Apple squandered the Holy Grail
xeiaso.net
xeiaso.net
One of the few examples I can think of however is Apple Maps. And it did get better; a lot better, some say better than Google Maps nowadays. So I generally do have hope for Apple Intelligence. At the end of the day, there are some disparate competing utilities in this class on the Samsung and Google phones, but no one is shipping something that is obviously game-changing and in first place; they all kinda suck, they're all tech demos, and it'll inevitably take many years to get this technology honed in to something that is truly useful to consumers.
iOS has a refinement that Android lacks but I am unimpressed with MacOS. Windows is stuffed full of terrible crapplets and Windows users largely recognize that these are terrible crapplets and don't use them. Apple users have a fixed belief that everything Apple does is brilliant and fashionable so they do use them which has a deadly effect on the market for third-party software. (No good music players for MacOS for instance)
Even Apple fans lately claim it's been getting worse in the last few years.
(That said, I love the innovation in the M-series chips from Apple just as much as I appreciate Microsoft's commitment to the long-term viability of Windows for all of us who invest in it. Occasionally at work we still use Access '98 to handle old files and it works great, the installer works great, in fact Office still tries to take the desktop over the way it did back in the day. Clippy still works. The borderless windows look just a little funny because the compositor changed. No way you could run Linux binaries or MacOS classic binaries from '98)
Many of the long term users of macOS/OS X/etc etc are highly critical of its downfalls, but still use it because of the available options they prefer it. Myself included. You can use something while also being aware of its shortcomings.
The key problem for desktop Linux is that nobody knows exactly how to build binaries that will run on any reasonable Linux desktop system today, so it's hard to keep that non-existent reasonable subset of ABI stable for an extended period.
That said, you CAN do this. The kernel itself does present a mostly pretty stable ABI to userland applications, so you can grab a Debian chroot from 1998 and be on your way. Debian even still serves repositories for everything on archive.debian.org, and Dockerhub has OCI images you can `docker run` for Debian from 1999, under the debian/eol repo. You can `docker run` and `apt-get install` 25+ year old binaries on modern Linux!
What would be sweet is if we could build and ship compatibility tools that make these old binaries work mostly transparently. Today, double clicking a binary on Linux won't do anything particularly sophisticated, and there are no compatibility options. But actually, it would be totally doable to write a variety of useful compatibility shims without doing anything horribly grotesque. The DT_INTERP and DT_NEEDED fields of binaries would often give sufficient information for how you might get such a binary to run. It's not like it would be that useful, but I would personally be very pleased if you could just double click e.g. some old Kylix application and have it just run, perhaps after downloading some (shims for?) old libraries. You could extend this to transparently running CPU emulators too, not unlike the tricks people do with binfmt_misc, just possibly with more batteries included (and a bit less transparency.)
Another really great feature would be useful error messages when executing an application fails. Today if the DT_INTERP is missing, it looks like the binary itself can't be found since it returns the same errno, and you won't see linker errors if you execute a file in a GUI file explorer. What a great improvement it would be if all of that could be fixed, and there is no technical reason it can't be.
Of course, frustratingly, for more reasons than just this, the more likely thing to happen is that nobody bothers since containers are the future anyways, and Win32 instead becomes cemented as the true stable ABI of Linux. Which, in my opinion, is a bummer. We could always have two stable ABIs of Linux...
Have to strongly disagree on this point. Cog¹ is my music player of choice on macOS; not only does it have a clean GUI, but it supports almost every format² I’ve ever wanted to listen to audio in, including game music in formats like GBS (Game Boy Sound System) and 2SF (Nintendo DS Sound Format).
――――――
Nah, Apple users knew from the beginning that Siri sucked and still sucks. Almost no one I know uses Siri except for setting alarms and asking for weather forecast.
assuming you mean on macOS, given your comment after this. (there is no iOS app called iTunes anyway)
yes, on both iOS and macOS, Apple intentionally hobbled iTunes/Music, incrementally making it worse each update, after Apple Music gained an initial foothold.
not that it was ever "great" after 1.0 maybe 2.0, but it certainly used to be good. Now, if you're not an Apple Music subscriber, you're left with a pretty basic player. I get that they want to segment the market, but to remove features and actively make it worse? horrible.
Can you give me a few examples of linux binaries from 98? I would like to give this a go, I think I have a pretty reasonable way to go about achieving this.
What are the good music players off MacOS?
I figured the entire field had withered and died from lack of public interest.
Slightly OT but Doppler is quite good in my opinion.
For one thing, the iTunes name doesn't technically exist anymore except on Windows. And anyone complaining about it being bad on Windows...I mean, that's like complaining that Microsoft Remote Desktop (Now called the Windows app for some reason) sucks on Mac, right? Like, can we just put the Windows version aside please? Even then, I'm not really sure what specific thing iTunes for Windows sucks at besides not looking like a Windows app. People just say that because they were saying it in 2005.
On Mac, the Music app (not to be confused with the streaming service) is fantastic and has supported Apple's "classic" digital music workflow longer than anyone else has been willing to support their users. The Apple TV app (again, not to be confused with TV+ subsciption service) is now the home for the music/TV show store/rental place and the home of your TV/movie library, which is a big improvement from shoving that functionality in iTunes. in that sense, Apple has cleanly separated use cases and functionality in a way that iTunes didn't previously, which is one reason why a lot of people said "iTunes sucks."
I have a family member who recently switched to Android because of frustration with Apple as a whole. They are a big digital music collector, they don't believe in streaming or "renting" their content.
I tried to help them with their music collection on Android. Theoretically it should be easier right? No weird restrictions on sync direction, basically dump your stuff on an SD card/transfer over USB-C and you're off to the races.
But still, they switched back to Apple secondarily because it's the only place left that actually makes that "purchased digital music" experience user-friendly, or possible at all. (Primarily they switched back to iPhone because the modem in their Google Pixel sucks and/or is poorly tested with their major US carrier and would drop international calls every 15 minutes exactly for no reason)
Google Play's music store doesn't exist anymore. Every jukebox app on Android depends on 100% manual file management. None of them have the polish of the Music app (the app not the service). Almost none of them have decent jukebox companion apps available on desktop computers. A whole bunch of other digital music stores have closed entirely.
