Linked List: May 6, 2024

Monday, 6 May 2024

Juli Clover, MacRumors:

Hovering over the Apple logo and moving the mouse allows the current artwork to be erased and replaced with a new logo design. Apple created a total of six logos for the May 7 “Let Loose” event, and the interactive eraser cycles through those options.

Would be cool if Apple’s managed to pull off being able to use the end of the new Pencil as an eraser, just like a real pencil.

Fitting Facts to the Narrative at The Washington Post ★

From a Washington Post story headlined “Apple Is Behind in AI and Killed Its Self-Driving Car Project. What’s Next?”:

The company’s Greater China region, which encompasses mainland China, Hong Kong, Macao and Taiwan, has long been one of Apple’s most crucial growth zones. But growing pressure from a handful local rivals — including Shenzhen-based Huawei, which surmounted U.S. sanctions aimed at slowing its advance in late 2023 by producing a smartphone with a domestically made processor — cut sharply into Apple’s market share in the region earlier this year.

Data from market research firm Counterpoint Research indicated that Apple’s sales in China dipped by nearly 20 percent in the first quarter of 2024, a shift that senior research analyst Ivan Lam attributed partially to “Huawei’s comeback.”

The full scope of the company’s decline in China became clear Thursday, when Apple reported an 8 percent revenue dip compared to a year earlier.

It’s inexplicable that the Post included a paragraph with projections from Counterpoint claiming iPhone sales in China were down 20 percent even after Apple reported its actual results for the quarter. Jason Snell, over at Six Colors:

Finally, I particularly enjoyed the exchange between Wells Fargo’s Aaron Rakers and Cook in which Rakers asked Cook to explain Apple’s results compared to the data reported by independent research groups that suggested iPhone sales were falling apart in China. Apple’s actual numbers weren’t that bad, and in fact, Apple trumpeted how well the iPhone was going in urban China.

“I can’t address the data points,” Cook said. “I can only address what our results are, and you know, we did accelerate last quarter. And iPhone grew in mainland China, so that’s what the results were. I can’t bridge to numbers we didn’t come up with.”

That’s about as savage a shade-throwing as you’ll get on an Apple analyst call.

iPhone grew in mainland China last quarter but even after Apple announced that — in a legally-binding context — they went with made-up projections from Counterpoint to fit their narrative that Apple is in trouble.

While I’m being grumpy, I’ll even take issue with the notion — which the Post leads with in its headline — that Apple is “behind in AI”? It is true that Apple doesn’t offer an AI chat product like OpenAI’s ChatGPT, Microsoft’s Copilot, or Google’s Gemini. But do we expect Apple ever to offer such a project? Apple doesn’t have a web search engine but no one is arguing that Apple is “behind” on search. (App Store search results quality is another issue.) Apple doesn’t offer turnkey cloud computing services like AWS or Google Cloud either. Are they “behind” on that? When it comes to the products Apple already sells, how are they “behind on AI”? Are iPhone users missing out on AI features available only to Android users? No. Are MacBook users missing out because Apple hasn’t added a dedicated AI key to their keyboards?

I get that people see AI as a frontier that is transforming the industry, and Apple hasn’t revealed any new plans or features yet. But I’d say Apple is silent on AI, not behind. When iOS and Mac users are missing out on features that are only available on other platforms, that’s when I’d say Apple is behind.

Logitech’s Mouse Software Now Includes ChatGPT Support, Adds Janky ‘ai_overlay_tmp’ Directory to Users’ Home Folders ★

Stephen Hackett, writing at 512 Pixels:

I know AI is all the rage right now and having a deal to bring ChatGPT into your software is trendy, but including a tool like this in what is basically a mouse driver is ridiculous. I’m not opposed to using AI in software. I’m just opposed to when it shows up as an unexpected, poorly-implemented feature in software that doesn’t need it.

At least Logitech’s Mac developers did such a bad job with it, that it was easy to spot.

