Vision Pro is an amazing product looking for a purpose and a market. I've opined that it's the 128K Mac of the 2020s -- an expensive toy of very little practical use on its own, that portends a revolution in computing, later versions of which really do change everything. I don't know if that's true anymore. It isn't the 1980s, and initiatives that don't make money over their first year or so tend to get axed. There's no Jobs to push the "vision thing" (no pun intended) anymore.
The Vision/Vision Pro product was also years behind schedule, based on original expectations, and some people feel that they had to launch something so that's why we got what we got.
We have to have thin glasses or contacts. No one wants to wear a bulky helmet thing. Until that happens it will forever be a hardcore enterprise or niche consumer product
I would say it's more like the internal Xerox machines that were built in the mid-1970s at PARC. Most people have no idea how to work with those machines. By the time the 128K Mac came out, people understood what a GUI interface was and knew how to use them - it's just that the Mac in particular was limited by it's hardware.
I wonder if there’s any “doers” in that top 100? Or is it all just “discussers” and “deciders”.
I’ve owned Apple products forever. But as long as Pournelles Law of Bureaucracy remains on full display, these shuffling of the titanic deck chairs will continue.
I wish it could be otherwise. I wish a company could introspect on itself and say “hey, our doer/decider balance is out of whack, let’s correct” but that just never seems to happen.
> Jobs was kind of infamous for making people put their app / product in front of him while he sat in silence and used it. Harder to fake.
Yes, faking that really well is basically equivalent to actually doing it for real.
Of course, there are many good and worthwhile things an exec could have done that wouldn't show up in such a demo.
Eg if you increase reliability of your cloud services from 99.9% to 99.9999% that could be huge, but most likely wouldn't show up in a short demo.
On the bullshitting front: I remember a particular re-organisation we went through at Goldman Sachs that the local bigwig was explaining to us, and all the benefits it would bring. I made the perhaps unwise decision to ask what observable measurements we could take in a few months to tell us whether the whole thing was a failure (or success).
(I suspect the actual main purpose of these semi-regular re-orgs is to shake things up enough so that after a few shuffles person A can eventually land ahead of his former boss B, without anyone ever losing face. And that's a good thing! But hard to admit to.)
A CEO just needs to make sure that stuff gets done; doing it personally is just one of the ways to get there. (And not a particularly scalable one.)
For Apple it probably made sense, because the whole company's image and reputation is built on that stick.
But eg for a toilet paper manufacturer or a producer of fighter jets, the CEO shouldn't spend too much time personally testing the products.
For the former, because presumably the product doesn't change that often.
For the latter, because there's not even a single 'user' of the product. The experience of the huge ground crew (with various specialised roles) is just as important as the experience of the guy in the cockpit, etc. No CEO, and actually no single person on earth, has the expertise to judge all of these aspects of use by themselves.
That doesn't mean the experience of the users is irrelevant. Just the opposite! But the CEO will have to intelligently delegate.
What do you mean by 'doing' rather than 'deciding' especially at the c-suite level ?
I thought that's what c-suites are paid to do - 'strategy' and 'capital allocation'.
Any example company you have in mind that has leadership actively 'doing' ?
I don't know why anyone believes that the push for Apple Intelligence was driven by middle management. Sure, engineers could have pushed back because they knew more about the limitations of the tech, but engineers aren't one to understand the macroeconomics driving industry-wide demand and long-term growth.
In this context, a "doer" might commit to an agenda, making ongoing decisions that furthered accomplishment and success on that agenda. While their nominal role is to decide, the decisions they make are organized to effect some end.
In contrast, a "discusser" or "decider" makes decisions in order to satisfy the social role of making decisions, but often with a lack of surety, clarity, follow-through or commitment. Perhaps in fear of missing some greater opportunity, or fear of being credited with some failure, their decisions are not organized in a way that actually effects some end.
Well, that is the problem with them making Siri useful for everyone - it is made for discussers and deciders.
Summarise a meeting here.
Summarise the report there.
Summarise the email from marketing.
Schedule a meeting with Jim.
Turn the lights on in the kitchen for the housekeeper.
While yes, there is some usefulness in having something summarise things for you, most people in their day to day lives don't handle a "100 emails from everyone" and "30 meetings with 40 different department heads" or whatever these people do.
Quite a big gap from the market they are targeting with Siri.
> Summarise a meeting here. Summarise the report there. Summarise the email from marketing. Schedule a meeting with Jim. Turn the lights on in the kitchen for the housekeeper.
Isn’t that what administrative assistants and hired help do? To handle the minutia of triage and actually doing the stuff you’ve decided upon? And they actually have to do their job well because (reputation loss, employment, and legal)?
Why is Apple struggling so hard with LLMs? I get that they're a hardware company, but surely that can't be the full story. They must be suffering from some case of extreme perfectionism or something like that. I would assume it'd be very difficult to get an LLM to perform consistently and reliably without making it useless.
Its wild to me that people actually believe the privacy marketing.
I guess not... there is a reason Apple is still in business, they are great at deception.
It just seems so silly to believe a blackbox company that was outed by Snowden in PRISM would actually be privacy focused. Or that a company that touts security has been hacked so easily by Pegaus that anyone even buys into the marketing.
But hey, people are religious, so I should be less surprised that people trust others.
No commercialized LLM product released so far would meet Steve Jobs' standard for a high-profile Apple product, and while that standard is clearly much more lax now, there still is one and still higher than what people accept from OpenAI et al. I wouldn't call that perfectionism, just a struggling brand standard that can't afford to lose even more face.
I think many inside and outside Apple hoped that the ways that they scoped their features in the Apple Intelligence announcement would help them pull off something duly reliable and practical, but it's not that surprising that even those ambitions might have bought too deeply into the hype.
> I wouldn't call that perfectionism, just a struggling brand standard that can't afford to lose even more face.
It's very well-known that Apple is perfectionistic. I'm not meaning to say that perfectionism is a negative quality or a bad thing, just that it takes a while.
Apple is not perfectionistic. Apple is performative. The entire company is performing software development instead of actually doing it.
Apple's development process is a marketing pitch-driven hallucination - project management by buzzword and individual career status progression.
It's almost entirely inward-looking. The connection to Rest of World is increasingly mythical and remote.
Some good work gets done in spite of this. But senior management doesn't understand quality - either in the internal sense of having bug-free robust product, or the design sense, where products meet real user needs in a satisfying, creative, and delightful way.
Nice graphic design though. Apple is still the leader there. Processor dev has also been exceptional.
IMO it's time for most of the C-suite to step down and let much younger talent take over and shake things up.
Has things gone cracy since I last used OSX for real in 2008 or whatever? Windows have become such a shitshow since Windows 7. I kinda assumed Apple didn't follow suite.
> Apple is not perfectionistic. Apple is performative.
My guy, Apple is autistic as shit. Just look at how in-house everything is. Every time Apple runs into someone doing it wrong, they do it themselves. You know the saying; if you want something done right...
Sure, they're performative as well, but that doesn't change that they are still in the process of doing absolutely everything from scratch because the existing solutions are not good enough for them.
I don't know if you've used the new Apple Silicon MacBooks, but I have a 128GB/8TB one lined up for me to pick up from the Apple Store in about an hour. It's certainly not the best that Apple offers (that would be the new 512GB/16TB Mac Studio) but it practically wipes the floor with every other laptop on the planet. Because Apple made their own chips, because everyone else was doing it wrong.
> No commercialized LLM product released so far would meet Steve Jobs' standard for a high-profile Apple product
I can put words into Steve Jobs' mouth as well!
Steve Jobs would reframe LLMs (and other ML-based solution) as creative assistants, insisting that their job is not to get the right answer, but to help you get your creative juice running.
To get this to happen, he would have personally convinced engineers that have made cool AI demos to work for Apple to turns these demos into features that form a new Apple creative suite.
If Apple released OpenAI's voice mode, without calling it "AI," referring to a "GPT" or a "model" -- if it was just integrated into the iPhone with a wake word, that absolutely would satisfy Jobs.
The problem with these companies is that they can't visualize how a product works for non-technical customers any more. Because everyone they know lives and breathes Silicon Valley. They see billboards for vector databases on their commute, so they think it's perfectly normal to name a product "GPT."
A product containing a language model should never, ever be called "AI" or "GPT" or even "Intelligence." 10 years ago, the only people who knew what the term "AI" even was were the nerdy readers of pulp scifi novels. I'm half joking but the whole town of silicon valley needs their glasses smashed and to be shoved into a locker.
Depends what you mean. Building an LLM seems to be easy enough that several companies have built one, and even Twitter was able to magic one out of the ether from pretty much a standing start given the resources.
Getting an LLM to do the right thing for you 99.999% of the time? Has anyone managed that yet? Frontier models still hallucinate the most basic shit, and consistency is very, very difficult.
Theory: a lot of this work was started a few years back, when it was a common belief that LLMs would continue to improve at a breakneck pace. That has now happened, so now they’re stuck with a lot of clearly Not Good Enough stuff.
People expect the llm version of siri to be an agent, not a chat bot. There's a very wide gulf between a chatbot like chatgpt and an agent like... well I haven't found any good agents I'd give write access to my calendar yet so I don't know what to cite.
John "JG" Giannandrea was hired over to Apple around 2018 to head Apple's new AI/ML division, which would include the Siri organization that predates Giannandrea's tenure at Apple. Giannandrea was previously the executive in charge of Google's Research division.
Mike Rockwell was hired into Apple from Dolby Labs around 2015 to head a group called "TDG" that would be the R&D and the product group for what we know today as Apple Vision, an AR/VR headset. The same organization had its own AI/ML applied research teams focused on computer vision problems, and had hired in folks from Microsoft's Hololens team.
From what the article describes, the Siri organization is being moved from Giannandrea's scope to sit under Rockwell, and Rockwell will be moving away from the Apple Vision organization.
Apple hired Giannandrea to build an organization like the one he led at Google, and appears to have found that that the AI/ML organization built at Apple struggled to manifest its work into successful product wins despite the massive financial investment. It's worth noting that at Google, Giannandrea was succeeded by Jeff Dean, and the organization was more recently reorged to become the "Google Deepmind" we know today also had the same struggles.
It feels like Apple had a dream of a granting easy AI integrations for many different apps and workflows, only to discover that very few people wanted any of those integrations from apple intelligence. At least, not the current iteration of apple intelligence.
I suspect the most significant thing holding Apple back in the realm of AI is the fact that Apple prides itself as a company that delivers revolutionary products with ‘wow factor’ that is leagues above the competition. Apple just doesn’t have anything like that with AI.
Furthermore, the challenge with AI is that even though it can deliver some shockingly impressive results, it also generates some bafflingly stupid responses, as well as answers that are seemingly correct but are never less wrong in ways that are hard to determine. Those hallucinations really shatter the illusion of quality and reliability, which ultimately undermines confidence in the functionality as a whole.
Given that, Apple is stuck trying to catch up to the rest of the products in the AI space while also realizing that everyday AI is nowhere near as polished as Apple would expect it to be. So the more Apple pushes AI, the more customers are likely to pick up on the errors and inadequacies of the product.
Even though other AI assistants suffer from the same fundamental problems, the ones from Apple Intelligence are going to seem more embarrassing because of Apple’s brand image.
Gemini has access to your Gmail, Calendar, Docs, Drive, Keep, and Tasks, as well as Flights, Hotels, Maps, YouTube, YouTube Music, OpenStax, and of course the entire web more generally.
At the end of the day, personal context is king for useful AI.
Which means Google or Apple. Or Microsoft on the business side.
Nobody sane is going to share all their personal data (about everything) with a fly-by-night random AI app/company.
Apple has time, in the same way it did with Maps, because despite the marketing push no one is buying phones solely because of AI features these days. They're still an "Oh, that's nice."
That 'wow factor' is a thing of the past, very firmly. When I put next to each other say Iphone 16 pro max and Samsung S25 ultra, none sticks out, none has anything 'wow' for regular users. Just different styles of working with device and small + and - all over the place. Same for tablets, notebooks, and so on. Vision pro seems like a failed product even according to Apple.
Its doesn't mean that each product doesn't have something a bit special but competition caught up long time ago, sometimes went ahead (better batteries, bigger cameras etc). These days there is much more 'yawn factor', and its across whole industry.
They could add some “wow factors”, but the rate of error is still too big to ship it. Like, I don’t think it would be impossible for them to have a live language converter using LLMs when you call somewhere in a different language. But being responsible for stuff is just a can of worms that nobody wants to sign up for.
>I suspect the most significant thing holding Apple back in the realm of AI is the fact that Apple prides itself as a company that delivers revolutionary products with ‘wow factor’ that is leagues above the competition.
I don't remember them ever doing that. Usually they start with something lacklustre then incrementally improve it until it is strong.
Maps is a great example, it was poor on launch compared to Google Maps but has now got to a level where most people don't bother with Google on the iPhone.
Google Maps largely exists because of Apple.
Do you remember when the iPhone was released?
AirPods? The iMac? The Mac? The Apple II?
Broaden scope a bit to understand what this comment meant. You’re right that Apple botches things in the small scale and especially in software, but they’ve been a great aggregator of amazing tech in the total lifespan.
Google Maps had already launched and had really surprised everyone at how much better it was than any other mapping tool a few years before iPhone launched. and just the interactivity of it in general independent of it being a “map” was a paradigm shifter.
Yes, the iPhone put it in everyone’s pocket, but if Apple had built a handheld device that was powerful enough to run any interactive mapping tool Google Maps was already the obvious choice.
It’s a minor quibble, but to say Apple played a majority role in Google Map’s success doesn’t ring true to me. In 2007 Google Maps was easily overtaking all other mapping tools with or without the iPhone.
It was great when I was visiting China. I only tried it because Google Maps was unavailable (nothing Google worked at all except push notifications, because those come from Apple), and was pleasantly surprised. Every train station and mall had detailed maps.
Apple Maps is also pretty good here in Canada, although I would never trust it to tell me with any accuracy what a given business's hours are, or whether it even still exists. Google is much better at that.
That is now, there was very little wow factor when Copland was being developed and they were at the edge of bankrupcy, long were the days of Mac Classic and LCs.
What is saving them this time around is the way they have been priting money with iDevices.
Well, the current iteration of Apple intelligence doesn’t enable personal context or 3rd party app related features at all. (https://www.macrumors.com/2025/03/07/apple-intelligence-siri...) So who knows whether people people want it or not; it’s not even out in the world! What Apple has actually discovered is that this is a hard problem to solve. But people do want a smarter Siri; they complain about it constantly.
The devil is in the details – you do shit clippy-like integration = nobody will use it, you do it well – everybody will, even though nobody currently does in any meaningful way beyond setting a timer.
„[…]struggled to manifest its work into successful product wins […]“
working in AI/ML, is there any recommended literature one may read on how to ensure efforts result in products? Would be interested to access more experience to benefit both an engineer and a manager pathway.
Given the size of Apple and these orgs, plausible. Siri also had a bunch of folks that came in from outside, through acquisitions, and did their own thing or tried to identify themselves and their roles as something different from Apple norms.
Apple had teams involved with AI/ML in several divisions also, from AI/ML (including Siri), TDG/VPG, SPG...
> "appears to have found that that the AI/ML organization built at Apple struggled to manifest its work into successful product wins despite the massive financial investment"
I would contest this statement very vigorously. Siri has been a disappointment but nearly every single Apple product has a load-bearing reliance on an AI/ML feature.
I'm continually frustrated by the (relatively recent, since LLMs) tic where just because a ML model isn't conversing with you or otherwise making its presence impossible to ignore, that it isn't delivering a massive amount of value. It is frustrating from laypeople, it is extra frustrating from industry insiders.
