, there exists two axes: qualified vs not, as well as Apple-blessed vs not. The Self Service Repair program highlights the absurdity of the emerging narrative: a random person at home, taking things carefully, can be trusted to do a job that an independent repair shop, which lives and dies by its reputation and literally does this all day can't be. Apple's previous position deserves some back-handed credit, in that at least "only we could do this" was free of such loops of logic – it was plainly and obviously wrong, but it was not that particular brand of ludicrous.

In reality, many people, from hobbyists to competent staff at repair shops in every village, town and city are perfectly well equipped to perform repairs, but are being prevented from doing so with original instructions and with original parts. While some would gleefully use knock-off parts to save a penny, most who do use them out of necessity, as a last resort.

There would be an instant and sharp drop in quality issues related to post-market repair if original parts and original instructions were simply available to everyone. This is not science fiction; with cheats, liars and penny-pinchers abound across all humanity, you can still repair cars without imminent physical danger, and off-network dealers or shops can still use original parts and instructions, at least in the EU.

In reality, miniaturization, water-proofing and minimalistic product design are not one-way funnels towards inexorable outcomes, but aspects in tension, to be managed. It is perfectly possible to make tiny, well-built objects with great fit and finish. In particular, it is perfectly possible to make minute changes to make disassembly enormously difficult and time consuming, or pedagogical and simple. (And as has recently been made clear, it is possible to needlessly entangle things that are completely disjoint, and undo this later.)

Just as it is possible to design a device that is already hard to make with a special focus on friendliness and usefulness, just as it is possible to ponder how to source the materials responsibly and sustainably, it is possible to make ease (or even possibility) of repair a priority, and it would serve everyone to do so, including Apple itself.

But this isn't about saving Apple money. It is about rejecting a simplified, convenient, harmful world view. It is true that not everyone is capable of repairing everything and not everyone wants to repair everything. But this is not an excuse or an instruction to let the whimsical, anti-customer demands of self-obsessed corporations with a control fetish dictate misery for everyone who does not live within five minutes of an Apple Store. (In Sweden, which globally speaking has an outsized number of locations proportional to our population, roughly 75-80% of the population live more than 1 hour from one.)

And it is particularly not an excuse to perpetuate the myth that knock-off parts are a choice in all cases, and are preferred by individual repair shops, which are therefore dishonest and/or not about to do a good job anyway. The entirety of the market, and the quality of its collective outcomes, depends on Apple's policies. They can change it tomorrow.

If we are lucky, Self Service Repair is not a cynical fig leaf in the face of approaching regulation, but the first step of accepting responsibility and extending to their customers the respect for their time, property and business that has been long due.

CNBC reporting on TSMC's new US factory

With software taking over so much of the world, hardware and particularly advanced chip manufacturing needs to become much more of an industry-wide discipline instead of the secret sauce of a handful of foundries. TSMC is doing a great job but the sparsity of serious competitors puts them under incredible pressure. (If you think the situation is dire now, when the Playdate's CPU was booked for two years solid and they had to switch for the new orders, imagine if TSMC messed up.)

I've had nightmares thinking of ways that China's increasing influence and hardline rhetoric would come back to bite both factories in China and factories in Taiwan, which would affect many companies the world over and many kinds of products. The clip argues that China knows that touching TSMC is bound to mess them up just as much, which is sobering, but also makes you wonder about the reaction to new factories in other locations.

Ben Eater on keyboards

I have recently been enjoying some of Ben Eater's videos on low-level hardware and implementing physical interfaces.

In particular, the series on how a USB keyboard works, how USB device discovery works and how n-key rollover works are measured and methodical – literally measured, as in hooking up the leads to an oscilloscope and going from electrical to digital.

My nose is sometimes deep into network/IPC protocol design these days, and so I appreciate the similarities and differences between software-only and protocols that have to bridge two pieces of hardware. Sockets can be frustrating, but you don't have to worry about wires turning into antennas, differential signaling, clock desyncing and bit insertion.

See also, or perhaps start with, the PS/2 keyboard interface video, which gets both practical and visually deadpan. (After watching it, I feel like I understand better what you'd use the Raspberry Pi Pico's PIO layer for.)

Wings of Pegasus: Visualization of auto-tune

This makes me really upset.