Apple's system for synchronizing content is actually pretty amazing for continuing to support an offline cloudless workflow. You still just hit one button/plug in your device to sync your music, movies, audiobooks, ebooks, and photos content. It also supports WiFi syncing, and it furthermore supports every iPod that ever existed so long as you have the right cable/adapter.
You can back up your iPhone's full image to your computer if you don't want to use iCloud backups just like it was an iPod. You can synchronize your Photos library and avoid iCloud storage fees, deleting synchronized photos from your phone to free up space to take new photos and videos. It works just like you were using a digital camera in 2005. Yep, you can still rip and burn CDs!
Furthermore, the way Apple moved device synchronization functions to Finder and split out Music from Podcasts and Audiobooks is helpful for organizing the whole process. It used to be that iTunes was the home for all this synchronizing of non-music-related content, but now it more sensibly exists in Finder.
I think a lot of people don't realize that Apple basically still allows you to send over personally owned non-DRMed or even pirated content to Apple's own modern apps very easily this way, you just have to be willing to synchronize using "the old way" like your iPhone is an iPod. They've even kept ancient hosted services like iTunes Match going just in case you still need that sort of thing (it essentially allows you to sync music to your iPhone that is either pirated or not part of a known label music catalog via a cloud service rather than having to do a local sync via cable or WiFi).
And this workflow is very simple for non-technical users who don't really know how to traverse complicated file management structures. Yes, I would really like if apps like Photos was more flexible on file management, but on the other hand if you follow the prescribed workflow the results are quite user friendly for someone who really doesn't want the cloud but also can't handle setting up a home NAS. In this use case you have a reasonable photo storage system by syncing your device and then backing up your computer in a relatively hands-off manner using Time Machine.
One final point here is that Apple Music the subscription service can be hidden entirely from the app. Apple will just give you a 100% owned music jukebox app. Google doesn't do that, and with Microsoft you're probably using a legacy app like Windows Media Player that looks like it belongs on Windows Vista.
You’ve referred to iTunes in the present tense when it hasn’t existed for years, refer to ‘Apple fans’ as some sort of completely separate group of clearly defined people, and spend an unjustifiable amount of your comment talking about your quite niche professional Windows backwards compatibility use case.
We don’t need to rehash this whole thing. Please. Don’t take all the oxygen out of the room.
This depends on where (which country) you live. For all the ways Apple has been vocal about the Indian market and local production, Apple Maps literally sucks even in major cities in India. Google Maps is decades ahead and gets updated very quickly. Apple Maps cannot even find regular addresses or places.
Apple has its share of incompetencies and willful blind spots, and that shows up in specific areas often related to its services (Apple Intelligence is also a service). The organization and its people are not built for handling these effectively or quickly.
That said, I have more hope in Apple Intelligence improving quicker (at least in English, while competitors are already ahead in other languages, including several Indian languages) than I have in Apple Maps improving in India.
The public transport part of Maps got much better in Germany, though.
A smart assistant, that can understand and speak to me like Advanced Voice Mode, use a vast knowledge database, is tailored to my needs and can act on my behalf.
And it would be great if it’s able to run locally.
Gemini's home automation works nicely and it can understand comments like it's too dark in here or it's cold inside and act appropriately. This is using the Android app as an assistant, not live mode.
OpenAI's implementation is apparently similar but I haven't tried the voice mode as a free user.
I haven't tried Apple Intelligence yet on my M1 and don't have an iPhone, so I can't compare.
I've been looking at offline capabilities with open weight models but they aren't there either. A full speech-to-speech model [1] working on an M1 Mac would be incredible.
I don’t think any company has a smart assistant that’s reliable enough to act on your behalf except for some very constrained tasks (examples: dish washers, auto-parking cars)
I think it is more complicated than that. I think the Apple Vision is a kind of albatross. No one wanted this thing. I happen to think the executives didn't want it either. For all the years and effort put into it (and, well, there was project "Titan" before that) killing it might have hurt worse than their lackluster shipping of it.
Flush with cash (and I can't think of a phrase that really carries the weight of just how flush with cash they are — embarrassingly wealthy?) it was a rounding error for Apple to hire everyone they could in The Valley and keep them busy (and filing patent applications as they worked). It kept them from the competitors.
And I don't believe you could have instead put the engineering hires to "fixing Maps" or whatever pet peeve you and I have about the current Apple ecosystem. You're 1) likely not hiring the type of engineers for those tasks and 2) just throwing more people on the thing is not necessarily going to be the right answer (The Mythical Man-Month, too many cooks (ha ha) and all that).
On the whole I think Tim has steered the Apple ship to align with the times we have been living in.
Apple Maps provides me with more accessible info. e.g. "turn right at the next traffic light", "stay in the second lane from the left" vs. "In 200 metres, turn right onto 1st Avenue" (where it's always off by 50m) and nothing about lanes
These type of rare but common enough edge cases make me super hesitant to use it in the Midwest.
Google maps won’t even work properly when there’s no data.
The very initial success of Microsoft was that everything was reliably mediocre. Most things Microsoft delivered that were truely bad were fixed within a few major versions. It was a superpower.
The same model works for most purchases on a bad|average|best spectrum: we never want to buy bad, best is difficult to buy, so we settle for average quality.
Aside: I think MS has gone downhill and is now bad on multiple dimensions for me
The problem is that this risks antagonising the everyone in arts/humanities, and most other use cases are really unneeded - who needs text summarizing for something as simple as personal texts from friends? casual use is not really complex enough to warrant an assistance.
I don’t anticipate this being a problem. Have you used generative fill in photoshop or lightroom? It’s a complete game changer. In Egyptian mythology they weigh your soul against a feather when you entered the afterlife, and with professional tools I think moral hangups about AI are going to get about the same weight. It’s just too good not to use.
Apple already has simpler tools aimed at non professionals, they don't need generative AI either.
It was actually quite common Apple during the days Steve Jobs was no longer at the helm, they weren't even able to create a new OS, had to buy another company to rescue them.
And had they gone with Be instead of NEXT, most probably we would be talking about Apple in the past tense nowadays.
Nowadays they might have more money than ever, but it won't last forever if they cannot do anything else than reboots of existing products.
If no one is good enough, does it really matter who's the worst?
I've been really enjoying the AI notification summaries, they're a nice combination of time saving and comedy
I chat with GPT (especially in the car) to get things done; assistant and a knowledgebase. Siri makes me have nerd rage (lol) trying to use her the same.