Just a few examples off the top of my head:
- AirPods with ANC and spatial audio. Both headline features of these products that are 100% AI/ML.
- Watch with heart arrhythmia detection, automatic workout detection, fall detection, etc. All are headline features (some literally life-saving) and are 100% AI/ML.
- iPhones have high-profile features that are 100% AI/ML: automatic car crash detection. Others are more subtle but IMO substantial differentiators - such as automatic image enhancements out-of-camera.
Again, I know in the age of ChatGPT we seem to have twisted ideas of what ML is, but "AI/ML" is not synonymous with LLM.
1) This appears to be what Apple has concluded by the action they're taking and implied by the premise of the article, versus my personal opinion.
2) The examples given may have been produced by AI/ML-based technologies developed at Apple, but may not have been the work of Apple's AI/ML Organization. Many "groups" (divisions) within Apple have their own teams using AI/Machine Learning (e.g. AI/ML Org coexisting with AI/machine-learning work and teams in TDG/VPG, SPG (special projects group: the one previously working on a car), etc.
imo, Apple are actually ahead when it comes to the hardware side of the whole thing. Their vertical integration gives them an edge not many can match, when it comes to running ML models. It's a no-brainer for Apple to reduce the barriers for devs to build really cool native Mac/i{Pad}OS applications and incentivize them to leverage the built-in AI/ML abilities to a greater extent. The iPhone in part took off _because_ of the whole app ecosystem that got built around it. Sure they might take a loss in their services revenue in the short term but they get to be _the_ AI platform for at least the next decade and half - both on-device and server side with their new Apple silicon servers.
It's just that most Apple software seems to suck in some fundamental way right now. I don't know if it's a technical issue (SwiftUI being meh when compared to UIKit for example) or a culture issue or the money coming in insulating management from accountability. Software execution has been lagging behind the excellent hardware execution for almost all of the Tim Cook era. They desperately need someone like Scott Forstall to come in, kick butts and get stuff going again.
They ideally have a couple of years while waiting for Moore's Law to catch up to turn around their software side. Otherwise, it's a real shame that all that great hardware is just being used to run Electron B2B SaaS apps.
A lot of people forget that it was only recently that the Photos app on your iPhone could run OCR text search on pictures in your phone. Google had that feature on their phones many years before Apple.
Having worked on Apple's TTS for more than a decade, I can state with confidence that this is utter bullshit and you don't have the slightest idea what you are talking about. Both in terms of quality, and of the underlying technology used, Apple's current TTS is in no way comparable to what existed 10 years ago (at Apple, or anywhere else in the industry).
I challenge you to find a 2014 recording that is on par with a contemporary Siri voice.
I have been playing recently with those enhanced TTS model and they are of similar quality like piper TTS models to me - not that good. StyleTTS 2 like kokoro sounds so much better for me and also run realtime on their devices. And when you compare their online models to not even what OpenAI have but some small recent startups like Sesame or open source models like Orpheus, Apple TTS sounds (pun intended) really behind.
I don't dispute your claim, just that I still find Alex voice to be the best, and it's been the same since over 10 years ago. The other voices have issues, they don't sound too good at 1.5x.
Alex was developed when VoiceOver (the screen reader) was the primary use case for text to speech. Consequently, it was optimized for low latency and robustness under rate changes.
The Siri voices sound much more natural at 1x and have a higher signal quality, but rate changes were a lower priority for this use case.
Fun fact: when we worked on Alex, many VoiceOver users stubbornly hung on to Fred (which is mostly using late 1970s technology). Screen reader users are not fond of switching voices; it appears their hearing locks in to a particular voice, so switching is costly.
> Recent advances in text-to-speech (TTS) synthesis, such as Tacotron and WaveRNN, have made it possible to construct a fully neural network based TTS system, by coupling the two components together. [...] However, the high computational cost of the system and issues with robustness have limited their usage in real-world speech synthesis applications and products. In this paper, we present key modeling improvements and optimization strategies that enable deploying these models, not only on GPU servers, but also on mobile devices
They also had (but squandered) the potential to be ahead on the software side. macOS is the only platform I'm aware of that has every app wired for scripting (AppleScript/Apple Events). And not only that, they already solved the issue of adoption since almost every well-behaved (read: non-electron) application has decent support for AppleScript.
It would take very little effort to put an LLM frontend on top of this, and yet they've not only abandoned applescript (or the underlying apple events) at a time when they could expose it to the masses, but have gone in the exact opposite direction with "Shortcuts".
Oh and the icing on the cake is that apple events can be sent over the network as well, and this infrastructure has existed since the early days of OSX.
I agree. The AppleScript/Apple Event Manager thing is an example of treating macOS like a second class citizen in the ongoing iOS-fication of the Apple Ecosystem. The point of macOS for me is that it's simple to use for most beginners but allows more advanced users to add complexity through tools like Apple Script and Automator and the underlying Unix base.
Like you say, an MCP server integrated with AppleScript/Apple Event Manager would instantly hook up any LLM with virtually all Mac apps. More Mac devs will then be incentivized to support these features. For people who find AppleScript un-intuitive, JavaScript is also supported. And in my view, this is a revolutionary way to use my computer - a very Star Trek way of using the computer.
> Their vertical integration gives them an edge not many can match, when it comes to running ML models.
They have an advantage when it comes to running them locally, but it feels like they're trying to push it onto consumer hardware before consumer hardware is at the point of actually being able to run useful LLMs.
You're right. The hardware right now can't run useful models.
But that's why I think they have a couple of years to sort out their software issues. When useful models can be run on their devices they have to be ready. The hardware advantage can only be an advantage when they have the software to run useful applications. Hopefully they don't get stuck in the typical big company bureaucracy and ego matches and instead can make a change for the better.
They made it less useful because they were greedy. Before M1 we used to have laptops that had 16GB RAM as a base. With M1 the made base back to 8GB. In PC world before M1 and even before apple started soldering everything you could have easily laptop extended up to 64GB RAM for a cheap price. Those ram sticks are not expensive in retail price and should be even less expensive if wholesale and not full sticks as bill of material but just memory modules.
If last year they would make each macbook air as a standard have 32GB RAM and iPhone 16GB RAM + 250GB SSD as a base they would have the most capable hardware with big user base.
Sure they loose some money of people having upgrade models but they would sell much more Macs. As a reference they sell each year only ~20M Macs comparing to 60M ipads and 240M iphones. Macs are having only like what ~10% market share worldwide? They could easily double it but they protect their profit margin like virginity.
> Before M1 we used to have laptops that had 16GB RAM as a base. With M1 the made base back to 8GB.
You're getting product segments mixed up. From what I can tell, the 13" MacBook Pro in the Intel era never started at 16GB; the last model still started at 8GB. That's what was replaced by the M1. The 15/16" Intel-based MacBook Pro models didn't get a proper replacement until the M1 Pro and M1 Max, which started at 16GB and 32GB respectively. The only regression I can find there is that the last 13" Intel MacBook Pro could be configured with up to 32GB, which wasn't available from the base M series chips until the M4 last year.
>imo, Apple are actually ahead when it comes to the hardware side of the whole thing.
It surely is just your opinion. Nvidia is king and Apple has found a way to market an integrated GPU and CPU RAM as something magnificent, rather than something that has existed since the dawn of computing.
There is a reason Nvidia is king. There is a reason corporations buy Nvidia and not Apple for their LLM uses.
You seem to think that the AI market consists entirely of the segment that NVIDIA dominates (datacenter training and inference) and that the segment where NVIDIA is absent (inference on battery-powered devices) doesn't exist.
The baffling part of Apple’s AI strategy - they continue to insist on using the “Siri” brand. That brand is damaged beyond repair and has been for many years, maybe since it was first launched 10+ years ago. The recent mistakes of over promising and under delivering have further eroded the brand to near zero. For a company that comes up with ridiculous branding of small features (Retina Display, Force Touch, Dynamic Island) the choice to keep “Siri” and its negative connotations is un-Apple like.
And here we're on an English speaking forum, I can tell you that no matter how bad you think Siri is, it's actually 2x worse in the other languages they support.
Agree, Siri is a damaged brand. Microsoft had the insight to drop Cortana when it flopped too.
In the age of LLMs I also increasingly view my experience with "Hey Google" and "Alexa... " with similar contempt. Their output feels so bad compared to what ChatGPT's voice gets me. All of the big tech companies seem to be struggling immensely to deploy the latest in AI advancements.
I never tried advanced things with "Hey Google" but it always gets the basics right for me:
"go home" opens maps app with a route to my home.
"alarm in", "alarm at", "what time is it?" works well too.
In fact I'm afraid they'll break it by shoving down Gemini.
Alexa is the same with basics, works well for me with "reminder in X hours", "reminder at 10am on every workday", "alarm at", "play The Strokes", "add coffee to the list" (groceries)
Alexa is really bad with any advanced stuff I've tried. Didn't try with "hey Google"
I agree Alexa/Hey Google are generally great at following commands.. but lately I find myself most wanting to use voice primarily for getting quick answers to factual or opinion based questions. There the results are abysmal vs ChatGPT voice.
It also doesn't help that ChatGPT's voice cadence is much closer to sounding like a real human.
Siri meanwhile is still far behind both Alexa and Google Assistant in its basic voice-to-text conversion capabilities -- a pain point I experience everytime I talk into my Apple TV remote to search for content.
It's not uncommon for technology companies to over promise and under deliver. We just aren't used to it from Apple because of their secretive nature. They released a new MacBook recently and we don't know whether that was on time or months late simply because Apple never mentioned it before the product was released.
Whether the fault lies with individuals or the structure of the organisation is a more interesting question. I wonder if American corporations would take a more individualistic approach to this compared to their European and Asian peers.
Apple knows how to call an OpenAI API as well as anyone. I don't think they're struggling with the LLM side of things. The problem is to ensure security (vs prompt injection), and to create an ecosystem of apps and developers who will provide hooks into their apps to let AI perform actions on behalf of the user.
IMO, Craig Federighi is quite good at presentations and his own style of humor, but he doesn’t seem to have a good grip on the quality of software across all of Apple’s (buggy) platforms and apps. I don’t believe that moving Siri under him (i.e., his reporting structure) is a good idea. Only time will tell.
There are far more bugs and issues that are to be fixed (but don’t get attention) than Siri. Anyway Siri has been close to useless for a decade and a half. It’s not a big deal if it remains so for another half a decade. Nobody outside of Apple has any expectations on it.
Edit: On a related note, this leak seems to be significant. The Apple I know of would fire the executives (among the “top 100”) who leaked this and probably make sure they’d pay some hefty monetary penalties too.
The fact Apple Music has notifications advertising it on my lock screen, or splash screens advertising it once a month when I open Music.app would have been unthinkable pre-Tim Cook.
None of my Samsung phones have ever advertised anything to me. Unlike iOS,there are any number of music player apps on Android that stay out of the way and don't try to be a "platform".
Galaxy S8 came out in 2017, almost a decade ago. I understand some budget and mid-range phones have ads, but I (in Canada) have used everything from S7, S8, S12, A51, Z Fold 3, and have never gotten one. There was a "Samsung Hub" app on the midrange A51 (similar to Google Discover), but that can be disabled.
Yeah people repeat advice like this but have no idea it makes no difference and even in search it defaults to "Apple Music" but shows no results even when that slider is turned off.
It used to work in the first few versions of Music.app but broke about 2 years ago and has never fully disabled it ever since.
You’re wrong. It displays Apple Music, then when you search it only pulls data from what it has access to, which are the free radio stations inside Apple Music.
They need to stop shooting for the moon and just release some incremental improvements.
Like you say, Siri is borderline completely useless. Just timers for me.
Even just the most basic LLM integration, without even giving that LLM any personal data would be a massive improvement. They even have this ChatGPT integration that Siri seems totally incapable of using in any meaningful way.
I might be completely wrong, but by making “pro-privacy bets”, they shot themselves in the foot. Average customer ended up really not caring about the benefits that we try to promote, which hinders the development and features in some sense. Also, it’s still a pain for them to integrate AI in their second biggest market — China.
Maybe go all in on local on device low latency LLMs? They make their own hardware so they could probably do it and do it well with enough focus and investment.
That seems like the opposite of what they were suggesting? Unless by “edge compute” you meant user’s devices, but I assume you intended the usual meaning of CDN edges.
8GB of RAM is not perfect for this. The M chips are only known for AI because unified memory allows you to get huge amounts of "VRAM", but if you're not putting absurd amounts of RAM on the phone (and they're not going to) then that advantage largely goes away. It's not like the NPU is some world-beating AI thing.
I care about privacy but Siri is so immensely frustrating and terrible that at this point I'd rather just prefer a competent integration with an AI that actually works.
I care about privacy too, that’s one of the reasons why I’m within Apple garden. But it definitely adds some hurdles when you want to do stuff like “literally take all of my data from every possible place and do stuff with it”.
Apple doesn't care about privacy in China. Even iCloud data on China was held by a Chinese custodian. The pain they experienced to integrate AI in China was because of the rule that Apple must choose a Chinese partner to integrate with. So they cannot choose OpenAI like they did in the rest of the world. They also couldn't choose Google Gemini or Anthropic obviously. They evaluated at least Alibaba's Qwen and DeepSeek and maybe some more.
> Even just the most basic LLM integration, without even giving that LLM any personal data would be a massive improvement.
I don't think Apple will ever use LLMs, given the offensive output they can generate sometimes. There is just no way to make them safe for all audiences.
> Nobody outside of Apple has any expectations on it.
Well, apparently some people have at least some expectations, after Apple pushed ads touting Siri's new abilities (and Apple AI) but failed to launch them on time, and now Apple been hit with a federal lawsuit against them: https://www.axios.com/2025/03/20/apple-suit-false-advertisin...
I’m one of them. I saw demos at _last year’s_ WWDC that I’ve been excited to get. And now with this announcement I don’t think that will happen before this year’s event.
Last year’s “demos” weren’t demos, but concept videos. I wouldn’t be surprised if it doesn’t happen this year nor next year, to the level implied by the ads. Incidentally, what the original Siri ads showed also never materialized.
Never mind this year's event - it's going to be multiple years before they have something of the quality they 'demoed' last year. They rushed some great concept videos and then thought they could actually build them in less than a year. And sold $1k+ devices based on it. I'd honestly say this is a bigger disaster than Apple Maps. At least in that case they were forced into doing something really big really quick (because Google was crippling the iOS Maps app). In this case Apple didn't need to do any AI. Absolutely nothing they have shipped is useful or innovative. It's one of the rare times they've followed the market instead of waiting to do it right. At the minute, it's not any better than the co-pilot rubbish MS have shoved into Windows.
Edit: I wonder if this will force them back to live events? Live demos are at least more believable and might win them back some trust.
I was hoping for more from the new Siri, but its just awful for me. Latest example:
My wife and I were watching Jeopardy late at night, and Pol Pot came up. I asked Siri something like "Siri, tell me about Pol Pot" and instead, it heard "Siri, Call Scott", and rang my friend at 1am.
Apple have never learnt how to build software sustainably. They let the bugs, misfeatures, and tech debt pile up over time. They addressed this once with a huge “0 new features” iteration with Snow Leopard to catch up on the worst of it, but never resolved the root of the problem, so it started to build up again. Now it seems like they don’t care enough about software quality to even do that. They need somebody else in charge of software who will actually change the way they do things.
Amen to that. Please for the love of all things remember your future defining ‘human design’ with brilliant UX in early macos and hypercard. Now you have unlimited money and don’t give a single f?
