78 comments

[ 3.0 ms ] story [ 161 ms ] thread
Wow. iOS 14 is hella bloated.

I would like the option to run an iPhone OS 1.0 sized OS on a modern iPhone.

But why? You'd have no 3rd party apps, gps, cell service, and Safari wouldn't work. To make it usable, Apple would have to put in a lot of effort.
Safari would work with my own site, and probably with Google Search.
But not with HTTPS.
I think HTTPS is overrated for most applications.

It has many drawbacks, too.

Define bloated
From experience it’s typically used to mean ”Something I dislike”
haha you got exactly what I was driving at. Wehn you're driving anyone passing you is a reckless maniac, and anyone slower than you is an idiot. When you're programming any convenience a language/standard library has more than your favorite is just Fisher Price childish handholding, and any missing feature is senseless busywork that should be built into the library.

And when it comes to OSes the word of choice is Bloated it would seem.

This is presumably part of the reason the iPhone 6S is not going to receive iOS 14 next year - since it was sold with 16GB as the lowest capacity. The next model along, the iPhone 7, had 32GB as the lowest capacity.
With all the horsepower directed at machine learning, why does it seem like text autocorrect hasn’t improved in the slightest?
Its general purpose hardware, which without a good model doesn't do much.

The keyboard is "good enough" and Apple is going to capture the market or retain users with a better ecosystem of apps. So they're investing in making it easy for other developers to build things using ML on the iPhone.

Autocorrect is fine for me in iOS 14. But Machine learning can also lead to regressions or edge cases. Here a fun rant about autocorrect in iOS 13:

https://www.smh.com.au/technology/iphone-autocorrect-driving...

The good:

> Machine learning has enabled the iPhone to correctly predict oddly spelt suburb names like Prahran and Woolloomooloo, to read the screen of an article and guess the surname or product I'm about to Google, and to suggest the spelling Bryan over Brian, because the former is a close friend.

The ugly:

> For at least two years now, my iPhone has insisted on autocorrecting Sydney to SYDNEY, every single time I type the word. After weeks of troubleshooting that included resetting the keyboard and uninstalling any transit apps I thought might be to blame, I realized the culprit. I have a few hundred contacts on my phone from a previous employer, all with UNSW SYDNEY as their address.

What should happen here with a few _hundred_ entries of "SYDNEY"? I guess it should be unlearned after the user entered a few times (once? two times?) by hand "Sydney" and user entered text privileged over other text corpora (even dictionaries?). But then you have other problems again like autocorrect learning potentially wrong spelling idiosyncrasies of the user.

> What should happen here with a few _hundred_ entries of "SYDNEY"?

I'd say, change the case in the database to proper "Sydney" and have whatever writes a piece of physical mail do the transformation to the correct format?

Because gboard predicts email addresses, and I have several Haskell-related mailing lists in my whitelist, gboard always tries to autocorrect "Haskell" to "haskell-mailing-list@..." when I am trying to text my pal about Haskell questions, so I get messages like (because I sometimes just swipe and send, without checking the message, like a moron):

"Hey, do you know in haskell-mailing-list@something.org if the Prelude..."

So, now I've turned off EVERY bit of AI bullshit everywhere, and I deal with mistakes myself.

I think the number one fix for autocorrect is adding an "I'm an adult, no I never mean duck" toggle
It could be personal preference, but I've found that since iOS7 or so it's actually become worse. Since then, I've found autocorrect and other text input related actions like selection and moving the I-cursor around became much more frustrating to do. There have been various tweaks since then, but it's never felt as seamless as it did in earlier versions of iOS.
I agree. I’m amazed how well my favourite IDE - Webstorm does spell correction. If you add keyboard shortcuts it’s just delightful.

Question about iOS I have is if it’s packed with vision models, why can’t I as a developer just feed an image and get tags / text?

As a developer, you can. The APIs exist (https://developer.apple.com/documentation/vision/vnrecognize...), seem simple to use (https://www.hackingwithswift.com/example-code/vision/how-to-...) and appear to perform decently (https://heartbeat.fritz.ai/comparing-apples-and-google-s-on-... compares with Google’s on-device OCR)

So, finally, they don’t step on the toes of third-party developers, and you still complain? ;-)

Thanks, was looking for it few weeks back and found you'd need to load up your own models.
I keep asking myself the same questions every time i read about modern (apple) hardware specs: what are they going to do with all this ML-dedicated hardware ??

I don't see anything smart with my iphone, nothing has changed since the introduction of Siri. I still can't use it seriously for anything else other than to set the alarm clock, and not a single app i download or use seem to do anything special needing ML running on the client side.

For folks heavily using their iPhone as a camera and photo album, there are lots of ML-aided things going on. You can search through photos by text without having to have previously annotated your photos, it automatically builds albums that contain people it's decided are important to you, it automatically pulls out the "best" photos for you which really helps if you like to take tons of photos, it can automatically create entire slideshows for you. This is probably just a subset of the things to which ML is applied in one app on the device.