If GPT came out with an AI Phone Apple would be out of my life. I want an AI Phone where on the lock screen I see a facetime like call with my AI Assistant (can skin how they look to be whoever). They do everything for me via voice, text, hand gestures, facial expressions and etc. It would be less icon focus and way more AI focused of a UX.
I agree Siri isn't good, but adding good AI into the existing ecosystem is clearly where the market is headed, and I don't think it will be long before Apple gets there.
The problem with interpreting AI through that lens is that AI, as it is being used here, is not an extension of your mind. Plenty of other things are (organizers for example), but AI does not extend your thoughts. It replaces them. Its notification summary feature does not improve your ability to quickly digest lots of notification information, it replaces it with its own attempt, which, not being your own judgment, can and does easily err.
There are some uses of AI that do act more like a McLuhanesque medium. Some copilot applications, in which suggestions are presented that a user accepts and refines them, are examples of this. But a lot of the uses of both image generation and LLM tools serve to limit what your mind does rather than expand it.
It's also worth noting that Apple traditionally is not a first mover and looks for "inspiration" from smaller competitors. In this case, there is no comp to reference. There is no startup mobile OS innovating in integrated AI. That, and the supposedly rushed timetable, probably explains a lot.
The lengths Apple went to build a secure and private system will make it stand out and help it hold up to regulatory scrutiny. Doing this now is better than doing it later.
In, let's say, 3 years, the features are more legitimately useful, the gatekeeping on new hardware will be a non-issue. In 3 years the majority of Apple's users will have an iPhone 15 Pro/iPhone 16 or newer. They are probably already mostly on M1 or newer Macs.
On the other hand, I totally agree that it's pretty useless as it stands right now. I also think that if their launch strategy is to have an amazing WWDC and then deliver 5-10% of the features in September, that's going to turn Apple into just another company that promises the moon and delivers gimmicks.
We have seen that play out with all the previous AI chips (mostly Google).
The software and its requirements are moving faster than devices can be shipped and accumulate significant marketshare.
This is not to say thay Google and Apple haven't or won't be able to ship some minor models such as voice recognition or translation, but for frontier level AI the local chips just won't suffice.
FALSE. Apple defines a photo as a record of something that actually happened. iPhones take photos. They doen't auto-swap a high-res moon in for the real one like Samsung phones do.
Clean Up (like crop) is just an editing feature, manually applied after a photo already exists, and using it effectively changes the image from a photo into an "edited image", the same way using Photoshop does.
Definitions of What a Photo Is:
Apple - "Here’s our view of what a photograph is. The way we like to think of it is that it’s a personal celebration of something that really, actually happened. Whether that’s a simple thing like a fancy cup of coffee that’s got some cool design on it, all the way through to my kid’s first steps, or my parents’ last breath, It’s something that really happened. It’s something that is a marker in my life, and it’s something that deserves to be celebrated." - John McCormack, VP of Camera Software Engineering @ Apple
Samsung - "Actually, there is no such thing as a real picture. As soon as you have sensors to capture something, you reproduce [what you’re seeing], and it doesn’t mean anything. There is no real picture. You can try to define a real picture by saying, ‘I took that picture’, but if you used AI to optimize the zoom, the autofocus, the scene — is it real? Or is it all filters? There is no real picture, full stop." - Patrick Chomet, Executive VP of Customer Experience @ Samsung
Google - "It’s about what you’re remembering,” he says. “When you define a memory as that there is a fallibility to it: You could have a true and perfect representation of a moment that felt completely fake and completely wrong. What some of these edits do is help you create the moment that is the way you remember it, that’s authentic to your memory and to the greater context, but maybe isn’t authentic to a particular millisecond." - Isaac Reynolds, Product Manager for Pixel Cameras @ Google
Definitions via https://www.theverge.com/2024/9/23/24252231/lets-compare-app...
> the Clean Up tool gives users a way to remove distracting elements while staying true to the moment as they intended to capture it.
idk, seems like the author described it the exact same way Apple does in their marketing copy.
https://www.apple.com/newsroom/2024/10/apple-intelligence-is...
It's one thing to use masks to edit highlights/shadows/color balance for certain areas (skies, buildings, people, etc) but it's an entirely different thing to completely replace the sky, or remove objects because they aren't "appealing"
Almost as long as we've had photos, we've been removing "unappealing" things from them. Famously Stalin had Nikolai Yezhov removed from a photo after he was "purged", but the Soviet Union in general is full of these instances.
More lightheartedly, Disney supposedly (though this seems to be subject to some debate) has airbrushed a number of photographs of Walt Disney to remove cigarettes from them. And perhaps most famously of all, Han only shot first if you were born before 1997.
Image quality on modern phones is in large part due to a lot of image processing done by the phone. Multiple photos being taken, combined for least blur and best dynamic range, colour balanced to best represent skin tones etc.
The line between the sort of algorithms that run on an iPhone and inserting a moon is largely philosophical rather than technical. It's an extremely important philosophical line! But the sort of things that have been added are the logical continuation of the sort of work that the camera teams have been doing.
[For context, I'm a Google Pixel owner, but of those three statements the Apple one is the one I agree the most with]
Easily the best camera app I've ever purchased. It makes me not want to pull out my mirrorless camera as much to get decent photos.
You can toggle "pro raw" in the default camera app. It captures a lot more information from the sensor. The files are a lot larger, it isn't throwing away information that isn't visible like in the shadows. This gives you more flexibility like changing color balance and exposure after the fact because of that extra data. But there is still some sharpening and post processing.
You can use the camera app inside lightroom or "procamera" or other apps and take raw photos, where it records all of the sensor data without any post processing. Most people don't want this, you need to develop the images using software like lightroom to look good.
An Apple marketing executive is smiling somewhere. Brainwashed another one!
(And outside of work, Apple punted on games, which means I will always buy a Windows or Linux computer.)
I recently tried Linux as my primary desktop after 10+ years. It’s amazing. It just works and there is no slop, no advertising, no ecosystem to log into, no notifications. So nice. I can actually get work done! And games often just work.
(Pop_OS on a System76 Adder)
And people always say that we risk our accounts being locked out, but when was the last time Apple was heard closing Apple Accounts?
Instead, all of these companies are effectively forced to play hype ball.
I disagree that Apple should have sat on their hands. LLMs are already shipping apps that are beginning to integrate with the OS (looking at you, ChatGPT icon in my MacOS menu bar). How much user privacy should Apple allow me (well, their customers generally) to cede before they step in?