Do better please! Where is the passion, catering to pro’s, any idealism? Hardware is great but software gets worse and worse..
You do remember the crash prone classic MacOS, the ridiculous Performa line, the Copland disaster and that Apple was almost bankrupt?
Also the classic MacOS where you (the user) had to pre allocate memory per application and where holding down the mouse button stopped any application from running in the background.
When I started at Apple it was to work on Quickdraw GX.
It was very cool tech. Closer to what Apple would get from NeXT years later.
My guess is as to why it failed (and this was also the common sentiment at the time) — it was an optional install and not a default. It's hard to imagine anyone writing an app around a component (extension) that few people had in the OS.
It's wild but my own issue free MacOS era was OS9 on my iBook G3 clamshell. Had to replace a harddrive (bounced down a stairwell, but held together) and upgraded the ram, but other than that fine experience.
My next Mac was a 2008 Macbook Pro (non-unibody) that had backlight issues out of the box, failed display cable two years in, and then the GPU failed. (Nvidia plague era, and Steve Jobs silence plague era of shit-ass fan curves)
To this day I still refuse to trust Apple's fan curves and install fan control software. Especially the M1 iMac with the cut down graphics that they only gave one fan and a smaller heatsink. Watched it break 100C regularly.
The m1/m2 mac mini might be the only one I've never had issues with the stock fan curves so far, but I still bump them up anyway for peace of mind. I don't see a reason to let an SoC hit 98C even if its rated to do so.
I skipped most of the x86 era. I had the first Intel Mac mini connected to my TV running Front Row in 2006. But after that I didn’t buy another Mac until the M2 MacBook Air in 2023.
At the end of the day, I was just not that impressed with the x86 based Macs. They had all of the drawbacks of x86 with all of the added disadvantages of Ive inspired design which led to poor heat dissipation, not enough ports and butterfly keyboards
They aren't even doing hardware well in some cases, their bread and butter.
Bluetooth has been broken on both editions of their Pro Max Phone (15 & 16). It cuts out and has all kinds of weird issues, but no fix has come for years now. This was why I always paid a premium for their phones, they did hardware and software well. Not anymore, their so focus has been on their processors.
Are you using any Bluetooth headsets designed for their special sync protocol? I noticed no issues with any certified headsets but any of my nice Bluetooth headsets from a few years ago that don’t leverage the newer protocol instead only standard Bluetooth have issues syncing and such.
This tends to be an Apple issue. They introduce proprietary thing and eventually unless your hardware supports said thing it will have a degraded experience in some way
I have no idea how Apple prioritise their bug fixing, but it really looks to me like bugs only get fixed when adding new features, if at all.
This is such a trivial big win as well, create a software dev team to focus on the outstanding bugs. Head it with one or two senior developers, make it the first step in onboarding new developers. What a great way to learn, and make a difference.
This is not rocket surgery, and Apple really do have a bad reputation here, even with the muggles.
> This is such a trivial big win as well, create a software dev team to focus on the outstanding bugs. Head it with one or two senior developers, make it the first step in onboarding new developers. What a great way to learn, and make a difference.
Sounds nice on paper. In practice, the seniors assigned to the team will see it as a career dead-end (because fixing bugs doesn't get management attention, but flashy new features do), juniors lack the experience to avoid edge cases or to navigate the project structure, and the seniors/intermediates on the "actual" development team don't have time to waste on reviewing and aiding code of juniors and bugfixes because then they have ownership over that...
Moreover, Mac OS X 10.6.0 was much buggier than its immediately predecessor Mac OS X 10.5.8. There were some really bad bugs in those early Snow Leopard releases.
The Snow Leopard that everyone remembers and loves was not version 10.6.0 but rather version 10.6.8 v1.1, which was released almost two years after 10.6.0.
Major OS updates invariably introduce more bugs than they fix. That's an iron law of software development. What Apple used to do, which they no longer do, is spend a long amount of time just fixing bugs in minor updates to the operating system. Now, unfortunately, Apple is on a rigid yearly OS update schedule. Worse, Apple doesn't just fix bugs between those yearly updates; instead they keep introducing new features even in the "minor" updates. So we never get a "stable" version. It's constant change for the sake of change.
> Yeah it seems like there’s a serious fundamental problem, or multiple, beyond just the yearly cadence.
My argument is that they were never building and have never built software at a sustainable pace, even before the yearly cadence. They race ahead with tech debt then never pay it off, so the problem gets progressively worse.
A while back, that merely manifested as more and more defects over time.
More recently, they began failing to ship on time and started pre-announcing features that would ship later.
And now they’ve progressed to failing to ship on time, pre-announcing features that would ship later, and then failing to ship those features later.
This is not the yearly cadence. This is consistently committing to more than they are capable of, which results in linear growth of tech debt, which results in rising defects and lower productivity over time. It would happen with any cadence.
Yeah, I personally remember "10.6.0 utterly fucking sucks" (though it got better) — hard to remember the details, which is probably in my personal best interest, but I remember fairly vividly shouting that to friends over beers somewhere.
And I also remember 10.7 actually being better and where I thought Mac OS X (and its macOS rebrand) peaked, for interactive/workstation use.
However I ran a 10.6.x virtual machine (that yes, ran Mac OS X 10.6.8 Update Combo v1.1[1] for the last decade of its service life) until 2021 for a bunch of long-timeline server functions, so I guess Snow Leopard was pretty good overall, in the end...
I can't remember which of these it was but one of these actually "bricked" my Mac Pro on the OS upgrade, at least from the casual Apple fan perspective or the Genius bar perspective.
I took it into the Genius bar and they had nothing other than "we will wipe it and get it working."
It turned out I was able to boot to single user unix console mode and clean up the issue in an old school style. I have no idea if that is even possible anymore, but in any case I have never had another issue with OSX as severe as that one.
Modern Mac OS got really quite high quality at some point but it's easy for people to forget it had a lot of growing pains between 1999-2008 or so.
I agree. About a year into the Intel transition, around 2007, it was incredibly stable.
I was doing hardcore computation 24/7 on a MacBook, including model checking real-time systems, and I vividly remember the exact day I experienced one kernel panic.
The rest of the time, everything was rock solid. It's sad both quality and aesthetics degraded during the 2010s, where the system somewhat converged to iOS UI patterns and lost skeuomorphism.
Apple did skeuomorphism really well, which is hard. No idea why they moved to flat designs, which are easy to copy by competitors.
This is far from isolated to Apple, too. It's a cancer that's infested virtually every single software company in the entire world at this point, and it's endlessly frustrating. If they don't have some New Shiny Version to show off to the tech media and blogosphere everyone starts acting like the software itself is a dead man walking. I also think it's partially down to all these same companies trying to justify subscription billing with a constant trickle of "new stuff" that the application in question does that you're meant to feel like you'd be missing out on.
I don't ascribe to this notion whatsoever. I in fact far prefer my software that barely changes at all over time. My most reliable and useful computer is my huge Ubuntu box that lives in my laundry room which handles... I mean frankly it would be easier to say what it doesn't handle, which is anything Apple adjacent because that's all proprietary, and it doesn't handle my security cameras because I couldn't find a linux-based solution for that. Literally every other automated function outside those two categories is handled by that machine, with a mix of software I've downloaded and configured and a fair bit that I've written myself.
I say all this to say that machine barely changes at fucking all. I do security updates of course and occasionally have to tweak something but the apps on it are the same, it serves the same internal websites, it runs the same scripts, like clockwork. I LOVE that thing and my life would be in an utter shambles if some overpaid tech executive had the power to randomly change how it works to justify his salary.
The way I see it is like fashion or fashion-adjacent, at least on the consumer side. One of the big accomplishments of Jobs direction is that they made computing cool, something you'd want to display instead of a functional beige box you hide away. It's not in the background, you have these colors, elegance or attention to detail on computing you want to be seen, and new major models need to be different to keep customers engaged along with being the status symbol.
With further progression on integration/miniaturization and lower power consumption/heat output for general computing, the devices have shrunk to their bare minimum as slabs, there's very little physical to display so the fashion needed to move to software (and what online services it can be a client for). Stasis is death when you rely on selling, and maintenance is a constant cost which no one really wants to pay for.
As a side note, it's interesting to compare attitudes to the physical side to PC, where for years there was almost a pride that PC was a minimalist functional box (besides the occasional fun/weird exception), but that contrasts against now where it's hard not to get components with windows or embellished with LEDs and logos on display (and there's another weird contrast, where on one hand you have internals on display, on the other trying to minimize or hiding away the cables as though the inner workings were ugly)
2 years ago I built an AI workstation (aka gaming PC) with all the RGB bling because it would have been harder to find parts that didn’t have the bling.
I think of Microsoft’s bad habit of introducing meaningless superficial changes to the interface but they also keep legacy software working for a long long time. (Office ‘97 works fine on Win 11)
The strangest thing about this is in the system configuration dialogs where they have a set of really gee whizzy modern configuration screens but you still have to go into 15 year old screens to do many common tasks —- they updated some but kept the old ones that they did not have time to update which is a little ugly but fine because they old ones work 100%.
I used to have “who moved my cheese?” moments all the time with Windows circa 2005 when I was admining Windows and having to deal with configuration screens in ‘98, 2000, XP and various versions of Windows Server, they couldn’t resist making superficial changes that annoyed me to no end.
Oh my god, as a UI designer by trade, the sheer STATE of the Windows OS will send me into a rant far too long (and that I'm far too sober for right now) for a hacker news comment. It's a TRAVESTY. If you try and modify certain system settings you will see, in no particular order, the slightly cleaned up Metro ala Windows 11 (I honestly don't know if this got a name), remnants of the original Metro from 8/10, the modified refreshed control panel from windows 7/vista, and as you point out, dialogs that are seemingly ported directly from as old as Windows 98.
I don't know who if anyone really is heading the UX department at Microsoft, but they should either be empowered to their job properly or fucking fired, and I couldn't tell you which.
Lots of people in this thread (correctly!) pointing out Apple's struggles with releasing high quality user software on MacOS. But at least when they changed the System Settings UI in MacOS Ventura (10.13) they managed to stick the landing. Not that I necessarily _like_ the new Settings UI, but there's no jumping into old UIs to configure things.
Microsoft released Windows 8 in 2012, and here we are 13 years later using interaction panels that have only gotten minor chrome updates since their original release in Windows 95 (looking at you, Advanced System Settings and Device Manager).
It really is the worst of both worlds. I don't like change for change's sake which we seem to agree on (though I do like the newer MacOS but we can agree to disagree there, that's strictly a matter of subjective taste) but it is INFURIATING and frankly should be embarrassing what a mess Windows is now. Change for change's sake is bad enough, they're currently doing... change for change's sake but not changing the whole thing because either they can't be fucked to finish the job or it was, for whatever reason, technically infeasible to complete it but they released it half-done anyway, in Windows Vista, 7, 8, 10 and now 11. How fucking embarrassing.
Yeah, but the fear I have is that if they put somebody like you in charge they'd prioritize visual consistency over functionality and things would just get dropped.
On some level it's an embarrassment, on the other level I never run into a "can't get there from here" experience though I had an awkward experience the other day when I brought home a bargain monster TV from the reuse center and found I had to look across a several different screens to find all the settings to get the video output properly matched to the TV.
> Yeah, but the fear I have is that if they put somebody like you in charge they'd prioritize visual consistency over functionality and things would just get dropped.
But the thing is functionality is already being impacted! I had an experience the other day where I was trying to manually configure an IP address for an ethernet interface to configure a piece of hardware, one of those situations where it lacks DHCP but I knew what the IP scheme was and so I just had to configure an address, gateway and subnet in order to be able to talk to this damned thing, which I was eventually able to do. But, crucially as I'm trying to do this, Windows pushes me towards a Modern/Metro UI version, which, fine, whatever, but when I entered my information it would then show a series of "in progress" dots across the bottom of the screen, presumably because it was attempting to verify it had internet access on this interface, which it obviously didn't and wouldn't, but it did have internet access via the WLAN card. This contradiction seemed to confuse the shit out of the Settings app, as it presented no error and eventually seemed to just... shrug and do the thing I'd asked.
Except it didn't work, and I spent an hour troubleshooting why I couldn't yet communicate to the thing, only to eventually out of exasperation go to the control panel to see if the old UI version of this (right click interface, properties, ipv4, properties) only to find out it NEVER SET THE ADDRESS I SPECIFIED AT ALL! Presumably because it couldn't verify it now had internet with those settings, or maybe because the UI was just broken, I don't know.
(Also interestingly CMD thought it had the address configured, but no subnet or gateway, and the old UI had none of it and it was still configured as DHCP)
And like, this is just bad software design, bordering on incompetent software design, for several reasons:
* We have an in-progress "I'm doing something" indicated from the OS, and I'm assuming it was in fact doing something, but it was not indicated at all what was being done, or why, or the result of what was done or what went wrong, which is simply horrendous.
* It was, I'm guessing, attempting to verify connectivity, but that was not explained to me, nor was it presented as an option as it is in the old UI (there's a checkbox that says "validate this when I'm done" and you can leave it off, which I did which then worked as I thought it would)
* Somehow, these three functions to show an interface's state (CMD, old UI, new UI) are all producing different outputs to one another about the same interface, which is just BONKERS in terms of software designed to manage a system, and implies severe fragmentation in the core of the OS.
So, just so I'm understood, I don't think Windows 11 is embarrassing because it has different UI conventions present simultaneously. I think it's embarrassing because of what that fragmentation means for lower, more important parts of the OS. It's a less important thing that points at a more important thing.
>The strangest thing about this is in the system configuration dialogs where they have a set of really gee whizzy modern configuration screens but you still have to go into 15 year old screens to do many common tasks
I much prefer the older screens. They are so much easier to use.
As the person who personally ran 10.6 v1.1 at Apple (and 10.5.8), you are wrong(ish).
The new version of the OS was always being developed in a branch/train, and fixes were backported to the current version as they were found. They weren't developed linearly / one after another. So yeah, if you are comparing the most stable polished/fixed/stagnant last major version with the brand new 1.0 major version branch, the newer major is going to be buggier. That would be the case with every y.0 vs x.8. But if you are comparing major OS versions, Snow Leopard was different.
Snow Leopard's stated goal internally was reducing bugs and increasing quality. If you wanted to ship a feature you had to get explicit approval. In feature releases it was bottom up "here is what we are planning to ship" and in Snow Leopard it was top down "can we ship this?".
AFAIK Snow Leopard was the first release of this kind (the first release I worked on was Jaguar or Puma), and was a direct response to taking 8 software updates to stabilize 10.5 and the severity of the bugs found during that cycle and the resulting bad press. Leopard was a HUGE feature release and with it came tons of (bad) bugs.
The Apple v1.1 software updates always fixed critical bugs, because:
1. You had to GM / freeze the software to physically create the CDs/DVDs around a month before the release. Bugs found after this process required a repress (can't remember the phrase we used), which cost money and time and scrambled effort at the last minute and added risk. This means the bar was super high, and most "bad, but not can't use your computer bad" bugs were put in v1.1...which was developed concurrently with the end of v1.0 (hence why v1.1s came out right away)
2. Testing was basically engineers, internal QA, some strategic partners like Adobe and MS, and the Apple Seed program (which was tiny). There was very little automated testing. Apple employees are not representative of the population and QA coverage is never very complete. And we sometimes held back features from seed releases when we were worried about leaks, so it wasn't even the complete OS that was being tested.
A v1.1 was always needed, though the issues they fixed became less severe over time due to larger seeds (aka betas), recovery partitions, and better / more modern development practices.