The example being used is Michael Bublé, who by all standards can sing on tune just fine, as mentioned by Fil. Someone, at some point, decided that wasn't good enough; that authenticity or mere human capacity was a weakness.

Auto-tune used as a parodic instrument of its own I understand. Used to correct outright mistakes in a pinch, fine, whatever, that's the deal with editing in the first place, and quite often you wouldn't come close to an "album version" by listening to any single take, even the best one.

But this, this fetishization of the perfect, digital hit of a key, roasting the analog and dynamic nature of not only the human voice but organic sound, even when the source material is a talented vocalist in real life, is just absurd.

Aditya Mukerjee: I Can Text You A Pile of Poo, But I Can’t Write My Name

My family’s native language, which I grew up speaking, is far from a niche language. Bengali is the seventh most common native language in the world, sitting ahead of the eighth (Russian) by a wide margin, with as many native speakers as French, German, and Italian combined.

[..] Until 2005, Unicode did not have one of the characters in the Bengali word for “suddenly”. Instead, people who wanted to write this everyday word had to combine three separate, unrelated characters. For English-speaking teenagers, combining characters in unexpected ways, like writing ‘w’ as ‘\/\/’, used to be a way of asserting technical literacy through “l33tspeak” – a shibboleth for nerds that derives its name from the word “elite”. But Bengalis were forced to make similar orthographic contortions just to write a simple email: ত + ্ + ‍ = ‍ৎ (the third character is the invisible “zero width joiner”).

Even today, I am forced to do this when writing my own name. My name is not only a common Indian name, but one of the top 1,000 names in the United States as well. But the final letter has still not been given its own Unicode character, so I have to use a substitute.

Worth reading in its entirety.

My prior view on the "Han unification" process was that it was undertaken to make the process of capturing all characters in use easier, and to thus increase the number of characters that end up in Unicode. But the comparison to a Latin/Greek/Slav equivalent gives me some pause as to whether it is an effective process, and whether people with deep knowledge would want to participate in it.

(Note that the original article is from 2015; things undoubtedly change over time, but I don't remember hearing about significant changes in the Unicode process recently.)

Review: MacBook Pro 16" (late 2016, Earth II)

I told myself I wasn't going to do this.

Just under two years ago, I had been holding out for years for Apple to introduce a MacBook Pro that undid the sins of the late 2016 model: Touch Bar, butterfly-mechanism keyboard and entirely Thunderbolt 3/USB-C. With rumors swirling all 2019, it took Apple until November 13th to introduce the first 16" model, which ditched the problematic butterfly switches, shrunk the Touch Bar to allow a physical Escape key and bumped the specifications.

With my then current pre-Touch Bar model crumbling, I bit the bullet and upgraded. It cost, by far, the most I had spent on a Mac, dongles to enable connectivity and compatibility and additional chargers not included. But it was a necessary investment, budgeted to last four to five years.

What none of us knew at the time was that work was already proceeding on its successor; the first Apple Silicon full-sized, full-featured MacBook Pro made for professionals, made to include all the features we had been clamoring for all these years.

The keyboard

The keys are, from all I can tell, the same type as the ones used in the 2019 model. Some say they can feel a difference between the older model MacBook Pros and these, and side by side I can maybe make it out, but they are what I want in a key. Snappy and distinct, with the right level of travel.

What's more, there are now 12 additional keys, taking the space of the Touch Bar, and stretching to fill a full-sized row, leaving the arrow keys (rightly) as the only half-height keys. After two years with a Touch Bar, being able to reach for and press a full-sized volume key feels almost ostentatiously luxurious.

Aesthetically, the keyboard well has been anodized a matching black, and it is a great decision that cleans up the visual clutter somewhat without sacrificing the legibility of the key shapes.

The function key system shortcuts are new, and the keyboard backlight controls are notably relegated, possibly in a message that it is time for the functionality to fade to the background and for the brightness to adjust itself. (You can still adjust the brightness in the Keyboard part of System Preferences.)

As for the rest, Do Not Disturb gets its own key, as does Spotlight and Siri. I don't use Siri and pressing its key (F5) is now permanently allocated to bringing up a window asking me if I'd like to change my mind. Together with Spotlight, which I always invoke faster with Cmd+Space, I would like to be able to reassign these keys to different functions.