Now they've gone and doubled[0] the recommended space AI needs on base model phones shipping with under spec'd storage in the first place. If it goes up any more you'll be looking at giving up nearly 10% of a base model 128gb phone's storage to AI.
[0]https://9to5mac.com/2025/01/03/apple-intelligence-now-requir...
The LLM hype is so powerful that it got people to think the market leader in consumer-facing AI features was falling behind because you couldn't ask your iPhone to summarize notifications poorly.
It's that sentiment all over again. Specifically, a naive belief that the specs/tech of something are just as important than the execution. Maybe they are if you are very tech online, but for most people, tech/specs don't matter as much.
Ultimately, this is another distortion of reality by the comment section.
Believe it or not, there are paradigm shifts in technology every now and then. Not everything is 3D television.
You mean like is documented on https://support.apple.com/guide/iphone/apple-intelligence-an...?
I wish they had but on the other hand, can you imagine how many think pieces there'd be about how Apple was stagnating and how they were "such, like, a 20th Century Company(tm)" and you'd probably get activist investors bleating about how Apple were leaving money on the table and and and etc.
Really loved this article overall, but I have to super-disagree here. The core of ChatGPT is you can have a conversation with a computer program.
Take away saving history and you can still have a conversation with a computer program (see ephemeral chats).
Take away typing one word at a time and you still have a conversation with a computer program (see non-streaming API / batch API).
But major props for writing this live on a twitch stream, benefit of the doubt there my friend.
I can’t help but wonder if the reason “agentic” systems seem so appealing to people is because as an industry we’ve spent the past fifteen years making software harder to use.
I have the latest Google Pixel, and was using Google Maps to navigate.
I pressed the "voice search" button from within Google Maps and said "What is the name of the mountain on the left that I'm about to drive past?"
Instead of a context-aware answer, my phone simply did a Google search for that exact phrase and showed the results to me. The top hit was a Reddit thread about some mountain near Seattle. :)
I guess we're entering an age where people might have a reasonable expectation that any app is a context-aware LLM, but personally I don't have that assumption yet.
It also depends on the context of the action. If you’re sitting in front of your computer, yeah, no big deal. Type “podcast” in the search field of the messages app and click the thing it finds. But if you’re busy cooking dinner or cleaning out the cat box, it’s a pain to get your phone out and poke around. The main draw of voice assistants (at least to me) is that they let you do things quickly when you don’t already have a device in your hands.
MacWrite was released 5 years after WordPerfect, which itself is predated by WordStar. I don't get why Apple fans have this obsession with pretending Apple invents these things.
Apple refines what others have attempted before, that's what they're good at. Part of the reason people are disappointed with Apple these days is because of this fantasy image of Apple as an inventor.
I don’t think cross-platform UI code really took off on the Mac until maybe 20 years later. Plenty of Mac apps were built that way before that point, but they tended to be ones in the “you’re stuck using this, so it doesn’t have to be very nice” category, like Word ended up being. It doesn’t really seem to have tipped until Electron came along, and somehow web apps that use half a gigabyte of RAM to show some text became totally accepted as good enough.
Incidentally, the Cocoa UI framework that macOS uses was originally a cross platform framework that could deploy to Windows and various other OSes, back when it was made by NeXT. Apple. Apple killed that off and it became Mac-exclusive. I wonder what the world would look like if they kept support for other OSes. Maybe we’d have a good selection of cross platform apps that actually look nice and perform well.
I mean, look, WordStar was a huge step up from a typewriter. It (and programs like it) made it possible for people like me to write.
WordStar let you get the words right. WordPerfect let you do "you asked for it, you got it" layout, which was a step up. But MacWrite and programs like it let you do WYSIWYG layout, which was huge. It was like going from typewriter to WordStar, but for layout and appearance. (The words are still more important, but the presentation also matters.)
WordStar and WordPerfect were for DOS. They were not WYSIWYG. Sure they were powerful word processors for professionals, but they were not “like” MacWrite. MacWrite was a tool for regular people.
I don't think I'm alone in that judgement, as Wiki says:
"WordStar was the first microcomputer word processor to offer mail merge and textual WYSIWYG." [1]
So... You might need to expand why you think that this is not true.
[0] https://cdn.arstechnica.net/wp-content/uploads/2017/03/words...
You said it.
WordStar did not have fonts. MacWrite did.
WYSIWYG without fonts is ... I don't know ... "textual WYSIWYG"? Whatever that is.
This is what enables you to do proper typesetting and page layout for a document, and using a PostScript printer such as the Apple LaserWriter you could do desktop publishing [1]. Desktop publishing was invented by Xerox PARC but the revolution began when Apple made it available to the masses with the Macintosh and LaserWriter.
Apple didn't invent any of these technologies but they were the first to put them all together into a package for the mass market and made them incredibly easy to use. Suddenly, grandma had a tool she could use to write and typeset the weekly church newsletter from home, and even print it on her LaserWriter at home. If she wanted to do a newsletter like that just two years prior she would have had to hire the services of a print shop to do both typesetting and layout as well as printing.
She could still have used WordStar or WordPerfect and printed with a dot matrix printer, but that doesn't get you large, proportional fonts or layout.
WordPerfect for Windows was released in 1991. Prior versions were MS-DOS and not graphical.
WordStar for Windows was released in about 1991. Prior versions were MS-DOS and not graphical.
What made MacWrite special, is that it was graphical, highly WYSIWYG (especially when paired with an Apple printer), had a lot of great fonts, and the software was intuitive.
Of course Apple didn't invent any of it, but they made one of the best word processing products at the time.
MacWrite was the best in a niche market of personal newsletters that wanted graphical elements. More professional desktop publishers used Aldus PageMaker. And big publishing houses didn't use PCs or software, they used offset lithography presses.
Microsoft Word for Mac was released in 1985. It was comparable to MacWrite.
Also Microsoft Write for Windows was released in 1985, but it was clunkier.
You realize there were other systems besides PC and Mac?
Signum for the Atari ST came out in 1986. It was a fully fledged WYSIWYG text processor with special printer drivers for regular dot matrix printers. Even with a 9-pin you could create great looking output, if you had the patience (a single page took minutes to print). Signum was way ahead of MacWrite and was very popular with people needing special fonts in science/math and the humanities (you could quite easily design your own fonts). Also, it allowed for Right-to-left text, and of course the Atari ST was way cheaper than the Mac.