Do you mean Mac OS X 10.6.8 v1.1? Otherwise, I don't know what you mean by "v1.1".
In retrospect, I'm guessing that you meant N.1, e.g., 10.6.1. But 10.6.1 certainly didn't solve all of Snow Leopard's problems. Like I said before, the "stable" Snow Leopard everyone loves was the result of almost two years of minor updates.
> you are wrong(ish).
> So yeah, if you are comparing the most stable polished/fixed/stagnant last major version with the brand new 1.0 major version branch, the newer major is going to be buggier. That would be the case with every y.0 vs x.8.
In other words, as you admit, I'm 100% right. ;-)
Every new major OS update introduces new bugs, that didn't exist in the previous OS, some of them horrible, and Snow Leopard was no exception.
Sorry, I thought I was using your terminology. I ran pretty much all the updates (switched between majors and security updates and others with my team) as well as helped with the custom OSes for hardware released between. I was talking about 10.6.1, which included any fixes in a 1.1 respin. Apple's updates could be versioned (so, for example, you can have the 10.5.8 update v1.0, v1.1, etc.)...I am specifically talking about the first software update to the new OS.
I still believe you are right and wrong (which is why I put the "ish" on wrong).
"Every new major OS update introduces new bugs, that didn't exist in the previous OS, some of them horrible, and Snow Leopard was no exception" is not what folks are noticing when talking about Snow Leopard. People don't say Snow Leopard was great because it had no bugs, they say it was great because it had fewer bugs they experienced at release and converged quickly to something that was very stable, much more quickly than previous releases (and Leopard was a particularly long stabilization cycle right before it).
I do remember that bug you link, and I remember is was via migration assistant and required some convoluted steps to get in that state (IIRC it largely cropped up with people migrating their home from extremely old versions continuously, which was one of the QA tests we actually did but this issue was missed). There was much debate as to just how widespread it was (Apple didn't have telemetry at the time and had to rely on forums and apple care reports). The majority of people had no issues with SL, which was not the case with Leopard. Leopard's development internally was also more chaotic FWIW.
During that time period my team and I triaged every single Mac OS X bug coming into the company every morning. Trust me, SL was of higher quality than Leopard.
Nowadays, shipping is less of a major event due to CI, automated tests, betas, no CDs/DVDs, recovery partitions, etc. We're in a much better world!
> I ran pretty much all the updates (switched between majors and security updates and others with my team) as well as helped with the custom OSes for hardware released between.
Could you clarify what you mean by "ran"? Originally I thought you meant that you were the manager in charge of the 10.6.1 and 10.5.8 releases: "the person who personally ran" is singular. Whereas now I'm getting the impression that you were just one of many engineers who installed and used internal builds of the updates?
> People don't say Snow Leopard was great because it had no bugs, they say it was great because it had fewer bugs they experienced at release and converged quickly to something that was very stable, much more quickly than previous releases
That's precisely where we disagree. My impression is that people are just looking back fondly at the later versions of Snow Leopard.
> (Apple didn't have telemetry at the time and had to rely on forums and apple care reports). The majority of people had no issues with SL, which was not the case with Leopard.
These two sentences seem a bit contradictory. ;-)
> Leopard's development internally was also more chaotic FWIW.
Externally too. Leopard was infamously delayed for iPhone.
> Trust me, SL was of higher quality than Leopard.
Well, from my perspective, "Leopard" and "Snow Leopard" are a bit amorphous. They are brand names, not specific pieces of software. 10.6.0 was definitely not higher quality than 10.5.8, as I think you've admitted already. And that's precisely why I object whenever people say, "We need another Snow Leopard". It's never the major releases that create quality, stable software but rather the long train of minor bug fix updates that occur afterward.
By the way, I also remember an NSURLConnection crasher that was introduced in 10.6.3 and IIRC plagued us for a number of 10.6.x versions until it was finally fixed. https://openradar.appspot.com/7841731
I certainly admit that Snow Leopard was less ambitious than Leopard, and thus 10.6.0 likely introduced fewer new bugs than 10.5.0. If you have to ship a major update, then obviously fewer new features is going to create less disruption than more new features. But the fundamental quality problem today is that Apple forces itself to ship new major updates every year. That's completely unnecessary, a self-inflicted wound. If you look back prior to the annual schedule, the major Mac OS X releases came 14, 18, 30, 22, and 23 months apart. That's much more reasonable.
> Nowadays, shipping is less of a major event due to CI, automated tests, betas, no CDs/DVDs, recovery partitions, etc. We're in a much better world!
You cannot be serious! As far as I see, Apple software quality is worse now than ever.
And just to compound the problems, we never get a "stable" version anymore, because Apple continues to ship new features throughout the year, even in "minor" updates, so even 11 months after the .0 release, we still get the bugginess, and after every .0 release Apple has to turn around relatively quickly and start work on the next major version in order to be ready for the next WWDC in June. It's a vicious cycle.
If anything, CD/DVD installers were good, because only the most eager users ran to the retail stores to buy them and jump onto the bleeding edge new versions, and bugs could be fixed before the majority of users eventually installed the major updates, whereas now Apple is pushes users hard to install major updates over the air immediately at .0.
> Now, unfortunately, Apple is on a rigid yearly OS update schedule
I cannot think of anything that less needs yearly major updates than macOS. It's a mature, 25 year old OS that, unlike iOS, is used for an enormous range of professional work. I need to it be fast, stable, and predictable. The last thing I want is change for the sake of change, which largely describes the last 5-10 years of the OS.
Wanna make Notes.app compatible with the new features in this year's iOS? Just update Notes. I don't need a whole new OS to get new features for a notepad.
Imagine that the last Major OS release was going to be a bug fix release ... but then AI happened. And imagine this happens every year with some new feature demanding it be shipped.
Having said that, here's the reality of bugs that are assigned to a team. Feature work comes first. The (older) engineers don't want this — they want to fix bugs, refactor code, get rid of tech debt. But ever since Jobs returned to Apple, it's become top-down and the engineers are no longer driving the ship. Management want the features.
So, the slippery part is this: once you defer a bug for an OS release and ship your product, deferring that bug in the next OS is a no-brainer. You already shipped with the bug, why is it so important this time around? So users/devs learn to work around the bug.
Occasionally, as an old-timer, I would on a whim just fix an arbitrary bug that was "bugging" me. Sadly though, the process had become so top-down though that I would also need: 1) someone to code-review the changes (and so already you are encountering your first obstacle requiring "buy in") and then 2) get permission from "management" (this might be the BRB, bug review board) in order to submit the fix and the Radar. (There are periods though when they are lax and bugs can be submitted with little justification — or we would just make up some bullshit to get the bug fix in).
I recall the early days with fondness when a dev just checked fixes in.
Me: Craig, are you working on the Crop tool in Photos?
Craig: Yeah, why?
Me: How does it pick the default for whether to use Portrait or Landscape? Half the time I think the default it picks is wrong. Why not look at the image bounds and default to Landscape if it is wider than it is tall?
Craig: Good idea. I'll make the change now and write a Radar for it later.
Also, a shout out to Brendan who would push back when in a feature meeting with something like, "Okay, which of these features can we drop? Because we have some serious technical debt we need to address that was not own your feature list."
> But ever since Jobs returned to Apple, it's become top-down and the engineers are no longer driving the ship. Management want the features
Were you using Macs before Jobs came back? The entire operating system for PowerPC based Macs was technical debt with parts of the OS running under emulation.
Yeah I had that game on my Mac LCII back in 1992 and Parena was great too.
The point still stands though. The operating systems were horrible back in the mid 90s and both Macs I had back then - the Mac LCII and PowerMac 6100 - were cripple by being a half speed bus. Back before it was necessary.
You're objectively right. I have been running emulators recently with System 6.0 and was surprised at how much crashing I must have just tolerated. I shouldn't have made it appear that I was defending the quality of pre-OSX Mac OS. (Although we were talking about how quality has come down of late.)
As an engineer before Jobs came back — and for a few years too after he was back — there was still a sense that engineers could call the shots with their own frameworks/apps. Copland, Pink, all that — you can dismiss them as failures but they were engineering trying to toss off the "technical debt" of no true multitasking. So, again, it was bottom up and was at least a joy to be an engineer working on the OS at the time.
As I say, for a few years Jobs let engineering call the shots as NeXT became integrated in the OS. I was on the Graphics team then and a whole new graphic architecture and window server were created more or less from scratch (heavily borrowing from NeXT of course since most of the graphics engineering team were NeXT).
At some point though the major changes were in and management began to take over. "Quality" was eventually measured in unit-test code coverage, for example. (Sigh.)
I was in college in the mid-late 90s. There was this brief period where all of a sudden it became painfully aware how much Mac OS was crashing. I was regularly using SGI, Sun, and IBM Unix workstations and Linux had just become available. I don't think I ever saw any commercial unix workstation "crash" and Linux the crashes were usually my fault.
If you were only using Windows you barely noticed, but if you used much Unix Mac OS got pretty painful. I mostly remember this from taking a digital music/video class and that was basically the only thing I used Macs for at that time. You never really got through an hour or two editing video or audio without a crash or two. We had Avid workstations where they controlled this by tightly controlling the software on the Macs, but I never actually got permission to use them.
There were also behavior patterns. If you only used Mac you were in a behavior pattern where you'd crash the machine less. If you were using Unix workstations a lot your behavior subtly changed and you tried to do more stuff at the same time. If you then went back over to the Mac and overloaded the Mac the same way it was a bad time.
> As an engineer before Jobs came back — and for a few years too after he was back — there was still a sense that engineers could call the shots with their own frameworks/apps
That’s a bug not a feature. Engineering led companies rarely ship good consumer products. What happened to Apple before there was a product guy at the top is a prime example.
Another recent example is Google. Google has the best technology in the industry but can’t ship and sustain a good product to save its life.
When your customers are nerds like you, I think engineers do fine.
I should qualify it somewhat though — I mean I was, for a spell, working on the ColorSync framework. So the "customer" was also an engineer using our API.
At the same time, when I worked on Preview initially I added features that I wanted and I think it dovetailed what users wanted. I am also a user I rationalized.
Further, many of the old timers like myself, were hung up on Tog and his Human Interface Guidelines. We would argue over lunch the Right way to handle UI for a specific feature, etc.
Jobs comes along and with Design at his side, and they decide that the address field in Safari should also be a progress bar (that appears to select the text as it loads?). Some poor engineer was told then to implement that. (I hated what I came to call one-off UI.)
Framing Snow Leopard as a "bugfix release" didn't happen for enlightened reasons, it happened because they cannibalized the Mac OS team and sent everyone to work on iPhone. The skeleton crew left behind lost the capacity to do anything but bugfixes.
This is why Snow Leopard was so buggy on release but unlike previous releases actually did get better: management stopped pushing new features, and even a skeleton crew was able to polish it over time. If you look at a map of tentpole Mac OS features, you can see that the ambition vanished with the iPhone's release and didn't recover for a decade.
This is also one of my only uses of Siri, and the only one it gets right the large majority of the time. I also try to use it to se reminders in the future, but it’s a crapshoot whether it will get the date and/or reminder text correct.
I may be remembering incorrectly, but I honestly think it was better when it was released years ago than it is today.
> I may be remembering incorrectly, but I honestly think it was better when it was released years ago than it is today.
I think you're remembering correctly. Siri used to work much better for basic tasks, and didn't seem to randomly fail for no discernible reason. I can't remember what iOS updates changed things, but there were at least a few that absolutely degraded Siri's functionality. One that stuck out to me was when Siri was one day suddenly unable to correctly respond to "play all songs shuffled" - it worked fine for years, but after an update, just decided it couldn't do that anymore.
My favorite is asking it to set a timer for n minutes, seeing the timer start and assuming it’s good, and only later realizing that it actually started a timer for n hours or n days. Learned that lesson with some burnt cookies.
Especially embarrassing when OpenAI's Advanced Voice can easily understand mixed-language contemporary slang on arbitrary topics, and even convincingly simulate it in the output. It can also do this at any volume, with "uhs" and "ahs", interruptions, corrections, and ignoring the accent.
> Siri has been close to useless for a decade and a half.
The concrete value-measure for me is whether I would feel any loss as a user if Siri simply disappeared. The answer is no, I would feel no loss.
But this is also true for me if the subject is Gemini, CoPilot, Alexa, &c. For all the money and energy burned, for all the advertising babble and grotesque valuation, I would not as a user be one jot less productive if the invisible Thanos hand of the market made them all disappear at the end of this sentence.
As an Android user I have to disagree, at least if we're comparing apples to apples (could not resist the pun).
Google Assistant with Gemini (or whatever its name is this month) is really remarkable at helping with mundane tasks, and I would really miss it if it were gone. Circle to search, asking questions about my agenda, and setting up new events on my calendar based simply on taking photos of documents or posters is so convenient that I really buy the concept of having an AI personal assistant.
And because this is a very sticky feature when delivered properly, if Apple is not involved it gives a very strong competitive advantage to Android flagships.
It's not so much a question of disagreement. The fact that you have found value as a user is fine by me, of course! Maybe your use-cases should be captured by the Siri development team.
If Siri has been useless for more than a decade, that also means that many users are barely using it, if at all. I'm not the only user of a system containing such applications that has found zero concrete value in them.
I use Siri for about three things: timers, reminders, and asking about the weather. But the reminders are mission critical. I use the watch, and any time my flighty brain hits on something, I set a reminder for it on the spot.
I literally couldn't get along without it. The only killer feature of Siri is accurate voice transcription, and that's all I need (and it could be more accurate, I just get used to comical translations when I use a rare word).
I guess I wouldn't mind if it were better, but I don't need it to be. Reminders are enough.
Don't you think that maybe the issues are rooted in the Apple's attempt to do everything by itself?
All things considered Apple actually delivers some great software, Apple is one of the few companies that built their own operating systems, their software UX is considered to be the gold standard. They pulled architecture transition processes that required great software engineering alongside with the hardware engineering and they did it amazingly well.
Apple's success in some areas suggests that the failure in other areas doesn't have to be about software development culture but simply wrong bets or policies.
Apple tomorrow can become the best platform for anything AI, they just have to open up their platform and let people on the cutting edge of AI to take over some functions.
IMHO Apple's greed is failing them. That revenue growth in the "Services" category on their quarterly reports is their Achilles heel. Locking down users on their platforms allows them access to a "market without competition" but there's actually a competition happening outside of this walled garden and that competition is fierce. People in the garden start thinking that maybe the grass is greener beyond the fence.
Ironically, there's a possibility EU saving them from themselves if they stop being just compliant enough to hold up in court and embrace the spirit of what DMA is trying to do. Throw a tantrum on EU and say that the lost revenue was because of the evil EU thingy and save this great brands and great company.
They've been very good at marketing, but it was often backed by good products. I still remember the feeling of getting my first MacOS X machine and what a breadth of fresh air it was. No drama or window focus stealing and a stable OS with relatively modern software that even had access to a POSIX shell! Wireless networking was simple, but you could still do advanced stuff!
But in those days apple was the underdog and HAD to do good by their users or there was no reason people would switch. Now it feels like their first thought is how they can monetize every step of the way - they're turning into Disney in that sense. There's often little money in fixing bugs anymore.
> It’s not a big deal if it remains so for another half a decade. Nobody outside of Apple has any expectations on it.
But when Apple advertises literally all their new devices as "Built for Apple Intelligence" without actually delivering on that promise, that's just false advertising.