Fully agreed on the Siri shortcomings, though! But, I'm not sure how much of that work is done on device with the neural engine vs in the cloud. Not like that matters when the end result is so bad.

One really silly bug (can't call it anything else) is the insistence to capitalise Apple product names like in their marketing speak.

When I'm writing in my native language, I might type a phrase like "dit is echt the next best thing" or "ik breng je naar Brussels Airport". My iPhone will 'correct' this to NeXT and AirPort.

Really interesting thanks for sharing. Its not surprising to see the size of iOS grow so dramatically over the last 10 years. Some may so bloat, but i'd say it reflects the iPhone/iPad's evolution into a device that is integrated deeply into our lives.

Just look at how much space fonts are taking up in iOS14. My hunch is that reflects a growing set of emoji and support for hundreds (thousands?) of alphabets. It's hard to call that bloat.

Also initially I was surprised to see the networking code take up so much volume, but that also makes sense. I can only imaging how much custom code is needed to ensure a solid connection with the hundreds of wireless carriers around the world.

> My hunch is that reflects a growing set of emoji and support for hundreds (thousands?) of alphabets. It's hard to call that bloat.

I find it incredibly easy to call emojis bloat.

I actually use some emojis in my notes to make them a bit more visual, and doesn't require to import an image so it's dead-simple to backup.
Emojis are incredibly well-loved by many people and used all the time. So bloat for you, perhaps, but a critical feature for many!

No kidding, it's the reason many people update their phones.

> it's the reason many people update their phones.

Not being able to install new fonts and being forced to ‘upgrade’ (in a non-modular fashion, being forced into a whole bunch of new ‘features‘ that push older devices to early deaths - planned obsolescence - which leads to tons of toxic e-waste [1]) is peak corporate black-box fuckery.

And before someone says “don't use it then, no one is forcing you to use it”, just remember that corps like Apple kick away the ladder they climbed up themselves: blocking others to create alternatives by privatizing and monopolizing new discoveries (often originally funded by government grants [2]) through the use of Patents and Copyrights and other ‘intellectual property’ protections given by the Intellectual Property office.

Public money, public code!

[1] https://youtube.com/watch?v=mleQVO1Vd1I

[2] https://time.com/4089171/mariana-mazzucato/

I think Apple is behaving reasonably here, at least, I don't think they're deliberately breaking older devices w/ updates, so I wouldn't apply the phrase "planned obsolescence," exactly. But I'll conceded that normal programmer laziness (i.e., optimize just enough to make it run acceptably on latest hardware) ends up having the same effect.

Software patents & much of copyright seem to me to be to be hurting more than they are helping.

I'd love to experiment w/ ditching those.

While we're at it, we should legally mandate that devices are open to 3rd party app stores and be "jailbroken" by anyone that cares to do so.

(It's pretty crazy that the dominant artist medium of our time is controlled by a single, outrageously censorious entity. Thank god we've got the web, at least. (For now.))

For me I think emojis help more clearly show the emotion of text being written.

Sometimes it’s not obvious in which way text can be interpreted so it helps give context that’s missing in non-face to face chat

We were constantly told in school "sarcasm does not communicate well over text!"

Between , , and WrItInG lIkE tHiS, the situation has greatly improved since then.

EDIT: Aparently HN is not displaying the eyeroll, facepalm and upside down smiley emoji.

> EDIT: Aparently HN is not displaying the eyeroll, facepalm and upside down smiley emoji.

IIRC HN strips out most things not in the BMP.

Why is it that you think emojis 'clearly show emotion'? Do they have well-defined meaning? Where does one look up the emotional context that a specific emoji adds?
Do facial expressions have well-defined meaning? Where does one look up the emotional context that a specific facial expression conveys when you say a thing to a person in real life?

How can you be sure that emojis aren't conveying emotional context to you? It's not possible to rewind reality to the time before you received that text, and replay it again without the emoji to see the difference. I posit that emojis do in fact add emotional context, and if you could run that experiment, that you would have a harder time deciding sarcasm, self-deprecation, humor in general, and a whole host of other emotions. They are becoming part of the textual grammar just like facial expressions provide grammatical context for sign language and can totally change the meaning of the same hand sign.

Either you learn the emotional contexts, or you don't and suffer the social consequences. I don't see much that's changed, just that technology is still way behind in-person communication. Seems to be catching up little by little, though.

>Do they have well-defined meaning?

Emotions don't have either. That's why emojis show them well. Plus people have learned for millenia to read other people, and face emojis are basically glorified magnified face expressions...

Directly from the article on iOS 14:

“Fonts are now counting for less than 6 % of the size”

And I assume emojis are part of that relatively small percentage. (‘Apple Color Emoji.ttc’ takes up 208 megabytes on my Mac, assuming that’s how they implement emojis.)

You’d fault wallpapers for bloat before emojis. Those things are massive.

That bloat is responsible for getting manufacturers to support Unicode, and getting consumers to update their OS (and often get critical security updates in the process).
A long time ago I was part of a team that did a competitive analysis of the first iOS when it came out. One of the key takeaways from that report that the team decided on was how much Apple had wasted space with unnecessarily high quality UI elements.