Claiming that Apple does not invent is as gobsmackingly wrong as claiming that Apple invented everything.
For example, you could claim that nothing new in CMOS manufacturing exists because it’s all just the existing idea of a transistor. Or the transistor is just a quantum mechanic version of the vacuum tube. Or the vacuum tube just an electric version of the Babbage machine. Repeat for Internet vs packet switching.
Basically come up with a definition that doesn’t require a “I know it when I see it step” and I’ll easily fit many things Apple did in there unless it’s such a restrictive definition that no one invents anything.
“If I have seen futher, it is by standing on the shoulders of giants”
This is entirely tangential and probably a pointless gripe in this thread, but...
For some reason, it's always really annoyed me that Apple took MP3 players, called them an 'iPod' and suddenly everyone ate them up like they were the second coming of christ and we'd never had them ever before.
To be clear, there were good non-Apple MP3 players, but they were either marketed poorly, or late arrivals (e.g. the Toshiba player that got rebadged into the Zune). By the time those existed (and tech companies started hiring UI/UX people), Apple was doing a complete reset of another product category: smartphones.
I suspect history would have been different had, say, MiniDisc hadn't failed horribly in America[0]. Pre-iPod, portable music in the US was either compact cassettes with all the downsides of tape, or CD players that could just barely fit in your pocket. The iPod was such a step up from either that it all but became a genericized trademark. Had we had a competing technology from not the 1980s, we probably wouldn't have thought the iPod was so great. Or at least, people I knew who had MiniDisc looked at the iPod like I look at all the e-waste that was trying to compete with the iPod.
[0] Yes, I know that Sony was basically trying to avoid a repeat of DAT getting banned
The iPod cam with a large screen and a click wheel. I could find songs on it. That was a revolution for me.
MP3 was the enabling technology (if you can't fit many songs on a small device, then this is moot).
Flash based players were smaller but limited in size and expansion media was expensive. Hard drive players were hobbled with USB 1.1 connections and an obsession with drag and drop for management.
The iPod by default just synched with your iTunes library. The FireWire (and eventually USB 2.0) did so quickly. The navigation was as good as the metadata which iTunes made easy to edit. The UX on the device made scrolling through long lists of songs very easy.
The iPod made using an MP3 easy and approachable for normal people. The Rio, Nomad, and a multitude of others did not. They included a bunch of checklist features but didn't focus on usability until Apple dominated the market.
Not to say MacWrite wasn't based on prior work (it was, though not actual products that I know of), but it isn't really comparable to prior text-based word processors.
I worked in a lab that used PCs while I owned a Mac Plus. There was no comparison.
Calling what they do as mere refinement may be equally understating it in some cases.
They get it right. For the masses. They create beginners.
Anyone can start with an Apple, because it’s what it’s designed to do.
I have my own biases that extreme usability started earlier with other movements like WebOS and Palm contributing to it as well.
Still, if we look for watershed moments where huge numbers of people in the mainstream adopt technology, whether it’s the iPhone, iPad, Apple TV, watch, laptops, they don’t need to be the first, just the best for the most number of people.
Being able to integrate hardware and software closely creates a different and reliable result for the many, as much as I might not like having complete agency.
If anything, Apple helps invent beginners in the mainstream.
I used to think that about Apple which is why I got the iphone 4s when it came out assuming that I would be able to use it via voice while driving and have a good experience. I disabled it after the first day and never went back. It was not remotely ready for a wide release. I still use a mac as my main device, but I've been on Android since the 4s.
Im referring to the general trend of adoption of smartphones being a certain way with Apple.
Like others, I’ve spent time back and forth between iPhone and Android while using Macs more.
Android dialled in was pretty comfortable for me, iPhone was a lot more taps and swipes for the same, but the vertical integration is harder to ignore.
Happens all the time with the iPhone. Apple gets celebrated as an innovator for adding features that have existed for years on Android.
Um, hello, Wang OIS? WordStar and WordPerfect didn't invent anything. They were copies of terminal-based word processors.
But MacWrite was different in two important ways. First, like Bravo and Gypsy before it, it was WYSIWYG, a million times better than WordStar/WordPerfect. And it worked with the LaserWriter. But more importantly: it was free. This made MacWrite revolutionary.
https://en.wikipedia.org/wiki/Wang_Laboratories
There's a certain power that comes when you can designate that something is a section header, or a title, or a paragraph, or an equation -- and that power is felt when you decide you want to change the style for all the titles and section headers, or change how you number and reference equations and theorems.
Word processors (to the best of my knowledge -- it's been years since I've used one) can't do this. This kind of power comes from markup languages, like LaTeX or HTML/CSS (if you organize your documents correctly).
Having said that, there are LaTeX editors that both provide a certain amount of "What You See Is What You Get" (I think one is called "LyX", and I think GUI Emacs provides this, too) but nonetheless allow you to use LaTeX directly when you need it; I guess this is equivalent to the "Show Codes" ability that WordPerfect provided. (I wouldn't know, though, because I prefer to use Vim to write my documents.)
Also, I would agree that when things become freely available, that can be a game-changer, but as someone who grew up with a terminal and a giant computer with 8-inch disks and an Atari 800XL, when a neighbor showed off his Macintosh, it would be out of the question for my family to think of that as "free".
In addition to Watson (https://en.wikipedia.org/wiki/Karelia_Watson), there's also Cover Flow (https://en.wikipedia.org/wiki/Cover_Flow), Shortcuts (https://en.wikipedia.org/wiki/Shortcuts_(Apple)), Konfabulator (https://en.wikipedia.org/wiki/Yahoo_Widgets), Growl (https://growl.github.io/growl/), and of course my favorite LaunchBar (https://www.obdev.at/products/launchbar/index.html, LaunchBar would be my pick as the most innovative app of the last 25 years), for apps that were incorporated into macOS.
While Excel, Photoshop, Illustrator, Sketch, Lightroom, Premiere, and PowerPoint are all examples of software developed and released first for the Mac that then went on to become software behemoths in their own right (too big to Sherlock). (Well Sketch turned out differently, because Figma happened, but it's still a great example of third-party innovation facilitated by "a set of software development tools, a platform, and an audience".)
The point here being I think of Apple as more providing a platform for innovation rather than innovating themselves (but I'm aware that's probably a minority opinion).