His delivery is too perfect and makes him sound like a really good text to speech program. It’s off-putting. The only other person I’ve heard that has this presentation style is David Miscavige. Perhaps some analogies could be made…
Apple doesn't fix bugs†, they just eventually replace the feature or ship a different way of doing the same thing, whose bugs hopefully don't overlap those of the first feature.
We're on Sequoia 15.3.2 now and the keyboard shortcuts of the big new window tiling feature still don't work on non-Apple keyboards (or Apple keyboards, if you happen to have remapped the globe key to something else, like Caps Lock).
Hopefully the new guys take the existing GenAI bullshit and fuck off from turning it back on after every update despite me turning it off 3 times already.
I'm really buffled what makes it for Apple so difficult to cut off Siri and replace it with state-of-art technology (LLMs). Siri is still akin to a "dumbphone" 4 years into the age of "smartphones" which is a good indicator for a lost case and when it's better to start a product from scratch without the tech debt.
Org level dysfunction. It is likely an exec's pet project.
Kind of like how Meta keeps the disaster that is Horizon Worlds going, but much worse because Apple has a monopoly on AI integrations in iOS, and nobody else is allowed to compete with their terrible implementation.
We're at an inflection point in the last few months that means Apple might finally be able to do something good with Siri.
Fundamentally, Siri's great promise (well in the last few years) has been that it could hook into all of the scripting facility made available by iOS apps. There are tons and tons of AppleScript hooks on many, many apps -- if only Siri could, you know, access them.
This is traditionally not Apple's strong suit - API access oriented thinking - but I will also say that it turns out to have been a pretty hard problem to actually parse out what 'hooks' to access.
We now refer to this as "tool calling", and know that it's a specific thing an LLM (or even MLM/SLM in many cases) can be trained to do a pretty good job at.
If Apple wants to do all this on device, I think we're not far off in terms of hardware and architectural knowhow to get this done now. I would bet that the Siri team spent a number of years on alternates, essentially relearning the bitter lesson.
I'd guess the best way to make this work on device would be to train and finetune a small model on a massive number of apple app store apps so that it gets good at figuring out which apps it will want to query for tooling functions calls - that's too much to inject in context directly especially on device - then it would call out for a list of available calls -> that's just code -> then it could use a local LLM to choose the proper call and call it -> then if needed it could offload for more heavyweight thinking. To start you could offload the tool call decisions as well, or really the whole thing, except for the local function call.
This would be vastly vastly better than what we have now, and I think does not require huge reintegration efforts from app developers.
"Siri what month is it?"
[... calendar_getcurrentdate requested ]
-> [ Mar 21 2025 ]
I worked at Apple for a couple years and it wasn't uncommon to see org level dysfunction.
But Siri is such a terrible product that it amazes me that the org hasn't been deleted, and that Siri hasn't been ripped out of products.
I have seen Apple delete bad orgs a couple times before so I find it strange they let Siri survive. It does not "have good bones," it is not a good product and it never will be.
It is actively frustrating to use and I deeply regret migrating to HomeKit as well, primarily because of the usage experience via Siri.
Truly a total/spectacular failure of an org and product. It ought to be a case study for how not to product manage.
While this all falls under the same product umbrealla, I'm of the opinion that the intractable problems around Siri have more to do with the lack of nuance in the underlying APIs.
Starting a specific album at a specific song, for example... well Siri handles it, to a point, but it has to interface with Spotify, and there you hit a wall of hot garbage, along with the sequencing and process-management/integrity issues of integration software. Garbage meets a money fire, in other words. High cost, brittle, constantly changing, glue code that is mostly depending on third parties also requires session management? Not a winning product formula.
Sprinkle in some inhuman HCI nightmares like not pausing speech appropriately when reading an album title and you've got a recipe for an expensive turd that annoys your customers, oh my, by a lot.
Kind of tangent, but it is often surprising what order the future arrives in.
From the perspective of 1960, early "robots" were expected to be physically capable and mentally feeble. Good at logic. Capable of making a sandwich. Weak at empathy and whatnot.
Even from the perspective of 2025... most people don't understand how slowly robotics has advanced. Human-level performance at laundry folding remains a distant dream. Empathy is increasingly trivial.
So Siri.. and voice UIs generally. The bottlenecks have been in unexpected places.
In general, we just don't have very good UI paradigms for voice. Voice recognition is finally good. LLMs theoretically add a lot of capability. But... there just isn't a great UI.
It's like trying to use a smartphone with a nipple mouse instead of touch. You can slowly hack your way to making specific tasks/features work... but there is no radiation event where lots of tasks become possible.
Siri's voice recognition has gone down a lot over the last 5 years actually, it used to be far more accurate, its speech has certainly improved to be more natural sounding. Its capabilities though... yikes
Siri wasn’t great but it was way, way more usable before the “machine learning” update whenever that was. Before it was a verbal command line, but at least you could learn some commands.
I think the problem is adding on to a clunky paradigm.
It wasn't a good verbal command line. It was usable because even an imperfect command line UI is inherently usable if use cases are limited. Building on to that though... you always overreach.
Personally, I always thought voice+screen interface was an under-explored. That (for example) would let you go from command line to command line navigator... a natural progression.
Voice UI kind of needs to be "one shot" to be pleasant: Play "Fortunate Son." Once you have to go back and forth, listen to list of options... it's no fun anymore. The Windows/Mac/Xerox GUI paradigm required a mouse, keyboard & screen. Voice only is limiting... perhaps.
They spent more time polishing voices than they did on actual features, if anything, regressing them to the stone ages in terms of its primitive form.
Plus, there’s no more humor. Trying to crack jokes at it used to give funny responses. Even old things like “I need to hide a body” — “I don’t know how to respond to that”. Insult it and it’s a “I won’t respond to that”. Lots of old funnies are now gone and stripped away.
All of them feel to me like they went backwards. In the beginning you could say you killed someone and it'd show you the nearest park/forest to bury them in. Now you can't even get them to play X on Spotify half the time.
I guess development is driven by military and commercial needs, rather than altruistic needs.
The futuristic ads were made for the consumer to feel safe and sound.
The actual money flowing went into military applications (drones, guidance systems, etc) and commercial applications (car factories)
For the above systems, they don't have to be physically capable, just have a restricted set of actions / outcomes, for example, flying a missile. But, they need to be mentally strong, to take decisions at split seconds.
I was unaware of Apple’s military drone and missile programs. Could you point me to more information on those? I’m having trouble even imagining what an Apple drone would cost.
The thing is though, nothing progressed in Siri since the original demo in 2011. Despite boasting desktop-level CPUs and never-before-seen dedicated "neural engine" co-processors.
It has only become worse in the past couple of years.
The inability to know that bedroom means bedroom [1], or the impossible task of adding several items to a list [2], or failing to answer what the current month is [3] have nothing to do with slow progress or UI.
I just wish they'd give up the interface already and open it instead of trying to reserve it only to be used by Siri for whenever they have a Siri personality-ified LLM or if ever the LLM works as a universal router for doing anything.
At a basic level i just want the interface to have good voice to text recognition. if it's a question ask an LLM, if it's a special recognized command run it instead.
At a dev/pro-user level let me define statement patterns (regex or semantics) where i can trigger a shortcut and parse some input from a single statement.
By all means, regen the custom responses devs and users setup with the Siri personality LLM so it sounds like Siri all the time. Just set the interface free already.
Windows Vista happened and Microsoft still exists. And many other products from huge companies failed hard. I think these things are quite common for large companies. The fact that it's Apple just makes it more interesting.
Honestly, Windows Vista was a victim of the computing landscape of the time. Virtually none of what made Vista hated was changed at all to Windows 7, except by then everyone's PC had caught up to what the OS expected or they were going to stick with XP until the bitter end.
No one at Apple uses Siri, that would be my guess and it provides no value to the company. Heck I'm surprised when anyone uses voice assistance in general.
Voice assistance was the hyped thing sometime before block chain and LLMs. Everyone needed one or was working on one. So Siri got stuck into the iPhone, like Apple Intelligence is now, because "the market" needed Apple to do so to keep stock prices going up. Then the hype died out Apple sort of needed to keep Siri around, because it is one of the interface to the HomePod and CarPlay. There's no reason do develop Siri beyond that, because very few people use it, not having a better version isn't going to sell less iPhone and they can't mine the data, because they sold everyone on Apple being privacy friendly.
There's absolutely no business case in having Siri do anything beyond "Call my wife", "Next song", "Play U2".
People don't use it because when it was first introduced, it responded "I found some web results for that" for the complex types of things people wanted to ask it.
The technology exists for much better responses to those types of requests today, but who knows if they can re-train their customers that it'll work for them now.
There are definitely a lot of complicated personal-assistant type tasks that they could do now that people would like.
Does Google have an LLM based AI experience for Android? What are they doing differently to avoid the problems plaguing Apple. The core technology in both can't be that different. Or is it an expectation thing?
I feel like they may be experiencing that issue where the product is quite good within some core set of features, but the edge cases are where all the paper cuts happen. And when its a singular product on a singular device made by a singular company, those paper cuts really add up. There must be a more succinct term for that.
Can it even be solved with the current state of LLMs?
Android has it, but it has much fewer features then the previous assistant. So what's the point? Yes, they were first, tothey beat apple. Except it doesn't work.
The Gemini Assistant is pretty close to feature parity now with the old Assistant, it got a lot better very quickly. And obviously it's a lot better at conversations.
The Android assistant with Gemini is quite good nowadays, very close to parity of features with the previous one. It's not there yet but Google decided to develop it in the open, similar to the old "beta" products. Hard to see Apple doing the same.
The main issue to Apple is the different mindset that takes to deliver hardware vs. services. Delivering hardware you have one shot to "wow" your customers, and Apple keep nailing this process in a way that the whole industry can't keep up with anymore. On software, that doesn't work that well. You can totally be disrupted by a startup that delivers an MVP faster with 30% of your perfect product that would take 3 years to deliver. In this way, Google's assistant strategy in the short term was spot-on compared to Apple's. You take the hits of bad feedback in the short term in the hope that being in the game can create enough momentum to get you to a solid position of features and know-how in a couple of years.
- They don't insist on being on-device although they seem to have on-device route.
- Their promise isn't as ambitious as Apples. It's basically Gemini + pre-existing Assistant features as plugins. Not trying to be the privacy-touching agent at least at the first cut. (Or they promised big, but they have trained the audience not take it as a fade value haha.)
- They ship the Gemini version as an opt-in for the Assistant replacement so that they can learn "iterate" on the adventurous enthusiast while leaving larger user base as is.
580 comments
[ 4.1 ms ] story [ 334 ms ] threadI’ve owned Apple products forever. But as long as Pournelles Law of Bureaucracy remains on full display, these shuffling of the titanic deck chairs will continue.
I wish it could be otherwise. I wish a company could introspect on itself and say “hey, our doer/decider balance is out of whack, let’s correct” but that just never seems to happen.
"Well, um, sir, we've realigned our strategy around generative AI and..."
"No no no, tell me what you've DONE. In the past six months. It shouldn't be hard."
But lots of execs don't know enough about their own products to interrogate like that. So they settle for BS presentations.
Yes, faking that really well is basically equivalent to actually doing it for real.
Of course, there are many good and worthwhile things an exec could have done that wouldn't show up in such a demo.
Eg if you increase reliability of your cloud services from 99.9% to 99.9999% that could be huge, but most likely wouldn't show up in a short demo.
On the bullshitting front: I remember a particular re-organisation we went through at Goldman Sachs that the local bigwig was explaining to us, and all the benefits it would bring. I made the perhaps unwise decision to ask what observable measurements we could take in a few months to tell us whether the whole thing was a failure (or success).
(I suspect the actual main purpose of these semi-regular re-orgs is to shake things up enough so that after a few shuffles person A can eventually land ahead of his former boss B, without anyone ever losing face. And that's a good thing! But hard to admit to.)
For Apple it probably made sense, because the whole company's image and reputation is built on that stick.
But eg for a toilet paper manufacturer or a producer of fighter jets, the CEO shouldn't spend too much time personally testing the products.
For the former, because presumably the product doesn't change that often.
For the latter, because there's not even a single 'user' of the product. The experience of the huge ground crew (with various specialised roles) is just as important as the experience of the guy in the cockpit, etc. No CEO, and actually no single person on earth, has the expertise to judge all of these aspects of use by themselves.
That doesn't mean the experience of the users is irrelevant. Just the opposite! But the CEO will have to intelligently delegate.
It’s a two way street. Managers can only see the forest, but engineers are capable of seeing both the forest and the trees.
In contrast, a "discusser" or "decider" makes decisions in order to satisfy the social role of making decisions, but often with a lack of surety, clarity, follow-through or commitment. Perhaps in fear of missing some greater opportunity, or fear of being credited with some failure, their decisions are not organized in a way that actually effects some end.
Summarise a meeting here. Summarise the report there. Summarise the email from marketing. Schedule a meeting with Jim. Turn the lights on in the kitchen for the housekeeper.
While yes, there is some usefulness in having something summarise things for you, most people in their day to day lives don't handle a "100 emails from everyone" and "30 meetings with 40 different department heads" or whatever these people do.
Quite a big gap from the market they are targeting with Siri.
Isn’t that what administrative assistants and hired help do? To handle the minutia of triage and actually doing the stuff you’ve decided upon? And they actually have to do their job well because (reputation loss, employment, and legal)?
I guess not... there is a reason Apple is still in business, they are great at deception.
It just seems so silly to believe a blackbox company that was outed by Snowden in PRISM would actually be privacy focused. Or that a company that touts security has been hacked so easily by Pegaus that anyone even buys into the marketing.
But hey, people are religious, so I should be less surprised that people trust others.
Are you serious? Did you miss the headlines around their news and text message summary products?
Actually, riddle me this: Do you really think they would just ignore how that went?
I think many inside and outside Apple hoped that the ways that they scoped their features in the Apple Intelligence announcement would help them pull off something duly reliable and practical, but it's not that surprising that even those ambitions might have bought too deeply into the hype.
It's very well-known that Apple is perfectionistic. I'm not meaning to say that perfectionism is a negative quality or a bad thing, just that it takes a while.
Apple's development process is a marketing pitch-driven hallucination - project management by buzzword and individual career status progression.
It's almost entirely inward-looking. The connection to Rest of World is increasingly mythical and remote.
Some good work gets done in spite of this. But senior management doesn't understand quality - either in the internal sense of having bug-free robust product, or the design sense, where products meet real user needs in a satisfying, creative, and delightful way.
Nice graphic design though. Apple is still the leader there. Processor dev has also been exceptional.
IMO it's time for most of the C-suite to step down and let much younger talent take over and shake things up.
My guy, Apple is autistic as shit. Just look at how in-house everything is. Every time Apple runs into someone doing it wrong, they do it themselves. You know the saying; if you want something done right...
Sure, they're performative as well, but that doesn't change that they are still in the process of doing absolutely everything from scratch because the existing solutions are not good enough for them.
I don't know if you've used the new Apple Silicon MacBooks, but I have a 128GB/8TB one lined up for me to pick up from the Apple Store in about an hour. It's certainly not the best that Apple offers (that would be the new 512GB/16TB Mac Studio) but it practically wipes the floor with every other laptop on the planet. Because Apple made their own chips, because everyone else was doing it wrong.
I can put words into Steve Jobs' mouth as well!
Steve Jobs would reframe LLMs (and other ML-based solution) as creative assistants, insisting that their job is not to get the right answer, but to help you get your creative juice running. To get this to happen, he would have personally convinced engineers that have made cool AI demos to work for Apple to turns these demos into features that form a new Apple creative suite.