As with all Apple Silicon Macs, the fn key now pulls double duty as the "Globe" (🌐) key, which when pressed alone brings up the Emoji palette. This feature has been available since macOS 11, as a checkbox option, but the key is now also referred to as the Globe key in the user interface of some settings related to keyboard layouts.

The ports

Blissfully, the sides now feature many things besides just plain Thunderbolt/USB-C. Let's start from the left.

First, you can find a new MagSafe 3 port for magnetically attached charging, which now involves a plain, braided and still expensive MagSafe-to-USB-C cable, but which at least is a separate purchase. Additionally, the 3.5 mm headphone jack has been moved back to the left edge (which I personally liked better) and it supposedly works better with high-impedance headphones.

Going around to the right, you will find an HDMI 2.0 port (and not HDMI 2.1), and an SDXC port capable of UHS-II, but not UHS-3. Neither are the absolute best, most up-to-date, most capable that could have appeared in their spot, which is frustrating considering the price and profile of the computer.

You will also find three Thunderbolt 4/USB 4 USB-C ports (two on the left, one on the right). What you won't find is even a single USB-A port (the traditional USB port). As someone who, even after years of accumulating USB-C devices, still needs to plug in USB-A devices, and still needs to do so much more often than use SD cards or HDMI, this is beyond frustrating. I would still have needed a dongle for the occasional Ethernet use or different video outputs, but I could have minimized my need to carry around, keep track of and remember to bring a dongle. After all, during the introduction of the MacBook Pro, Apple itself claimed to heed the wishes of professionals to not need adapters. With that goal in mind, eschewing even a single USB-A port is a very odd trade-off, since it, unlike Ethernet (which many production professionals also need) would have fit on the side.

The display

Returning to 2019 for a moment, it also saw the tremendously expensive Pro Display XDR, aimed to act as a less expensive alternative to professional production-level reference monitors. Whereas the Pro Display XDR had 576 local dimming zones powered by LEDs, whose need for constant cooling informed the lattice pattern on the back of the display, the new MacBook Pro display is one of the first mini-LED displays, packing in 2500 local dimming zones (with 4 mini-LEDs each), into an assembly that is millimeters thin, and still delivering the same color space, 1600 nits peak and 1000 nits sustained brightness specifications.

The new display also picks up 120 Hz and adaptive refresh rates (ProMotion), putting it still behind many PC laptops and desktop monitors (some of which go for 360 Hz), but matching this year's iPhone 13 Pro and 2017's iPad Pro, and besting the 60 Hz Pro Display XDR (which audience is often reticent to venture above film refresh rates at all).

In a sentence, the quality of the display is probably the best display you've ever seen on a laptop. OLED-level inky deep blacks with sharp contrast and even higher resolution. The bezels are even narrower, with the top corners even following the curve of the casing. You may even call it…

A top notch display

Smack dab in the top middle of the display is a notch; an area where the display ceases, for the benefit of the camera and related sensors. I think notches are an eyesore that at best you learn to ignore, but in this case I see a number of things working to its advantage.

Without the notch, the display would be roughly equal in proportions and area as the prior 16" MacBook Pro, and would match a full 16:10 display. The additional area is extra space and houses the menu bar, which for me personally is sparsely populated in almost every application. In applications with a large number of menus (which many professionals could need to use), or in any application for people who use dozens of the top right "menu extras", you do start to run out of space, with menu items wrapping around or their titles being truncated. This is not a good look, and it may be a better look to enable the quizzically titled Get Info option "Scale to fit below built-in camera", which may as well be named "Move below the notch", which causes the display rows alongside the notch to be ignored.

In iPhones, full screen widescreen video brings out the worst of the notch, making it cut out an unsightly bite of the display. Here, the aspect ratio of videos work to its advantage. Most are 16:9 or wider and are already centered in the display with letterboxing, putting them far away from the notch. The mini-LED nature of the display also leaves the letterboxing areas black and the notch barely distinguishable at all.