> the new standard that companies like Samsung and Google would clone the same way they cloned the hardware and software design of the iPhone.
That's misrepresenting history in ways it's not even funny.
Stopped reading at that point. The article took too long to even sketch it's main point anyway.
- No physical keyboard + touch keyboard
- Modern OS kernel (not embedded specific kernel)
- Desktop browser engine
- Capacitive touchscreen + finger instead of stylus - one or two phones had capTouch before, but they were far from standard, and they still had physical keyboards for typing
- Vertical by default orientation
- Short 1 day battery life in favour of more power/features (weird to list, but was a bold move everyone mocked then followed)
They totally took out the existing market (Blackberry, Windows Mobile, Symbian, a variety of OEM OSs). Android succeeded but came after, still had keyboards on its flagships the years after iPhone came out (G1, Droid), and took these design cues from iPhone.
The Mac GUI with mouse+keyboard+windows was also huge. Admittedly not first to invent it (Xerox PARC), but first to ship it as a package is still hugely impressive. Few people commercialize a new product before it existed in some lab.
They didn't clone everything, but they cloned a lot in the early days. Rounded corners was another one. Now it seems like Apple is cloning Google/Samsung more.
This is the famous "reality distortion field" you may have heard of. It's basically a form of tribalism taken to an absurd level.
Exactly. And often the first version is often kind of meh (iPhone, Watch, Vision Pro) but they keep iterating and later versions become really good. Sometimes it's a hit directly (M1), it's still very iterative on whatever came before.
Natural language as input doesn't give you any information about where the boundaries are or what's possible. Meanwhile natural language can express anything, most of which any current implementation won't be able to do.
So the user gets a blank canvas and all the associated problems with learning what to do, except it's worse because many things they think up will fail.
And the main tool we have to guide the user through this fraught path is LLM output. Oof.
I am hopeful that Apple will demonstrate their expertise in using some traditional UI to help alleviate some of these problems.
I’m not aware of any competition that’s perfect.
I guess you meant to say the competition is better. I think it is, but that doesn’t mean their product isn’t a huge improvement.
A simple example is text search in Photos.app. It probably misses text in some photos, but it helps me find quite a few photos. Similarly, face recognition in Photos.app is far from perfect, but way better than not having that feature.
> I want to be able to easily create calendar events and interact with other native iOS APIs
It’s not in Apple’s DNA to release a product here that mostly works. Chances are they’re working on something like that but don’t find it good enough to release it yet.
What surprises me, though, is that they released the “get an AI summary of this web page” feature. That definitely produces some results that are very bad.
(I get that not everybody bothers doing vaguely complex mental arithmetic, but dividing by two? Come on!)
In my view, Private Cloud Compute and Apple Intelligence, together with the ubiquity of Apple devices, position Apple as a leading candidate to realize the widespread, AI-enabled transformation of personal computing discussed in the post — with tasks requiring less energy and cognitive load than they do today for the general consumer.
steve jobs understood you have to do really hard things very well to do anything simple beautifully. I'm pretty sure most of their AI play is still in front of them.
> I want the data coming off of the sensor to be the data that makes up the image. I want to avoid as much processing as possible and I want the photo to be a reflection of reality as it is, not reality as it should have been. Sure, sometimes I'll do some color correction or cropping in post, but that doesn't change the content of the image, only its presentation.
First nit: the iPhone camera, and all digital cameras, are deeply influenced by computational photography techniques. What this means is that you essentially never get the raw pixel values, although there are exceptions. The image you get is already significantly manipulated.
Second nit: color correction, color in general, dynamic range, focus, depth of field, and more are all manipulations made by default, even long before digital cameras when film was king. There is no "correct" image version of what our eyes see, there is only pleasing to the photographer and the audience.
An example: the negative for Ansel Adams' well known "Moonrise Over Hernandez , New Mexico" looks like, at first glance, something a professional would trash for lacking detail.
Here's contact print vs the version most of use will probably recognize: https://images.squarespace-cdn.com/content/v1/5f5fe5ca8d6a35...
Here are four different versions Adams printed over the course of 3 decades: https://images.squarespace-cdn.com/content/v1/5f5fe5ca8d6a35...
I will mention, but won't even get into a topic that will surely bait HN commentors: Kodak designed and standardized its color film to represent Caucasian skin tones. It wasn't until chocolate and furniture makers complained that everything looked like the same gross mud in their expensively-produced product catalogs that Kodak took a look at rendering dark brown/red/yellow tones more pleasingly. Notice I said "more pleasingly", not "correctly".
That may be an urban legend. There was a popular reference card that had a light-skinned woman, but that wasn't the problem. This is closer to the old Technicolor vs. Eastmancolor debate.
Here's the trailer for Disney's "Song of the South" (1946) [1] That's three-strip Technicolor. Good dynamic range, with each of three colors on its own strip of black and while film. Here's the trailer from "Shaft" (1971) [2]. That's single-strip Eastmancolor. Dynamic range is not as good, as you can see in the street scenes where lighting wasn't controlled. Eastmancolor took over in the 1950s because the cameras are much smaller and production is easier and cheaper.
NTSC Color TV really did have something to standardize skin tones. NTSC color TV transmits an intensity value and two vector components which determine the color. The color vector components are rather low bandwidth (only about 10 full range color changes across the width of the tube), and so some NTSC receivers had a gimmick which, when the vector is near the "skin tone line" for a standard skin tone, pulled it to a fixed value.[3] The effect was that all faces had roughly the same skin tone in UV color space, but intensity could vary.
[1] https://www.youtube.com/watch?v=NxwqH47Ne70
[2] https://www.youtube.com/watch?v=pFlsufZj9Fg
[3] https://bramstout.nl/en/webbooks/vectorscopes/#yiq-and-the-s...
LLMs make different mistakes than I do so I've thought about using one as a copy editor but I've had terrible experiences with copy editors: I've hired more than one when I was writing marketing copy who injected more errors than they fixed. (A friend of mine wrote an article for The New York Times that got terribly mangled and barely made sense after the editors made it read like an NYT article.)
In the meantime, they’re shipping the best non-workstation computers, by far, to run models locally.
They don’t have to be the ones to implement all of this themselves, you can install ollama right now, and BoltAI can integrate those models into other parts of the OS. And Apple will watch, and Sherlock the best parts of what others do into the OS, and sell gobs of machines.
They haven’t squandered anything, the foundations are still there.
