WWDC24 keynote reactions.
My early question was "why the slightly wide 'WWDC24' logo with the overlapping 'W's?". I still don't know about the overlap, but the wider font is a theme thing that carries through the developer content, possibly related to variably-weighted fonts, which were featured here and there in more animated fashion.
Good to see Phil again – looks like he's doin' well. The quick Airplane Mode toggle and other little modern Apple nods were cute.
This seems like the first time a WWDC keynote has ever had an opening title! Well deserved, especially considering they're more like movies now. (And I'm glad they've never piped had the cliché and audible "ladies and gentlemen, please welcome…" for presenters, as though they're worried the audience might fail to.)
Tim Cook seems to be channeling the late Billy Mays during recorded events now, tessitura-wise; perhaps a nice unintentional tribute. (He's still softer-spoken during interviews and such.)
Seems like a while since anything's been diamond-chamferred. What are they doing with the diamonds? Working on a way to recycle them?
Lovely intro for the calculator app, which in fact belied its power. We'll come back to that.
Control Center: circles amid rounded rectangles? Not my favourite. I think I wish it went the other way: all just slightly less-rounded rectangles.
There's always fine print during Apple events, but none of it seemed to warn about attempting activities such as parkouring down concrete flights of stairs or jumping from airplanes. ("Treat the user as smart"?)
Freeform app: apparently still no dark background.
Over-trodding terminology around "AI" has already proven more tiring than exciting, and "helpful" was the one I was hoping Apple wouldn't stoop to harping. (Google seemed to think a few minutes without it constituted failure.) They did use the word, but at least with more moderation and less of a sense of verbal abuse, if you will.
"Compute," as a mass noun, seems another trendy and otherwise displeasing term. So "Private Cloud Compute," seemingly Apple's outstanding example to the industry, deserves a better name.
The Watch's "vitals" app seems cool. Footnote on Apple site, though: "The Vitals app is for wellness purposes only and not for medical use." (There's "wellness," a term with more liability-related baggage.)
Apple Intelligence is "coming in beta this fall." I don't remember the keynote framing the fall release as "beta" at all, but it's all over the site.
How was the keynote overall?
The presenters spoke of "a new era," and "AI" has been the recent source of much public pleasure and concern, but for me, this didn't feel like a big one – certainly not like the introduction of the headset device last year. The first half of the presentation was dedicated to Apple's general platform updates, with a few nice new features for each. Not that I'll be in possession of a Vision Pro for some time, but the wraparound Mac display looks fantastic to me, and it lifted my heart to hear that foviation was performed on the Mac itself. But overall, the upgrades felt "incremental."
Refreshingly so. I miss that. The introduction of Snow Leopard (with its emphasis on refinements from the preceding Leopard, rather than further bounds forward) is one of the keynotes I still review most. Taking care of existing customers, optimizing and striving to perfect the flagship operating system's simplicity and stability, should remain in focus whether the company is new or dominant. Even during the "big" upgrades, I find myself seeking, noticing and appreciating the refinements.
I think what disappointed me about the "intelligence" angle was that Apple's been doing "artificial intelligence," in the broadest sense, for well over a decade. Apple's people know what they're doing, and their custom chips execute it well. But it's just been regarded as part of the product – it's never really needed a brand, and it still doesn't feel like it should need one. "Well, the whole industry is doing this, so I guess we have to do this" feels like the kind of thought Apple is generally wise enough to ignore.
Second-most disappointing? Now that they're doing it, they've featured much of the same stuff as everyone else. Rewrite your e-mail in a different tone? Okay, I guess. Little generated images in the style of Pixar characters (or, as we'll call it, "animation")? At least they have the nice interface and everything, I guess. It's all so impressive that any of this can be done at all – a form of magic even to ourselves a decade ago – which makes it all the more "impressive" that it's somehow already so natural to shrug at.
Out of all the generative features, a few struck me as genuinely good and useful, and none of them are the ones "the industry" seems to emphasize:
One is proof-reading. Just plain, simple proof-reading, where grammatical issues are not just surfaced, but explained. I'd like to think that might help more people genuinely be better language-wielders, and in particular, additional-language-learners.
"Genmoji" struck me as an actually-interesting version of image generation. Constraints can stoke creativity and satisfaction, and emoji have a well-established forms, styles, and, necessarily, gaps between standard glyphs which, even involving a thoughtful committee, are too tethered to time and place not to leave gaps that anyone might want to fill. The idea of a working generated emoji standard makes me think Gemnoji could be the first thing to capably succeed emoji.
The third is the truly amazing "Math Notes," capable of performing operations on handwritten math forms, supposedly even to the extent of inferring values from other handwritten variables and from geometric scribblings. That was announced in parallel with "Smart Script," letting the technology learn, replicate and refine your handwriting. Great stuff.
(And a better Siri wouldn't hurt, but I doubt anyone's holding their breath.)
But I think the real highlight was the promise behind the shudder-inducingly-named "Private Cloud Compute." With all this remote computation performed on people's data, Federighi is right that potential data misuse is nothing to ignore. Apple claims that off-device requests, when leveraged, are encrypted, only use the relevant user data, and don't store it. But the real achievement is server software which can be independently audited, and with which users' devices are designed not to communicate unless the software has been logged as independently auditable. I'll be waiting to hear how that goes, but it sounds like an actual step towards balancing power with responsibility.
Access to ChatGPT is also handled by the Apple platforms, but it comes in the form of Siri explicitly asking, per-request, whether you'd like to relay it. And each response comes with a reminder to check it for mistakes. The interface reminds me of Apple's (rather ephemeral) system-level integration of Facebook and Twitter from last decade.
One of my favourite reactions to the week's announcement has been this interview with Steve Wozniak; his enthusiasms and hesitancies match mine pretty closely.