So is the notch needed from a technical standpoint? We've all seen phones; why can't you just fit the camera up in the bezel area? Because on phones, you still get 2-4x the depth to work with, and cameras can be small due to fitting all the circuitry behind the lens and the housing. Here, the module needs to be spread out instead, which means placing it entirely inside the bezel area probably did not work well at all. I'm sure you could fit some form of camera in the space allotted, but I'm not sure it would be one you'd want to use.

Given the result is both a good camera and additional screen space, and given that I don't suffer from stuffed menu bar syndrome, I'm willing to forgive the notch. My main complaint is that it's not possible to, out of the box, set the menu bar to pitch black, which would hide the notch much better. This is not a guess; opening an application in full-screen does this, and it is indeed almost not noticeable.

Of cores

I wanted to maintain my 64 GB of memory, which meant I had to pick the M1 Max, which is the only variant that allows for 64 GB. The heaviest pressure I put on the GPU in daily use is to have way too many windows and tabs open. On the previous MacBook Pro, the GPU had 8 GB of memory to work with, and I would not be surprised if most of it was populated with texture caches for all the various surfaces to draw.

Historically, shared memory between a CPU and an integrated GPU has been a feature of the low-end processors, or at least of the energy-efficient processors. Those tend to work with miniscule amounts of RAM and over slow buses, and the circumstances have added up to a poor experience – a slow GPU, stealing a static amount of memory from an already constrained CPU.

"Unified memory", as Apple terms it, on the most fundamental level is not different from these shared memory shenanigans. But as implemented on the M1 Pro and M1 Max, the balance shifts. The memory bandwidth is now 200 or 400 GB/s, which is in line with what purposely high-bandwidth directly-attached GPU memory provides. The access latency is very low, and the throughput is significant. Most of the GPU rendering is also probably being done by system frameworks using Metal, which is optimized to use representations for pixel buffers or textures that the GPU cores natively need, which means the CPU can produce an image and the GPU can use it without transforming it into another representation, and without even needing to copy it into another location. On the scale of a system, all of this adds up. Instead of two cooks trying to work separately in a galley the size of a closet, it ends up more being two cooks trying to cooperate, preparing and picking up from one another, in a sizable and well-kitted-out kitchen.

I have not used an M1 Mac, but have heard tales of a single core saturating the memory bandwidth, and one of the reasons I stayed away was because of the limited connectivity. The architectures of the M1 Pro and M1 Max, keeping the fundamental CPU cores themselves but expanding on everything else, shows that these shortcomings were not intentional declarations of good enough, but transitional pains that will eventually go away. Depending on the benchmark chosen, many things can be shown, including that the GPUs are far from adapted to be competitive with desktop GPUs at gaming. But the collected performance of the CPU cores is competitive with many desktop CPUs from Intel and AMD, and this bodes well for the future.

All of the above in an architecture that in both an absolute and relative sense sips power and thus both gives long battery life and minimizes fan noise and heat makes it hard to be calm and reasoned about the performance of the MacBook Pro.

Industrial design

In 2015, the 12" MacBook started a new era of industrial design and of priorities for Apple. For one and a half years, its impact on the upper end of the MacBook family was unknown, but the 2016 MacBook Pro brought in slimmer lines, the butterfly keyboard and entirely USB-C I/O.

As of last week, this era is over and buried, and the 2021 MacBook Pro, being the first major redesign, defines a new picture. The Pro products are allowed to be more substantial, in every sense of the word. Thicker and heavier are not absolute terms of revulsion, but axes along which the product can be measured. There are laptops three times the thickness of the 2021 MacBook Pro, and half the thickness as well. It is a reasonable size and weight for its intended use case, and its format allows it to pack in features that are important. Focusing, more than just saying "no" as many times as you can to various impingements of a designed dreamed up in isolation and divorced from purpose and mechanical realities, is the process of choosing tradeoffs.

Design is how it works, and the 2021 MacBook Pro, through its 100% recycled aluminum shell, through its comfortable keyboard and outstanding trackpad, through its sharp display, through its groundbreaking M1 Pro/Max system-on-chip, through even its camera notch and still-too-few ports, works very well for its intended use. It knows its market – the people who need to or who want to do a lot of things with their computers – and is designed for them.

It is also expensive. More than any high-end Mac in a decade, it provides a mix of features and capability that could actually command a premium, even if we weren't in a worldwide semiconductor crisis. And it finally catches up to high refresh rate displays.