Parsing text for variables when it sees an equals sign and running basic calculations on them? I feel that could have been a novel feature 30 years ago.
The whole article feels like it suffers from a similar lack of coherence. Ex - I am hardly an apple fanboy (I strongly dislike the company) but the complaints here are basically
The service sometimes has outages
The image gen is not as customizable as he'd like
He's morally opposed to cleanup in photos
Notifications summaries are bad (and how dare I get my texts 5 seconds slower).
---
None of that is really related in any way to the security footprint of the tooling he discusses up front, and it's also hardly distinct from most other current AI offerings, and it's not really a consistent complaint about the tech.
My opinion of Apple is that they do a crappy job with the vast majority of their apps...
They build good hardware, and they abuse their small hardware footprint to make decent device experiences and a decent (but getting worse) OS - but their actual applications are generally mediocre at best (mediocre copies of a previously successful, usually better, app that they will put out of business through shady store practices if I'm being blunt).
---
If anything, the failure here is that Apple marketed a thing that AI can't really do (yet, maybe at all), and most of the things AI can do without being incredibly invasive aren't actually all that useful to most folks. Very useful to a handful of power users in specific circumstances, but otherwise essentially novelty apps.
So... it's not an implementation failure. It's a marketing failure. And this is hardly unique to Apple right now. The only difference is that usually Apple doesn't play their hand until this inflection point with new tech is over, so it's more obvious this time around just how bad the product fit is for general use.
Most of the intro to this is credulous hooey. Macs weren't "bicycles for the mind" in some magic way that was different from PCs. What the early ones had was 1) a better and much more standardized interface, and 2) task switching that worked.
As for the AI tools, image generation might occasionally be useful for a D&D game, but otherwise nothing on offer at the moment has much value. And the value of image generation (for me) is pretty small.
No, it's not. The sensor in an iPhone is AI/ML'd up the ass to hide all the noise because it has 1µm sensor wells.
A Panasonic video-oriented mirrorless micro 4/3rds (so not even anywhere near 35mm) like the GH5 is 3-4x that.
A Sony Alpha 7 III? six times the sensor well size.
I don't care how many megabits of video bandwidth you throw at it or how fancy you think "raw" shooting is, or how fancy your sensor technology is; nobody these days has anything that is even close to 2x better than anyone else. The top sensor from all the major players are pushing the limits of physics, and have been for a long time.
No amount of AI/ML shit will give you depth of field and bokeh that looks as nice as a big sensor and a fast lens with nice shutter leaf shape.
Original prompt to Image Playground: "Enron logo" with the Enron logo as img2img input https://bsky.app/profile/yasomi.xeiaso.net/post/3lf472e2yfc2...
His first instinct was right. It seems impossible, because it is. Unless I can run the entirety of the "Private Cloud Compute" on my own hardware in my own firewalled network, I 100% believe that the pipeline is compromised; our data is siphoned off and sold off to advertisers, especially now that they know they can do it and get less than slap on the wrist: https://news.ycombinator.com/item?id=42578929
This is sardonic, as yes, Apple could have chosen different monopolies than it currently has, at different points, and had a different (maybe not better?) trajectory, some of of us are old enough to remember the antipathy towards Microsoft when to the Office suite it added default Explorer,
But also, maybe our system fundamentally rewards the "wrong" things if one's definition of "right" includes things like innovation. Or maybe, the welfare of the commons and the common good.
I’m hoping with what they’ve built in infrastructure and custom chips is a step towards making personal LLM also highly available to non-technical people. I think this is where Apple has always shined - making things not just better, but accessible and grokable for normal people.
Generative AI is a dud. ML has a lot of applications. I use my smartphone for calls, chat and banking apps. I use my iPad only for drawing, mail and some games.
For everything else, I have computers. With real OS.
Sorry, what?
Apart from the level of dream-detail recalled being highly dubious, quoting your own hallucination of Steve Jobs to help with your argument about generative AI being useless (and missing the irony) is downright weird.
Also Math notes is basically the same thing search engines have been able to do for over a decade now. Enter a sum, get an answer.
I figured you'd rather read something my brain made up (albeit unconsciously) than something a machine made up using linear algebra without understanding any of the words that it's using.
Also, it's almost exactly what Solver has done for years.
Ultimately Apple’s strategy of a privacy focused AI will be a winner for a consumer device with access to sensitive personal information. It’s a question of whether they can pull it off technically.
Having said that, I actually paid attention to the image playground criticism. Image playground is literally a playground. It is meant to make fun, low-effort images for friends and family, largely for social type interactions.
"It uses a placid corporate artstyle and communicates nothing." It's a hot taco holding a beer. What is it SUPPOSED to communicate? Looks like a pretty great image to me. But of course this piece was leading into the anti- angle, so suddenly it's "horrifying". I guess I didn't get the special training to understand what was wrong with a clearly lighthearted, fun image.
Similar asinine, overly-jaded complaints about the cartoonish, memoji style portrait generation. I think the image is actually pretty hilarious. Actually used image playground to make my social media image, and I care not what this guy thinks about it, or that it is "soulless" (as if a cartoonish representation is supposed to be soulful?)
That is basically the entire point of a cartoon, so yeah
* The taco holding the beer glass to its lips and sipping on it to smoke it
* Consistent eye shapes (likely the eyes would be closed to smoke the beer)
* Better bokeh for the elements in the background (if that is the stylistic choice I'd go for)
* Have the smoke coming out of the beer, not out of the taco "taco smoking beer"
* Stylize the image such that it has individual flair, there's something about the Apple Intelligence artstyle that just has an unperson corporate vibe that I don't like.
* The levity comes from "hey tacos don't have faces, hooves, arms, or legs and you can't smoke a beer", this would be used to communicate absurdism https://en.wikipedia.org/wiki/Absurdism, specifically by means of the taco smoking a beer whilst holding it in its hooves
Maybe I've just been exposed to way more AI imagery than you, but guacamole does not look like that in the image. There's more fever dream images that I have locally, but I didn't want to saturate my article with them and haven't fully implemented "image gallery" support yet.
And yes, a cartoon is normally meant to communicate something, quite literally the definition of soulful. Look at this for example: https://bsky.app/profile/yasomi.xeiaso.net/post/3ldgzieehjc2... When I made it, I was trying to communicate a lo-fi peaceful vibe accentuated through traditional artstyles. In a more finished piece I'd probably recreate this through watercolor in Procreate and apply a bokeh effect (emulating the depth of field for a subject with forward light being looked at from an 85mm portrait lens at about f/2.8).