Logitech committed a bunch of sins with this mouse driver. First, it just seems ridiculous to add an AI prompt feature to a mouse driver. Second, no matter what the feature, it’s wrong to add a top-level folder to a user’s home directory — and it’s especially wrong to give such a folder a dumb name like “ai_overlay_tmp”.

It’s more common for poorly-programmed Mac software to create such folders with a leading dot in their names, an age-old Unix convention that tells the Finder to treat them as “invisible”. But that’s poor form on MacOS too. Support folders should be organized in standard sub-folders inside the user’s Library folder. Open Terminal and type ls -a at the root of your home folder and you’ll probably see a lot of detritus that ought to be inside your Library folder.

Hackett has switched from Logitech’s mouse software to the excellent SteerMouse, an excellent $20 mouse driver that supports just about every mouse in the world. I’ve been using and wholeheartedly recommending SteerMouse for nearly 20 years.

It’s also the case that even with a third-party mouse, you might not want any third-party driver software at all. MacOS’s built-in mouse software recognizes most mice. I rely on SteerMouse not because my mouse has lots of buttons (it doesn’t), but to get fine-grained control over the speed and acceleration of the pointer. SteerMouse lets me set my mouse to go way, way faster than the built-in Mouse panel in Settings does — something I’ve done for decades to reduce wrist fatigue and pain. I can move my pointer from corner to corner across my Studio Display by moving my mouse just a few centimeters.

The problem is, many people — perhaps especially people whose computing experience was forged on Windows — wrongly assume that if you buy a Brand X mouse, you need to install Brand X’s software to use it. For Logitech mouse users, that means AI software they neither want nor need running in the background, and an ugly cryptically-named temp folder stinking up their home directory like a squatter.

Yours Truly on ‘First Ones’ ★

Brian McCullough at the Techmeme Ride Home podcast has a new YouTube series, “First Ones”, where he asks his guests for their firsts — first computer, first movie in a theater, etc. Guests so far, in addition to me, include Guy Kawasaki, Bluesky CEO Jay Graber, Ed Zitron, and more. Very fun.

Apple Watch Ultra’s Best Feature: Battery Life ★

Jason Aten, writing for Inc.:

I don’t run marathons or climb mountains or dive. I don’t often find myself in remote locations in precarious situations. I joked that, for me, the Apple Watch Ultra will be the perfect thing to wear while other people are working out. Like, for example, when I’m keeping track of my daughter’s cross-country race splits.

When I first started wearing the Ultra, I was sure the most useful feature was what Apple calls the Action Button. I still think the ability to assign a dedicated button to things like starting a workout or a stopwatch is great, but there’s a far better feature, and the most surprising thing is that Apple isn’t making a big deal of it at all.

That’s the fact that the Apple Watch Ultra has ridiculous battery life, at least by Apple Watch standards. Yes, Apple has said that the Ultra has the best battery life of any Apple Watch. But Apple is dramatically underselling the battery life on its new flagship wearable, claiming it gets 36 hours. In my experience, it got more than 60 hours.

Two thoughts about this:

Perhaps Apple underplays the Ultra’s battery life so as not to make the battery life on the regular Series 9 models look bad? From my experience while reviewing the original Ultra, I could get two days of battery life on a single charge even while wearing it to sleep.
The battery life on all Apple Watch models from the last few years offers a stark contrast to the dedicated AI gadgets we’re starting to see, like Humane’s AI Pin and Rabbit’s R1 handheld dingus. It goes under-remarked-upon that Apple is really really good at making computer hardware, and particularly at making small computer hardware.

Boring News: Vision Pro Sales Are Going Just About as Expected ★

Ming-Chi Kuo, two weeks ago:

Apple has cut its 2024 Vision Pro shipments to 400–450k units (vs. market consensus of 700–800k units or more). Apple cut orders before launching Vision Pro in non-US markets, which means that demand in the US market has fallen sharply beyond expectations, making Apple take a conservative view of demand in non-US markets.