If Apple released OpenAI's voice mode, without calling it "AI," referring to a "GPT" or a "model" -- if it was just integrated into the iPhone with a wake word, that absolutely would satisfy Jobs.
The problem with these companies is that they can't visualize how a product works for non-technical customers any more. Because everyone they know lives and breathes Silicon Valley. They see billboards for vector databases on their commute, so they think it's perfectly normal to name a product "GPT."
A product containing a language model should never, ever be called "AI" or "GPT" or even "Intelligence." 10 years ago, the only people who knew what the term "AI" even was were the nerdy readers of pulp scifi novels. I'm half joking but the whole town of silicon valley needs their glasses smashed and to be shoved into a locker.
Depends what you mean. Building an LLM seems to be easy enough that several companies have built one, and even Twitter was able to magic one out of the ether from pretty much a standing start given the resources.
Getting an LLM to do the right thing for you 99.999% of the time? Has anyone managed that yet? Frontier models still hallucinate the most basic shit, and consistency is very, very difficult.
Meta, OpenAI, and others out-compensate Apple for ML talent by a factor of 2-4x.
Especially after China has entered the game and DeepSeek disintegrating most of the moat that existed around those companies.
And Apple can simply use what others created instead of burning billions into R&D which might not come up with anything noteworthy.
John "JG" Giannandrea was hired over to Apple around 2018 to head Apple's new AI/ML division, which would include the Siri organization that predates Giannandrea's tenure at Apple. Giannandrea was previously the executive in charge of Google's Research division.
Mike Rockwell was hired into Apple from Dolby Labs around 2015 to head a group called "TDG" that would be the R&D and the product group for what we know today as Apple Vision, an AR/VR headset. The same organization had its own AI/ML applied research teams focused on computer vision problems, and had hired in folks from Microsoft's Hololens team.
From what the article describes, the Siri organization is being moved from Giannandrea's scope to sit under Rockwell, and Rockwell will be moving away from the Apple Vision organization.
Apple hired Giannandrea to build an organization like the one he led at Google, and appears to have found that that the AI/ML organization built at Apple struggled to manifest its work into successful product wins despite the massive financial investment. It's worth noting that at Google, Giannandrea was succeeded by Jeff Dean, and the organization was more recently reorged to become the "Google Deepmind" we know today also had the same struggles.
Furthermore, the challenge with AI is that even though it can deliver some shockingly impressive results, it also generates some bafflingly stupid responses, as well as answers that are seemingly correct but are never less wrong in ways that are hard to determine. Those hallucinations really shatter the illusion of quality and reliability, which ultimately undermines confidence in the functionality as a whole.
Given that, Apple is stuck trying to catch up to the rest of the products in the AI space while also realizing that everyday AI is nowhere near as polished as Apple would expect it to be. So the more Apple pushes AI, the more customers are likely to pick up on the errors and inadequacies of the product.
Even though other AI assistants suffer from the same fundamental problems, the ones from Apple Intelligence are going to seem more embarrassing because of Apple’s brand image.
Oh, of course. Like...? Nothing else has access to relevant data. That's like 95% of the pitch of siri.
Voice assistants are just fundamentally embarrassing to use, imo. The only exception of which I can think is while driving.
> Nothing else has access to relevant data.
Simply not true.
Which means Google or Apple. Or Microsoft on the business side.
Nobody sane is going to share all their personal data (about everything) with a fly-by-night random AI app/company.
Apple has time, in the same way it did with Maps, because despite the marketing push no one is buying phones solely because of AI features these days. They're still an "Oh, that's nice."
They absolutely will if the tech giants refuse to work with each others' data silos.
Its doesn't mean that each product doesn't have something a bit special but competition caught up long time ago, sometimes went ahead (better batteries, bigger cameras etc). These days there is much more 'yawn factor', and its across whole industry.
I don't remember them ever doing that. Usually they start with something lacklustre then incrementally improve it until it is strong.
Maps is a great example, it was poor on launch compared to Google Maps but has now got to a level where most people don't bother with Google on the iPhone.
Yes, the iPhone put it in everyone’s pocket, but if Apple had built a handheld device that was powerful enough to run any interactive mapping tool Google Maps was already the obvious choice.
It’s a minor quibble, but to say Apple played a majority role in Google Map’s success doesn’t ring true to me. In 2007 Google Maps was easily overtaking all other mapping tools with or without the iPhone.
Apple Maps is also pretty good here in Canada, although I would never trust it to tell me with any accuracy what a given business's hours are, or whether it even still exists. Google is much better at that.
What is saving them this time around is the way they have been priting money with iDevices.
Which are historically the hardest problems for engineering orgs to ship: precisely because they cut across team/org boundaries.
Suddenly the Maps team need to give a shit about the Calendar team, etc.
And both teams tend to exaggerate and say 'Sure, our side of the integration is finished... it must be a problem on their side.'
Ironically, Amazon-as-a-company probably has the most experience retooling this way -- from the Bezos API memo. Because that's what it will take.
working in AI/ML, is there any recommended literature one may read on how to ensure efforts result in products? Would be interested to access more experience to benefit both an engineer and a manager pathway.
Apple had teams involved with AI/ML in several divisions also, from AI/ML (including Siri), TDG/VPG, SPG...
I would contest this statement very vigorously. Siri has been a disappointment but nearly every single Apple product has a load-bearing reliance on an AI/ML feature.
I'm continually frustrated by the (relatively recent, since LLMs) tic where just because a ML model isn't conversing with you or otherwise making its presence impossible to ignore, that it isn't delivering a massive amount of value. It is frustrating from laypeople, it is extra frustrating from industry insiders.
Just a few examples off the top of my head:
- AirPods with ANC and spatial audio. Both headline features of these products that are 100% AI/ML.
- Watch with heart arrhythmia detection, automatic workout detection, fall detection, etc. All are headline features (some literally life-saving) and are 100% AI/ML.
- iPhones have high-profile features that are 100% AI/ML: automatic car crash detection. Others are more subtle but IMO substantial differentiators - such as automatic image enhancements out-of-camera.
Again, I know in the age of ChatGPT we seem to have twisted ideas of what ML is, but "AI/ML" is not synonymous with LLM.
2) The examples given may have been produced by AI/ML-based technologies developed at Apple, but may not have been the work of Apple's AI/ML Organization. Many "groups" (divisions) within Apple have their own teams using AI/Machine Learning (e.g. AI/ML Org coexisting with AI/machine-learning work and teams in TDG/VPG, SPG (special projects group: the one previously working on a car), etc.
https://news.ycombinator.com/item?id=43425082
It's just that most Apple software seems to suck in some fundamental way right now. I don't know if it's a technical issue (SwiftUI being meh when compared to UIKit for example) or a culture issue or the money coming in insulating management from accountability. Software execution has been lagging behind the excellent hardware execution for almost all of the Tim Cook era. They desperately need someone like Scott Forstall to come in, kick butts and get stuff going again.
They ideally have a couple of years while waiting for Moore's Law to catch up to turn around their software side. Otherwise, it's a real shame that all that great hardware is just being used to run Electron B2B SaaS apps.
I challenge you to find a 2014 recording that is on par with a contemporary Siri voice.
Alex was developed when VoiceOver (the screen reader) was the primary use case for text to speech. Consequently, it was optimized for low latency and robustness under rate changes.
The Siri voices sound much more natural at 1x and have a higher signal quality, but rate changes were a lower priority for this use case.
Fun fact: when we worked on Alex, many VoiceOver users stubbornly hung on to Fred (which is mostly using late 1970s technology). Screen reader users are not fond of switching voices; it appears their hearing locks in to a particular voice, so switching is costly.
> Recent advances in text-to-speech (TTS) synthesis, such as Tacotron and WaveRNN, have made it possible to construct a fully neural network based TTS system, by coupling the two components together. [...] However, the high computational cost of the system and issues with robustness have limited their usage in real-world speech synthesis applications and products. In this paper, we present key modeling improvements and optimization strategies that enable deploying these models, not only on GPU servers, but also on mobile devices
It would take very little effort to put an LLM frontend on top of this, and yet they've not only abandoned applescript (or the underlying apple events) at a time when they could expose it to the masses, but have gone in the exact opposite direction with "Shortcuts".
Oh and the icing on the cake is that apple events can be sent over the network as well, and this infrastructure has existed since the early days of OSX.
Like you say, an MCP server integrated with AppleScript/Apple Event Manager would instantly hook up any LLM with virtually all Mac apps. More Mac devs will then be incentivized to support these features. For people who find AppleScript un-intuitive, JavaScript is also supported. And in my view, this is a revolutionary way to use my computer - a very Star Trek way of using the computer.
They have an advantage when it comes to running them locally, but it feels like they're trying to push it onto consumer hardware before consumer hardware is at the point of actually being able to run useful LLMs.
But that's why I think they have a couple of years to sort out their software issues. When useful models can be run on their devices they have to be ready. The hardware advantage can only be an advantage when they have the software to run useful applications. Hopefully they don't get stuck in the typical big company bureaucracy and ego matches and instead can make a change for the better.
If last year they would make each macbook air as a standard have 32GB RAM and iPhone 16GB RAM + 250GB SSD as a base they would have the most capable hardware with big user base.
Sure they loose some money of people having upgrade models but they would sell much more Macs. As a reference they sell each year only ~20M Macs comparing to 60M ipads and 240M iphones. Macs are having only like what ~10% market share worldwide? They could easily double it but they protect their profit margin like virginity.
You're getting product segments mixed up. From what I can tell, the 13" MacBook Pro in the Intel era never started at 16GB; the last model still started at 8GB. That's what was replaced by the M1. The 15/16" Intel-based MacBook Pro models didn't get a proper replacement until the M1 Pro and M1 Max, which started at 16GB and 32GB respectively. The only regression I can find there is that the last 13" Intel MacBook Pro could be configured with up to 32GB, which wasn't available from the base M series chips until the M4 last year.
It surely is just your opinion. Nvidia is king and Apple has found a way to market an integrated GPU and CPU RAM as something magnificent, rather than something that has existed since the dawn of computing.
There is a reason Nvidia is king. There is a reason corporations buy Nvidia and not Apple for their LLM uses.
In the age of LLMs I also increasingly view my experience with "Hey Google" and "Alexa... " with similar contempt. Their output feels so bad compared to what ChatGPT's voice gets me. All of the big tech companies seem to be struggling immensely to deploy the latest in AI advancements.
"go home" opens maps app with a route to my home.
"alarm in", "alarm at", "what time is it?" works well too.
In fact I'm afraid they'll break it by shoving down Gemini.
Alexa is the same with basics, works well for me with "reminder in X hours", "reminder at 10am on every workday", "alarm at", "play The Strokes", "add coffee to the list" (groceries)
Alexa is really bad with any advanced stuff I've tried. Didn't try with "hey Google"
Any other basics that work?
i have a pixel 9 with gemini. i absolutely hate it. google assistant is miles ahead.
and btw, i'm not the only one complaining https://www.reddit.com/r/GooglePixel/comments/1jdpvnq/gemini...
But it's only a matter of time until some exec removes it to "improve Gemini adoption".
https://www.androidauthority.com/google-killing-assistant-35...
It also doesn't help that ChatGPT's voice cadence is much closer to sounding like a real human.
Siri meanwhile is still far behind both Alexa and Google Assistant in its basic voice-to-text conversion capabilities -- a pain point I experience everytime I talk into my Apple TV remote to search for content.
And Siri will continue to be dumb regardless of whatever "AI" monicker they slap on it.
https://old.reddit.com/r/iphone/comments/1jehkpm/apple_intel...
Meanwhile LLM's came along and shifted peoples expectations.
Just tried this on Android, and it works (though it gives the full date)
Whether the fault lies with individuals or the structure of the organisation is a more interesting question. I wonder if American corporations would take a more individualistic approach to this compared to their European and Asian peers.
It seems like most tech corporate structures are to figure out the appropriate place in the org chart to wield the axe, so long as it's lower down.
you’d think that would be a lesson well learned by now
Because now the perception is that "Apple Intelligence" is something distinct from the state of the art.
There are far more bugs and issues that are to be fixed (but don’t get attention) than Siri. Anyway Siri has been close to useless for a decade and a half. It’s not a big deal if it remains so for another half a decade. Nobody outside of Apple has any expectations on it.
Edit: On a related note, this leak seems to be significant. The Apple I know of would fire the executives (among the “top 100”) who leaked this and probably make sure they’d pay some hefty monetary penalties too.
I had a Galaxy S8. It definitely did.
> Unlike iOS,there are any number of music player apps on Android that stay out of the way and don't try to be a "platform".
iOS has other music players too. It was a choice that e.g. doubleTwist stopped supporting iOS when pivoting to CloudPlayer.
Let’s define “ad”.
https://www.howtoisolve.com/how-to-turn-off-or-disable-apple...
It used to work in the first few versions of Music.app but broke about 2 years ago and has never fully disabled it ever since.
Like you say, Siri is borderline completely useless. Just timers for me.
Even just the most basic LLM integration, without even giving that LLM any personal data would be a massive improvement. They even have this ChatGPT integration that Siri seems totally incapable of using in any meaningful way.
I don't think Apple will ever use LLMs, given the offensive output they can generate sometimes. There is just no way to make them safe for all audiences.
Well, apparently some people have at least some expectations, after Apple pushed ads touting Siri's new abilities (and Apple AI) but failed to launch them on time, and now Apple been hit with a federal lawsuit against them: https://www.axios.com/2025/03/20/apple-suit-false-advertisin...
Edit: I wonder if this will force them back to live events? Live demos are at least more believable and might win them back some trust.
My wife and I were watching Jeopardy late at night, and Pol Pot came up. I asked Siri something like "Siri, tell me about Pol Pot" and instead, it heard "Siri, Call Scott", and rang my friend at 1am.
Do better please! Where is the passion, catering to pro’s, any idealism? Hardware is great but software gets worse and worse..
It died 14 years ago.
I think Apple started to die with the iPod and got worse the more successful it became.
Then again, maybe I always prefer the underdog.
Also the classic MacOS where you (the user) had to pre allocate memory per application and where holding down the mouse button stopped any application from running in the background.
The resuscitation was truly miraculous.
It was very cool tech. Closer to what Apple would get from NeXT years later.
My guess is as to why it failed (and this was also the common sentiment at the time) — it was an optional install and not a default. It's hard to imagine anyone writing an app around a component (extension) that few people had in the OS.
And I just thought about the other abandoned tech that came with System 7.1 Pro - PowerTalk.
My next Mac was a 2008 Macbook Pro (non-unibody) that had backlight issues out of the box, failed display cable two years in, and then the GPU failed. (Nvidia plague era, and Steve Jobs silence plague era of shit-ass fan curves)
To this day I still refuse to trust Apple's fan curves and install fan control software. Especially the M1 iMac with the cut down graphics that they only gave one fan and a smaller heatsink. Watched it break 100C regularly.
The m1/m2 mac mini might be the only one I've never had issues with the stock fan curves so far, but I still bump them up anyway for peace of mind. I don't see a reason to let an SoC hit 98C even if its rated to do so.
At the end of the day, I was just not that impressed with the x86 based Macs. They had all of the drawbacks of x86 with all of the added disadvantages of Ive inspired design which led to poor heat dissipation, not enough ports and butterfly keyboards
Bluetooth has been broken on both editions of their Pro Max Phone (15 & 16). It cuts out and has all kinds of weird issues, but no fix has come for years now. This was why I always paid a premium for their phones, they did hardware and software well. Not anymore, their so focus has been on their processors.