The point of art is to communicate something. If a work does not communicate something, it is categorically not art.
What they released as Apple Intelligence wasn’t a well-planned, cohesive product as much as the only thing they could possibly do, given the timelines they were up against. Maybe they’ll catch up, but they’re definitely behind and it’s a shocking thing to behold.
But the demands of intelligence and the general trajectory means no amount of hardware - storage, RAM or battery size would be enough to generate the high fidelity experiences or solutions that fans and customers have come to expect from the company.
They are dedicating “AI” hardware to their current devices, and are building a framework for many different futures. In regards to software, they are first building the safety, then MVP’s to provide APIs to their chips.
So yah, maybe they aren’t holding the holy grail at the moment. At least we won’t be drinking from lead goblets .
They have set themselves up for a loser in the next year or two, because they can’t double their resources to catch back up to a normal release schedule.
Didn’t realize how widespread that type of spam was until now. Why hasn’t someone implemented better spam detection at Apple like we have for email? It would be nice if they could classify texts as spam, promotions, etc and organize them the way Gmail does.
Because that’s a laughably false premise.
"Apple Intelligence" is less than a year old. Give it some time, for chripe's sakes.
Apple has done such a good job with marketing that my 10 year old thinks that AI stands for Apple Intelligence.
We live in the Bay Area so she's seen a bunch of billboards with that. I have to constantly remind her that that is not what it means in most cases.
The Cursor editor literally has this.
In the end, even if the features aren't perfect, they still raise the bar for competitors, so Apple is less in danger of being disrupted.
Also, there is plenty of AI driven features that people do not talk about, but those "just work" so you don't see them as well.
Honestly I'm not sure that they do. Everything I've seen with notification summaries for instance has given me the feelings of "wow I guess I'm really not missing much". LLMs as an answer engine has a big benefit that it feels fast and fluid even when its wrong (and many won't bother verifying) but with notification summaries most users in messaging contexts will eventually go back to the conversation and see the responses in full detail. Mistakes in that context are identifiable by mainstream users as having made the product worse.
Having used "smart devices" since the Apple Newton 2.0 days, followed by Windows Mobile, a very brief Android excursion (Motorola Milestone - early enough Android that I was often frustrated trying to copy/paste text between apps), then another side-pivot into Windows Phone for awhile (mainly because the development was incredibly easy - and Microsoft gave me a free one), I have been in the iOS mobile phone ecosystem ever since the iPhone 6.
And - the software has gotten increasingly better over time - I wouldn't have (for me) alot of content/subscribers on TikTok, if iMovie on my phone did not exist - attempting to edit videos using OpenShot was taking forever (While I have Davinci Resolve installed, it seems "daunting" for someone who doesn't want to be a professional videographer/editor).
But then I tried iMovie "Magic Movie" on my phone and ... "it just works". Still not great for long-form YouTube style content, but for quick things, slice-of-life videos - it does the job rather well.
... I expect that Apple will improve these AI offerings dramatically over the next couple of years as people upgrade their devices.
e.g. https://imgur.com/a/1Y7HviK - what rules govern this half-arsed speckling of road names?
(Apple Originals, their production house, is also an example. A huge bank of original prestige TV, subsidized by iPhones... they're just still finding a way to market it.)
Has it improved that much? Does anyone remember what I'm talking about?
I mean you can hack it the same way you would hack any other Darwin platform
By which I think he means the AI stuff runs on your machine rather than the cloud. For me that's not a holy grail at all or even something I'm terribly interested in. I downloaded Apple AI on the macbook, found it quite meh and am now seeing if there is a way to remove it as it uses quite a lot of GB or memory. I can see for someone wanting to use LLMs on confidential corporate data that would be important but that's a specialist use case that I don't think Apple Intelligence is particually good for.
Before the AI craze you could search Photos on iOS based on content and metadata. You could lift subjects off photos with a long tap and copy them, recognize faces and make montages based on inferred relationship with them. And all of this is done on-prem, on your device.
These are very subtle, nice features. Apple had to put a name on all these otherwise there would be no marketing material.
I remember reading it was a post-Jobs transition thing, where the key message about products transitioned away from vibes and overarching slogans ("shuffle", "the internet comuter"), to features ("iPhone X", "iPhone XS", "iPhone XR", "iPhone XS Pro (??)").
I'm sure that a lot of the old-guard exec team knows it's a loss... I'm sure how they feel about it or why.
https://simson.net/ref/sbook5/
(and the source of which is available)
Given the lack of a single good supporting example (what, did PCs have no word processors that reacted to backspace keys?) it seems like these are fantasy bicycles...
And since no evidence is needed to believe, you can of course believe in Intelligence that can act better than your brain (what are "those pics from San Francisco", you've snapped a hundred there, which 5 would you like to post?)
And yet the disillusionment comes a bit faster than expected, why not give Apple a few more decades to iron out some kinks on such a revolutionary fantasy path?
How is apple intelligence going to help me with that ?
LLM are built on data, and copious amounts of it. Apple has been on a decade long marketing campaign to make data radio active. It has now permeated the culture so much so that, Apple CANNOT build a proprietary, world-class AI product without compromising on their outspoken positions.
It is a losing battle because the more apple wants to do it, the users are gonna punish them and meanwhile, other companies (ChatGPT, anthropic) are gonna extract maximum value.
(Meta can probably train on private data but OpenAI and Anthropic seem to be doing OK without it as far as we know.)
So that means all of the big tech companies are going down: Facebook, Google, Apple.
Only Microsoft remains strong though for how much longer remains to be seen.
A great time for startups.
But yes, it's a great time for startups. I'd argue it always is and always has been.
I think they're all doing pretty well at the moment.
They don't want to scare any part of their audience away from future uses of Apple intelligence. Their audience is tech and non-tech folk alike.
If the tech folk say it's safe and the non-tech folk get comfortable with the basic AI features then they're onto a winner.
How many people's parents/grand parents have iPhones because they're simpler for them to understand who are also scared or don't understand this 'AI thing'. I think Apple have been quite savvy in introducing it slowly and are probably watching the metrics like a hawk!
I suspect image playground is so creepy in an attempt to mark the images as clearly AI generated when they get posted to social media?