Neil Cybart, on Twitter/X:

Ming-Chi Kuo’s numbers and statements regarding Apple Vision Pro sales don’t make sense. [...]

Interestingly, when framing a 400K to 450K unit sales figure for Vision Pro in 2024 (which would actually be a good result), Kuo compares the range to a made up consensus figure of 700K to 800K unit sales. I don’t recall anyone running with such a high sales range for Vision Pro in 2024. Instead, Kuo himself actually claimed back on Feb 28th that a few suppliers were planning for 700K to 800K production. That’s not the same as a sales forecast. Far from it. Kuo would know that too, so one is left to assume he’s purposely being misleading.

Ming-Chi Kuo occasionally uncovers legitimate scoops from Apple’s Asian supply chain. Ming-Chi Kuo also regularly inserts his own name into the news when he has no legitimate scoops. This is one of the latter. There was no “market consensus” that Apple would sell 700–800K Vision Pro units in 2024. In fact, the reporting has been around the 400–450K range since last summer.

The Financial Times, back on 3 July 2023:

Two people close to Apple and Luxshare, the Chinese contract manufacturer that will initially assemble the device, said it was preparing to make fewer than 400,000 units in 2024. Multiple industry sources said Luxshare was currently Apple’s only assembler of the device. Separately, two China-based sole suppliers of certain components for the Vision Pro said Apple was only asking them for enough for 130,000 to 150,000 units in the first year.

TheElec reported last June that Sony only had the capacity to manufacture 900,000 OLED panels per year for Vision Pro, which, if true, would cap Vision Pro headset production at 450,000 units. The Information reported in August that this display bottleneck “is one reason why Apple plans to make fewer than half a million Vision Pros in the first year of production”.

Meta Threatens to Pull WhatsApp From India Over Encryption Battle ★

Russell Brandom, reporting for Rest of World:

IT rules passed by India in 2021 require services like WhatsApp to maintain “traceability” for all messages, allowing authorities to follow forwarded messages to the “first originator” of the text.

In a Delhi High Court proceeding last Thursday, WhatsApp said it would be forced to leave the country if the court required traceability, as doing so would mean breaking end-to-end encryption. It’s a common stance for encrypted chat services generally, and WhatsApp has made this threat before — most notably in a protracted legal fight in Brazil that resulted in intermittent bans. But as the Indian government expands its powers over online speech, the threat of a full-scale ban is closer than it’s been in years. [...]

It’s not clear how the courts will respond to WhatsApp’s ultimatum, but they’ll have to take it seriously. WhatsApp is used by more than half a billion people in India — not just as a chat app, but as a doctor’s office, a campaigning tool, and the backbone of countless small businesses and service jobs. There’s no clear competitor to fill its shoes, so if the app is shut down in India, much of the digital infrastructure of the nation would simply disappear. Being forced out of the country would be bad for WhatsApp, but it would be disastrous for everyday Indians.

One thing to remember is that this isn’t so much a conflict between what the law demands and what Meta chooses to do, but rather a conflict between what the law demands and the secure-by-design nature of WhatsApp. There is no “traceability” switch that Meta could flip but is choosing not to. They’d have to build a completely new, insecure-by-design, protocol to comply with this law.

Kolide ★

My thanks to Kolide for sponsoring last week at DF. Deepfakes are good and only getting better. In real life, people only detect voice clones about 50% of the time. You might as well flip a coin. And that makes businesses extremely vulnerable to attacks.

In the “classic” voice clone scam, the caller is after an immediate payout (“Hi, it’s me, your boss. Wire a bunch of company money to this account ASAP”). Then there are the more complex social engineering attacks, where a phone call is just the entryway to break into a company’s systems and steal data or plant malware (that’s what happened in the MGM attack, albeit without the use of AI).

But the good news is that we can be trained to learn how to identify suspicious phone calls — even when the voice sounds just like someone we trust. If you want to learn more about Kolide’s findings, read their report exploring the details of audio deepfakes.