This tends to be an Apple issue. They introduce proprietary thing and eventually unless your hardware supports said thing it will have a degraded experience in some way
Apple has the MFI program for these things. You can get certified and tested in relatively short order.
They also aren’t removing the original protocol support or anything like that.
What I will say is that I think they don’t devote as much QA resources to getting it right as they do their own protocols, hence the difference.
With Microsoft it was very different. They wanted to subsume or shutdown whatever they targeted with EEE
This is such a trivial big win as well, create a software dev team to focus on the outstanding bugs. Head it with one or two senior developers, make it the first step in onboarding new developers. What a great way to learn, and make a difference.
This is not rocket surgery, and Apple really do have a bad reputation here, even with the muggles.
Sounds nice on paper. In practice, the seniors assigned to the team will see it as a career dead-end (because fixing bugs doesn't get management attention, but flashy new features do), juniors lack the experience to avoid edge cases or to navigate the project structure, and the seniors/intermediates on the "actual" development team don't have time to waste on reviewing and aiding code of juniors and bugfixes because then they have ownership over that...
This is a myth, though a myth created by Apple itself and Bertrand Serlet's tongue-in-cheek keynote slide. Snow Leopard actually had a number of new features: https://en.wikipedia.org/wiki/Mac_OS_X_Snow_Leopard#New_or_c...
Moreover, Mac OS X 10.6.0 was much buggier than its immediately predecessor Mac OS X 10.5.8. There were some really bad bugs in those early Snow Leopard releases.
The Snow Leopard that everyone remembers and loves was not version 10.6.0 but rather version 10.6.8 v1.1, which was released almost two years after 10.6.0.
Major OS updates invariably introduce more bugs than they fix. That's an iron law of software development. What Apple used to do, which they no longer do, is spend a long amount of time just fixing bugs in minor updates to the operating system. Now, unfortunately, Apple is on a rigid yearly OS update schedule. Worse, Apple doesn't just fix bugs between those yearly updates; instead they keep introducing new features even in the "minor" updates. So we never get a "stable" version. It's constant change for the sake of change.
Because they still managed to fix all the egregious bugs by August every year, as late as Mojave in 2019.
From my recollection at least.
My argument is that they were never building and have never built software at a sustainable pace, even before the yearly cadence. They race ahead with tech debt then never pay it off, so the problem gets progressively worse.
A while back, that merely manifested as more and more defects over time.
More recently, they began failing to ship on time and started pre-announcing features that would ship later.
And now they’ve progressed to failing to ship on time, pre-announcing features that would ship later, and then failing to ship those features later.
This is not the yearly cadence. This is consistently committing to more than they are capable of, which results in linear growth of tech debt, which results in rising defects and lower productivity over time. It would happen with any cadence.
And I also remember 10.7 actually being better and where I thought Mac OS X (and its macOS rebrand) peaked, for interactive/workstation use.
However I ran a 10.6.x virtual machine (that yes, ran Mac OS X 10.6.8 Update Combo v1.1[1] for the last decade of its service life) until 2021 for a bunch of long-timeline server functions, so I guess Snow Leopard was pretty good overall, in the end...
[1]: https://support.apple.com/en-us/106449
I took it into the Genius bar and they had nothing other than "we will wipe it and get it working."
It turned out I was able to boot to single user unix console mode and clean up the issue in an old school style. I have no idea if that is even possible anymore, but in any case I have never had another issue with OSX as severe as that one.
Modern Mac OS got really quite high quality at some point but it's easy for people to forget it had a lot of growing pains between 1999-2008 or so.
I was doing hardcore computation 24/7 on a MacBook, including model checking real-time systems, and I vividly remember the exact day I experienced one kernel panic.
The rest of the time, everything was rock solid. It's sad both quality and aesthetics degraded during the 2010s, where the system somewhat converged to iOS UI patterns and lost skeuomorphism.
Apple did skeuomorphism really well, which is hard. No idea why they moved to flat designs, which are easy to copy by competitors.
This is far from isolated to Apple, too. It's a cancer that's infested virtually every single software company in the entire world at this point, and it's endlessly frustrating. If they don't have some New Shiny Version to show off to the tech media and blogosphere everyone starts acting like the software itself is a dead man walking. I also think it's partially down to all these same companies trying to justify subscription billing with a constant trickle of "new stuff" that the application in question does that you're meant to feel like you'd be missing out on.
I don't ascribe to this notion whatsoever. I in fact far prefer my software that barely changes at all over time. My most reliable and useful computer is my huge Ubuntu box that lives in my laundry room which handles... I mean frankly it would be easier to say what it doesn't handle, which is anything Apple adjacent because that's all proprietary, and it doesn't handle my security cameras because I couldn't find a linux-based solution for that. Literally every other automated function outside those two categories is handled by that machine, with a mix of software I've downloaded and configured and a fair bit that I've written myself.
I say all this to say that machine barely changes at fucking all. I do security updates of course and occasionally have to tweak something but the apps on it are the same, it serves the same internal websites, it runs the same scripts, like clockwork. I LOVE that thing and my life would be in an utter shambles if some overpaid tech executive had the power to randomly change how it works to justify his salary.
With further progression on integration/miniaturization and lower power consumption/heat output for general computing, the devices have shrunk to their bare minimum as slabs, there's very little physical to display so the fashion needed to move to software (and what online services it can be a client for). Stasis is death when you rely on selling, and maintenance is a constant cost which no one really wants to pay for.
As a side note, it's interesting to compare attitudes to the physical side to PC, where for years there was almost a pride that PC was a minimalist functional box (besides the occasional fun/weird exception), but that contrasts against now where it's hard not to get components with windows or embellished with LEDs and logos on display (and there's another weird contrast, where on one hand you have internals on display, on the other trying to minimize or hiding away the cables as though the inner workings were ugly)
The strangest thing about this is in the system configuration dialogs where they have a set of really gee whizzy modern configuration screens but you still have to go into 15 year old screens to do many common tasks —- they updated some but kept the old ones that they did not have time to update which is a little ugly but fine because they old ones work 100%.
I used to have “who moved my cheese?” moments all the time with Windows circa 2005 when I was admining Windows and having to deal with configuration screens in ‘98, 2000, XP and various versions of Windows Server, they couldn’t resist making superficial changes that annoyed me to no end.
I don't know who if anyone really is heading the UX department at Microsoft, but they should either be empowered to their job properly or fucking fired, and I couldn't tell you which.
Microsoft released Windows 8 in 2012, and here we are 13 years later using interaction panels that have only gotten minor chrome updates since their original release in Windows 95 (looking at you, Advanced System Settings and Device Manager).
On some level it's an embarrassment, on the other level I never run into a "can't get there from here" experience though I had an awkward experience the other day when I brought home a bargain monster TV from the reuse center and found I had to look across a several different screens to find all the settings to get the video output properly matched to the TV.
But the thing is functionality is already being impacted! I had an experience the other day where I was trying to manually configure an IP address for an ethernet interface to configure a piece of hardware, one of those situations where it lacks DHCP but I knew what the IP scheme was and so I just had to configure an address, gateway and subnet in order to be able to talk to this damned thing, which I was eventually able to do. But, crucially as I'm trying to do this, Windows pushes me towards a Modern/Metro UI version, which, fine, whatever, but when I entered my information it would then show a series of "in progress" dots across the bottom of the screen, presumably because it was attempting to verify it had internet access on this interface, which it obviously didn't and wouldn't, but it did have internet access via the WLAN card. This contradiction seemed to confuse the shit out of the Settings app, as it presented no error and eventually seemed to just... shrug and do the thing I'd asked.
Except it didn't work, and I spent an hour troubleshooting why I couldn't yet communicate to the thing, only to eventually out of exasperation go to the control panel to see if the old UI version of this (right click interface, properties, ipv4, properties) only to find out it NEVER SET THE ADDRESS I SPECIFIED AT ALL! Presumably because it couldn't verify it now had internet with those settings, or maybe because the UI was just broken, I don't know.
(Also interestingly CMD thought it had the address configured, but no subnet or gateway, and the old UI had none of it and it was still configured as DHCP)
And like, this is just bad software design, bordering on incompetent software design, for several reasons:
* We have an in-progress "I'm doing something" indicated from the OS, and I'm assuming it was in fact doing something, but it was not indicated at all what was being done, or why, or the result of what was done or what went wrong, which is simply horrendous.
* It was, I'm guessing, attempting to verify connectivity, but that was not explained to me, nor was it presented as an option as it is in the old UI (there's a checkbox that says "validate this when I'm done" and you can leave it off, which I did which then worked as I thought it would)
* Somehow, these three functions to show an interface's state (CMD, old UI, new UI) are all producing different outputs to one another about the same interface, which is just BONKERS in terms of software designed to manage a system, and implies severe fragmentation in the core of the OS.
So, just so I'm understood, I don't think Windows 11 is embarrassing because it has different UI conventions present simultaneously. I think it's embarrassing because of what that fragmentation means for lower, more important parts of the OS. It's a less important thing that points at a more important thing.
I much prefer the older screens. They are so much easier to use.
The new version of the OS was always being developed in a branch/train, and fixes were backported to the current version as they were found. They weren't developed linearly / one after another. So yeah, if you are comparing the most stable polished/fixed/stagnant last major version with the brand new 1.0 major version branch, the newer major is going to be buggier. That would be the case with every y.0 vs x.8. But if you are comparing major OS versions, Snow Leopard was different.
Snow Leopard's stated goal internally was reducing bugs and increasing quality. If you wanted to ship a feature you had to get explicit approval. In feature releases it was bottom up "here is what we are planning to ship" and in Snow Leopard it was top down "can we ship this?".
AFAIK Snow Leopard was the first release of this kind (the first release I worked on was Jaguar or Puma), and was a direct response to taking 8 software updates to stabilize 10.5 and the severity of the bugs found during that cycle and the resulting bad press. Leopard was a HUGE feature release and with it came tons of (bad) bugs.
The Apple v1.1 software updates always fixed critical bugs, because:
1. You had to GM / freeze the software to physically create the CDs/DVDs around a month before the release. Bugs found after this process required a repress (can't remember the phrase we used), which cost money and time and scrambled effort at the last minute and added risk. This means the bar was super high, and most "bad, but not can't use your computer bad" bugs were put in v1.1...which was developed concurrently with the end of v1.0 (hence why v1.1s came out right away)
2. Testing was basically engineers, internal QA, some strategic partners like Adobe and MS, and the Apple Seed program (which was tiny). There was very little automated testing. Apple employees are not representative of the population and QA coverage is never very complete. And we sometimes held back features from seed releases when we were worried about leaks, so it wasn't even the complete OS that was being tested.
A v1.1 was always needed, though the issues they fixed became less severe over time due to larger seeds (aka betas), recovery partitions, and better / more modern development practices.
Do you mean Mac OS X 10.6.8 v1.1? Otherwise, I don't know what you mean by "v1.1".
In retrospect, I'm guessing that you meant N.1, e.g., 10.6.1. But 10.6.1 certainly didn't solve all of Snow Leopard's problems. Like I said before, the "stable" Snow Leopard everyone loves was the result of almost two years of minor updates.
> you are wrong(ish).
> So yeah, if you are comparing the most stable polished/fixed/stagnant last major version with the brand new 1.0 major version branch, the newer major is going to be buggier. That would be the case with every y.0 vs x.8.
In other words, as you admit, I'm 100% right. ;-)
Every new major OS update introduces new bugs, that didn't exist in the previous OS, some of them horrible, and Snow Leopard was no exception.
Remember this one? It's hard to find a worse bug than total user data loss: https://forums.macrumors.com/threads/snow-leopard-bug-respon...
I still believe you are right and wrong (which is why I put the "ish" on wrong).
"Every new major OS update introduces new bugs, that didn't exist in the previous OS, some of them horrible, and Snow Leopard was no exception" is not what folks are noticing when talking about Snow Leopard. People don't say Snow Leopard was great because it had no bugs, they say it was great because it had fewer bugs they experienced at release and converged quickly to something that was very stable, much more quickly than previous releases (and Leopard was a particularly long stabilization cycle right before it).
I do remember that bug you link, and I remember is was via migration assistant and required some convoluted steps to get in that state (IIRC it largely cropped up with people migrating their home from extremely old versions continuously, which was one of the QA tests we actually did but this issue was missed). There was much debate as to just how widespread it was (Apple didn't have telemetry at the time and had to rely on forums and apple care reports). The majority of people had no issues with SL, which was not the case with Leopard. Leopard's development internally was also more chaotic FWIW.
During that time period my team and I triaged every single Mac OS X bug coming into the company every morning. Trust me, SL was of higher quality than Leopard.
Nowadays, shipping is less of a major event due to CI, automated tests, betas, no CDs/DVDs, recovery partitions, etc. We're in a much better world!
Could you clarify what you mean by "ran"? Originally I thought you meant that you were the manager in charge of the 10.6.1 and 10.5.8 releases: "the person who personally ran" is singular. Whereas now I'm getting the impression that you were just one of many engineers who installed and used internal builds of the updates?
> People don't say Snow Leopard was great because it had no bugs, they say it was great because it had fewer bugs they experienced at release and converged quickly to something that was very stable, much more quickly than previous releases
That's precisely where we disagree. My impression is that people are just looking back fondly at the later versions of Snow Leopard.
> (Apple didn't have telemetry at the time and had to rely on forums and apple care reports). The majority of people had no issues with SL, which was not the case with Leopard.
These two sentences seem a bit contradictory. ;-)
> Leopard's development internally was also more chaotic FWIW.
Externally too. Leopard was infamously delayed for iPhone.
> Trust me, SL was of higher quality than Leopard.
Well, from my perspective, "Leopard" and "Snow Leopard" are a bit amorphous. They are brand names, not specific pieces of software. 10.6.0 was definitely not higher quality than 10.5.8, as I think you've admitted already. And that's precisely why I object whenever people say, "We need another Snow Leopard". It's never the major releases that create quality, stable software but rather the long train of minor bug fix updates that occur afterward.
By the way, I also remember an NSURLConnection crasher that was introduced in 10.6.3 and IIRC plagued us for a number of 10.6.x versions until it was finally fixed. https://openradar.appspot.com/7841731
I certainly admit that Snow Leopard was less ambitious than Leopard, and thus 10.6.0 likely introduced fewer new bugs than 10.5.0. If you have to ship a major update, then obviously fewer new features is going to create less disruption than more new features. But the fundamental quality problem today is that Apple forces itself to ship new major updates every year. That's completely unnecessary, a self-inflicted wound. If you look back prior to the annual schedule, the major Mac OS X releases came 14, 18, 30, 22, and 23 months apart. That's much more reasonable.
> Nowadays, shipping is less of a major event due to CI, automated tests, betas, no CDs/DVDs, recovery partitions, etc. We're in a much better world!
You cannot be serious! As far as I see, Apple software quality is worse now than ever.
And just to compound the problems, we never get a "stable" version anymore, because Apple continues to ship new features throughout the year, even in "minor" updates, so even 11 months after the .0 release, we still get the bugginess, and after every .0 release Apple has to turn around relatively quickly and start work on the next major version in order to be ready for the next WWDC in June. It's a vicious cycle.
If anything, CD/DVD installers were good, because only the most eager users ran to the retail stores to buy them and jump onto the bleeding edge new versions, and bugs could be fixed before the majority of users eventually installed the major updates, whereas now Apple is pushes users hard to install major updates over the air immediately at .0.
I cannot think of anything that less needs yearly major updates than macOS. It's a mature, 25 year old OS that, unlike iOS, is used for an enormous range of professional work. I need to it be fast, stable, and predictable. The last thing I want is change for the sake of change, which largely describes the last 5-10 years of the OS.