In 2019, the 16" MacBook Pro told us: hey look, at least they're trying again. In 2021, the 14" and 16" MacBook Pro tell us: breathe a sigh of relief. The MacBook Pro has rediscovered what it is.

The hard part now will be going four years without getting the next one.

Good Things Take Time


I spent part of 2019, all of 2020, and most of 2021 working on the new MacBook Pros announced today! I designed the M1 Pro and M1 Max dev boards. then I was the M1 Pro and M1 Max system integration lead, and new miniLED and camera system integration lead.

Reminder to myself and to other people: good things take time. Especially trying to make something brand new come to life for the first time.

Voicing complaints is easy. Building things is hard. Thank you, Colin, and all the people you work with, and every other person out there who sits down with a blank sheet of paper and slowly and steadily makes new things happen.

So Ordered

Clear, reasoned, calm thinking does not emanate within an hour of an announcement. I'll do my best, but let's instead call this a celebration.

  • MagSafe is back, at the cost of one Thunderbolt/USB-C port.

  • All Thunderbolt ports are now Thunderbolt 4, instead of the "USB 4/Thunderbolt 3" of the M1 Macs.

  • The Touch Bar is gone, with full-height function-row keys to take its place, and leaving Touch ID alone. Could it ever have worked? Maybe, but Apple was interested only in planting the flag and claiming victory instead of acknowledging its failures and working around them. The world is not full of successful touch surfaces that you use by feel, without looking at and which contents change constantly. Considering its mention in the presentation, even Apple eventually thought getting back to what a keyboard is all about was a defendable move.

  • Out of the fog of "M1X" steps M1 Pro and M1 Max, née Jade C-Chop and C-Die. M1 Pro has 16 GPU cores and up to 32 GB memory; M1 Max has 32 GPU cores and up to 64 GB memory. I (and my wallet) might have settled for M1 Pro with 64 GB of memory, but that is not how these things work, and now with some technical justification.

  • M1 Max's GPU chops are apparently "comparable" to laptop-level discrete "Max-Q" chips (GeForce RTX 3080 Laptop, in their comparison). In other words, it's not a bloodbath in terms of maximum performance, and it is not yet a credible alternative to the highest performance discrete desktop GPUs. But if your needs are lower than this, it looks promising.

  • That said, the unified memory strategy means that, as long as you don't want to max out both CPU memory and GPU memory, the ability to address a ton of it over a fast (200 or 400 GB/s) bus is a sleeper hit of this architectural choice. (And even if you do; 32 GB of each does not come easily in most PCs.)

  • The more square-ish form factor reminds me of the aluminum PowerBook G4, and is probably great for airflow.

  • HDMI, SDXC, good. I could have done with a USB-A port or two; them and wired Ethernet are prime reasons to not throw away your dongle.

  • 120 Hz, ProMotion/adaptive refresh rate, mini-LED display, with higher pixel density and display brightness fit to deliver good HDR. I would have been happy to have two or three of these; a notch for the webcam is a negative, but much less of a negative on an OS with a menu bar (assuming the OS knows not to put anything underneath, of course).

  • With M1 Pro and M1 Max, the mirage of an entire product line supported by essentially the same M1 chip is now killed; arguably, it was killed by the M1 itself taking the place of A-series chips in the iPad Pro earlier this year.

At some point in the past, from all I hear, the PowerBook G4 was faster than the then anemic Intel mobile chips despite its lower clocks. When I had mine, its own anemic 167 MHz front-side bus put a damper of any claims to supreme (or at times even acceptable) performance. The switch to Intel and the then-new Core Duo brought a welcome jolt, but at the expense of "just" being as good as the rest of the portable PC market.

Today, for the first time, Apple steps into the brave new world of not just terrific performance per watt, but assuming all the claims hold water, a level of laptop performance that is barely achievable with machines that are several times thicker, heavier, louder, and last for a fraction of the time on battery.

Today, with some caveats, Apple delivered a MacBook Pro that was for actual Pros again, for the people who need more ports, more performance, more capable hardware; who want their computers to be computers, instead of scaled-up app machines.

And today, at long last, Apple listened to its users once again, took their needs and their advice at their word, and made a better computer.

← Earlier posts