Wanna make Notes.app compatible with the new features in this year's iOS? Just update Notes. I don't need a whole new OS to get new features for a notepad.
Having said that, here's the reality of bugs that are assigned to a team. Feature work comes first. The (older) engineers don't want this — they want to fix bugs, refactor code, get rid of tech debt. But ever since Jobs returned to Apple, it's become top-down and the engineers are no longer driving the ship. Management want the features.
So, the slippery part is this: once you defer a bug for an OS release and ship your product, deferring that bug in the next OS is a no-brainer. You already shipped with the bug, why is it so important this time around? So users/devs learn to work around the bug.
Occasionally, as an old-timer, I would on a whim just fix an arbitrary bug that was "bugging" me. Sadly though, the process had become so top-down though that I would also need: 1) someone to code-review the changes (and so already you are encountering your first obstacle requiring "buy in") and then 2) get permission from "management" (this might be the BRB, bug review board) in order to submit the fix and the Radar. (There are periods though when they are lax and bugs can be submitted with little justification — or we would just make up some bullshit to get the bug fix in).
I recall the early days with fondness when a dev just checked fixes in.
Me: Craig, are you working on the Crop tool in Photos?
Craig: Yeah, why?
Me: How does it pick the default for whether to use Portrait or Landscape? Half the time I think the default it picks is wrong. Why not look at the image bounds and default to Landscape if it is wider than it is tall?
Craig: Good idea. I'll make the change now and write a Radar for it later.
Also, a shout out to Brendan who would push back when in a feature meeting with something like, "Okay, which of these features can we drop? Because we have some serious technical debt we need to address that was not own your feature list."
Were you using Macs before Jobs came back? The entire operating system for PowerPC based Macs was technical debt with parts of the OS running under emulation.
It was a buggy unreliable crash prone mess.
I think he was :)
https://en.wikipedia.org/wiki/Glider_(video_game)
Yeah I had that game on my Mac LCII back in 1992 and Parena was great too.
The point still stands though. The operating systems were horrible back in the mid 90s and both Macs I had back then - the Mac LCII and PowerMac 6100 - were cripple by being a half speed bus. Back before it was necessary.
As an engineer before Jobs came back — and for a few years too after he was back — there was still a sense that engineers could call the shots with their own frameworks/apps. Copland, Pink, all that — you can dismiss them as failures but they were engineering trying to toss off the "technical debt" of no true multitasking. So, again, it was bottom up and was at least a joy to be an engineer working on the OS at the time.
As I say, for a few years Jobs let engineering call the shots as NeXT became integrated in the OS. I was on the Graphics team then and a whole new graphic architecture and window server were created more or less from scratch (heavily borrowing from NeXT of course since most of the graphics engineering team were NeXT).
At some point though the major changes were in and management began to take over. "Quality" was eventually measured in unit-test code coverage, for example. (Sigh.)
If you were only using Windows you barely noticed, but if you used much Unix Mac OS got pretty painful. I mostly remember this from taking a digital music/video class and that was basically the only thing I used Macs for at that time. You never really got through an hour or two editing video or audio without a crash or two. We had Avid workstations where they controlled this by tightly controlling the software on the Macs, but I never actually got permission to use them.
There were also behavior patterns. If you only used Mac you were in a behavior pattern where you'd crash the machine less. If you were using Unix workstations a lot your behavior subtly changed and you tried to do more stuff at the same time. If you then went back over to the Mac and overloaded the Mac the same way it was a bad time.
That’s a bug not a feature. Engineering led companies rarely ship good consumer products. What happened to Apple before there was a product guy at the top is a prime example.
Another recent example is Google. Google has the best technology in the industry but can’t ship and sustain a good product to save its life.
I should qualify it somewhat though — I mean I was, for a spell, working on the ColorSync framework. So the "customer" was also an engineer using our API.
At the same time, when I worked on Preview initially I added features that I wanted and I think it dovetailed what users wanted. I am also a user I rationalized.
Further, many of the old timers like myself, were hung up on Tog and his Human Interface Guidelines. We would argue over lunch the Right way to handle UI for a specific feature, etc.
Jobs comes along and with Design at his side, and they decide that the address field in Safari should also be a progress bar (that appears to select the text as it loads?). Some poor engineer was told then to implement that. (I hated what I came to call one-off UI.)
This is why Snow Leopard was so buggy on release but unlike previous releases actually did get better: management stopped pushing new features, and even a skeleton crew was able to polish it over time. If you look at a map of tentpole Mac OS features, you can see that the ambition vanished with the iPhone's release and didn't recover for a decade.
I may be remembering incorrectly, but I honestly think it was better when it was released years ago than it is today.
I think you're remembering correctly. Siri used to work much better for basic tasks, and didn't seem to randomly fail for no discernible reason. I can't remember what iOS updates changed things, but there were at least a few that absolutely degraded Siri's functionality. One that stuck out to me was when Siri was one day suddenly unable to correctly respond to "play all songs shuffled" - it worked fine for years, but after an update, just decided it couldn't do that anymore.
The concrete value-measure for me is whether I would feel any loss as a user if Siri simply disappeared. The answer is no, I would feel no loss.
But this is also true for me if the subject is Gemini, CoPilot, Alexa, &c. For all the money and energy burned, for all the advertising babble and grotesque valuation, I would not as a user be one jot less productive if the invisible Thanos hand of the market made them all disappear at the end of this sentence.
Google Assistant with Gemini (or whatever its name is this month) is really remarkable at helping with mundane tasks, and I would really miss it if it were gone. Circle to search, asking questions about my agenda, and setting up new events on my calendar based simply on taking photos of documents or posters is so convenient that I really buy the concept of having an AI personal assistant.
And because this is a very sticky feature when delivered properly, if Apple is not involved it gives a very strong competitive advantage to Android flagships.
If Siri has been useless for more than a decade, that also means that many users are barely using it, if at all. I'm not the only user of a system containing such applications that has found zero concrete value in them.
I use Siri for about three things: timers, reminders, and asking about the weather. But the reminders are mission critical. I use the watch, and any time my flighty brain hits on something, I set a reminder for it on the spot.
I literally couldn't get along without it. The only killer feature of Siri is accurate voice transcription, and that's all I need (and it could be more accurate, I just get used to comical translations when I use a rare word).
I guess I wouldn't mind if it were better, but I don't need it to be. Reminders are enough.
All things considered Apple actually delivers some great software, Apple is one of the few companies that built their own operating systems, their software UX is considered to be the gold standard. They pulled architecture transition processes that required great software engineering alongside with the hardware engineering and they did it amazingly well.
Apple's success in some areas suggests that the failure in other areas doesn't have to be about software development culture but simply wrong bets or policies.
Apple tomorrow can become the best platform for anything AI, they just have to open up their platform and let people on the cutting edge of AI to take over some functions.
IMHO Apple's greed is failing them. That revenue growth in the "Services" category on their quarterly reports is their Achilles heel. Locking down users on their platforms allows them access to a "market without competition" but there's actually a competition happening outside of this walled garden and that competition is fierce. People in the garden start thinking that maybe the grass is greener beyond the fence.
Ironically, there's a possibility EU saving them from themselves if they stop being just compliant enough to hold up in court and embrace the spirit of what DMA is trying to do. Throw a tantrum on EU and say that the lost revenue was because of the evil EU thingy and save this great brands and great company.
But in those days apple was the underdog and HAD to do good by their users or there was no reason people would switch. Now it feels like their first thought is how they can monetize every step of the way - they're turning into Disney in that sense. There's often little money in fixing bugs anymore.
But when Apple advertises literally all their new devices as "Built for Apple Intelligence" without actually delivering on that promise, that's just false advertising.
His delivery is too perfect and makes him sound like a really good text to speech program. It’s off-putting. The only other person I’ve heard that has this presentation style is David Miscavige. Perhaps some analogies could be made…
We're on Sequoia 15.3.2 now and the keyboard shortcuts of the big new window tiling feature still don't work on non-Apple keyboards (or Apple keyboards, if you happen to have remapped the globe key to something else, like Caps Lock).
† Hyperbole, in case it wasn't obvious
I wonder how much U2 had to bribe someone at Apple for that
That's a page right out of the Windows Update playbook
Kind of like how Meta keeps the disaster that is Horizon Worlds going, but much worse because Apple has a monopoly on AI integrations in iOS, and nobody else is allowed to compete with their terrible implementation.
Me: Hey Sir, set a timer for my management meeting in 10 minutes.
Siri: I've deleted your minutes from the last 10 management meetings.
Fundamentally, Siri's great promise (well in the last few years) has been that it could hook into all of the scripting facility made available by iOS apps. There are tons and tons of AppleScript hooks on many, many apps -- if only Siri could, you know, access them.
This is traditionally not Apple's strong suit - API access oriented thinking - but I will also say that it turns out to have been a pretty hard problem to actually parse out what 'hooks' to access.
We now refer to this as "tool calling", and know that it's a specific thing an LLM (or even MLM/SLM in many cases) can be trained to do a pretty good job at.
If Apple wants to do all this on device, I think we're not far off in terms of hardware and architectural knowhow to get this done now. I would bet that the Siri team spent a number of years on alternates, essentially relearning the bitter lesson.
I'd guess the best way to make this work on device would be to train and finetune a small model on a massive number of apple app store apps so that it gets good at figuring out which apps it will want to query for tooling functions calls - that's too much to inject in context directly especially on device - then it would call out for a list of available calls -> that's just code -> then it could use a local LLM to choose the proper call and call it -> then if needed it could offload for more heavyweight thinking. To start you could offload the tool call decisions as well, or really the whole thing, except for the local function call.
This would be vastly vastly better than what we have now, and I think does not require huge reintegration efforts from app developers.
"Siri what month is it?"
[... calendar_getcurrentdate requested ] -> [ Mar 21 2025 ]
"It's March".
But Siri is such a terrible product that it amazes me that the org hasn't been deleted, and that Siri hasn't been ripped out of products.
I have seen Apple delete bad orgs a couple times before so I find it strange they let Siri survive. It does not "have good bones," it is not a good product and it never will be.
It is actively frustrating to use and I deeply regret migrating to HomeKit as well, primarily because of the usage experience via Siri.
Truly a total/spectacular failure of an org and product. It ought to be a case study for how not to product manage.
Starting a specific album at a specific song, for example... well Siri handles it, to a point, but it has to interface with Spotify, and there you hit a wall of hot garbage, along with the sequencing and process-management/integrity issues of integration software. Garbage meets a money fire, in other words. High cost, brittle, constantly changing, glue code that is mostly depending on third parties also requires session management? Not a winning product formula.
Sprinkle in some inhuman HCI nightmares like not pausing speech appropriately when reading an album title and you've got a recipe for an expensive turd that annoys your customers, oh my, by a lot.
From the perspective of 1960, early "robots" were expected to be physically capable and mentally feeble. Good at logic. Capable of making a sandwich. Weak at empathy and whatnot.
Even from the perspective of 2025... most people don't understand how slowly robotics has advanced. Human-level performance at laundry folding remains a distant dream. Empathy is increasingly trivial.
So Siri.. and voice UIs generally. The bottlenecks have been in unexpected places.
In general, we just don't have very good UI paradigms for voice. Voice recognition is finally good. LLMs theoretically add a lot of capability. But... there just isn't a great UI.
It's like trying to use a smartphone with a nipple mouse instead of touch. You can slowly hack your way to making specific tasks/features work... but there is no radiation event where lots of tasks become possible.
Now it’s a crapshoot.
It wasn't a good verbal command line. It was usable because even an imperfect command line UI is inherently usable if use cases are limited. Building on to that though... you always overreach.
Personally, I always thought voice+screen interface was an under-explored. That (for example) would let you go from command line to command line navigator... a natural progression.
Voice UI kind of needs to be "one shot" to be pleasant: Play "Fortunate Son." Once you have to go back and forth, listen to list of options... it's no fun anymore. The Windows/Mac/Xerox GUI paradigm required a mouse, keyboard & screen. Voice only is limiting... perhaps.
Plus, there’s no more humor. Trying to crack jokes at it used to give funny responses. Even old things like “I need to hide a body” — “I don’t know how to respond to that”. Insult it and it’s a “I won’t respond to that”. Lots of old funnies are now gone and stripped away.
which is equally funny I think
https://en.m.wikipedia.org/wiki/Moravec%27s_paradox
You have no idea how much "hole" you just filled in my lexicon. Have been articulating this all wrong for years.
The futuristic ads were made for the consumer to feel safe and sound.
The actual money flowing went into military applications (drones, guidance systems, etc) and commercial applications (car factories)
For the above systems, they don't have to be physically capable, just have a restricted set of actions / outcomes, for example, flying a missile. But, they need to be mentally strong, to take decisions at split seconds.
It has only become worse in the past couple of years.
The inability to know that bedroom means bedroom [1], or the impossible task of adding several items to a list [2], or failing to answer what the current month is [3] have nothing to do with slow progress or UI.
[1] https://x.com/dmitriid/status/1781030511336456612
[2] https://mastodon.nu/@dmitriid/114172857097176113 and https://mastodon.nu/@dmitriid/114180042031267746
[3] https://news.ycombinator.com/item?id=43432945
At a basic level i just want the interface to have good voice to text recognition. if it's a question ask an LLM, if it's a special recognized command run it instead.
At a dev/pro-user level let me define statement patterns (regex or semantics) where i can trigger a shortcut and parse some input from a single statement.
By all means, regen the custom responses devs and users setup with the Siri personality LLM so it sounds like Siri all the time. Just set the interface free already.
Voice assistance was the hyped thing sometime before block chain and LLMs. Everyone needed one or was working on one. So Siri got stuck into the iPhone, like Apple Intelligence is now, because "the market" needed Apple to do so to keep stock prices going up. Then the hype died out Apple sort of needed to keep Siri around, because it is one of the interface to the HomePod and CarPlay. There's no reason do develop Siri beyond that, because very few people use it, not having a better version isn't going to sell less iPhone and they can't mine the data, because they sold everyone on Apple being privacy friendly.
There's absolutely no business case in having Siri do anything beyond "Call my wife", "Next song", "Play U2".
The technology exists for much better responses to those types of requests today, but who knows if they can re-train their customers that it'll work for them now.
There are definitely a lot of complicated personal-assistant type tasks that they could do now that people would like.
I feel like they may be experiencing that issue where the product is quite good within some core set of features, but the edge cases are where all the paper cuts happen. And when its a singular product on a singular device made by a singular company, those paper cuts really add up. There must be a more succinct term for that.
Can it even be solved with the current state of LLMs?
I don't miss the old Assistant at all now.
The main issue to Apple is the different mindset that takes to deliver hardware vs. services. Delivering hardware you have one shot to "wow" your customers, and Apple keep nailing this process in a way that the whole industry can't keep up with anymore. On software, that doesn't work that well. You can totally be disrupted by a startup that delivers an MVP faster with 30% of your perfect product that would take 3 years to deliver. In this way, Google's assistant strategy in the short term was spot-on compared to Apple's. You take the hits of bad feedback in the short term in the hope that being in the game can create enough momentum to get you to a solid position of features and know-how in a couple of years.
- Their promise isn't as ambitious as Apples. It's basically Gemini + pre-existing Assistant features as plugins. Not trying to be the privacy-touching agent at least at the first cut. (Or they promised big, but they have trained the audience not take it as a fade value haha.)
- They ship the Gemini version as an opt-in for the Assistant replacement so that they can learn "iterate" on the adventurous enthusiast while leaving larger user base as is.