Even at the time, I remember thinking this is what makes it look so special, we shouldn't write it off just because we're used to scraping for every byte at the cost of shiny graphics.

Sadly the optimization masters won the day, and didn't see the writing on the wall that Apple did.

Where was this, a competitor like Nokia?
Can someone explain how Apple is full-tilt on ML and yet still has the worst personal assistant (Siri)?
I would say one of the main reasons is data - Google just has a lot more of it.
My experience with Alexa has been more positive than with Siri, and I'm a full-bore Apple user.
Siri isn't the best but it's far from the worst - Bixby owns that category by a mile.
I don't even know what Bixby is, so you're probably right.
I'm pretty sure they killed it (Bixby) a few months ago
(comment deleted)
There’s a lot more to ML than just Siri. Photos, for example, utilizes it pretty heavily. Here’s an interview with their AI chief to give you an idea of the scope: https://arstechnica.com/gadgets/2020/08/apple-explains-how-i...
No that's my point. They seem fairly competent at doing ML in basically every other application they're using it for.

But Siri still sucks. If they were generally bad at ML I'd understand. But they're not. They make their own TPU-type things FFS.

All I have are second-hand accounts as to why. They have a sizable Siri team in Boston and a good friend of mine interviewed there. This friend asked a number of questions regarding tech stack and their SCM only to discover what a complete mess it was. It was enough so that he was turned off by it. Again, second-hand account. It leads me to suspect that the struggle has always been to technical issues rather than product.

It also might be that Apple hasn’t seen voice control as an area of growth or enhancement to experiences long term. Maybe those who have Alexas and all of its processing capabilities end up just using it for timers and dictionary definition lookups.

The only hits I get for SCM are Supply Chain Management, which doesn't make a ton of sense to me in this context. What do you mean by that acronym?
My guess would be not collecting data on users.

Google's entire model is collecting massive user data which they can then use to train their models.

It's a lot harder to do that kind of thing when privacy is a big part of your brand (thankfully). I prefer the tradeoff of Siri being behind the others if that's what it takes.

Sorry what data they don’t collect? Audio is being sent to servers, videos on icloud, etc.

It’s either they suck at it or pretend to suck so people think their data isn’t “collected”...

I don't think Siri is the worst if you're familiar with ways that you can work with it. Google is probably the best at discerning intent from abstract speech but Siri has never not worked for me when I've told it to do something, even going as far as HomeKit and trivia which, I would imagine, would have a lot of ways to misinterpret.
From my experience, Siri is only good at the things we've never needed ML for (beyond MLs role in speech recognition).

"Set me a timer" or "call my brother" both work great, but you can make something like that with off-the-shelf software and libraries in like an hour of hacking.

They've integrated Siri in speakers, computers, phones, and more, and yet somehow it's worse than Alexa, where most of the exposure to that is just through echo devices.

Can you give an example of something that you can do with Alexa that you've had trouble with on Siri? I'd like to test that. I don't ever have issues with Siri and I'm struggling to think of something that Alexa would be that much better with.
I have personally found that Alexa is much better at general info questions (what is the capital of Djibouti?) and is much more responsive as a Smart Home orchestrator.

Here's a face-off article[1] that might help.

[1] https://www.tomsguide.com/us/alexa-vs-siri-vs-google,review-...

Interesting. I just tried that question and got the correct response and I use Siri for all my home automation stuff. I even have triggers that go off when my Apple Watch connects to certain things and I use my "Hey Siri, let's watch a movie" and "Hey, Siri, let's watch TV" scenes nearly every day.

Edit: For reference, her answer was "Djibouti is the capital of Djibouti"

It’s particularly bad on macos as it seems never know when to stop listening. “Call dad” never works on my macos for some reason, just crashes quietly.

iOS or latest iPhone it slowed down massively most of the time hitting a race condition...

1/4th of iPhone OS 1.0: fonts
I wanna know what that one huge file that takes up most of "Health" is.
I would imagine it's the HealthKit framework that enables devices and apps to share health data.
Health is big, it tracks heart rate data for years, has step analysis to calculate how much movement is being done(gps data being loaded as well as accelerometer data) Also sleep analysis. So I assume lots of custom libs apple has along with storage of these health stats and client(iphone) side analysis(health privacy).
Samsung somehow fits all of that on their Galaxy Watch though - what's the space taker here?
Your comment implies that the Galaxy Watch has a small memory size because it's... small?
No, it doesn't do that at all - I'm just wondering why the difference since Galaxy Watch has only 4GB of total flash storage on board (which isn't filled up by the OS) and provides this functionality.

Health is a significantly larger piece of software, so there must be more to the story.

Machine learning model for analyzing movement data?
Interesting correlation between hardware and software, the iOS14 (2020) tree map looks complicated, and requires a great processor 5nm M1 SoC with 16 billion transistors (2020), with harware tree map is even more complex with. Probably, the future software iOS30 (2030) tree map will look as complex as the M1 (2020) chip itself and will require an <5nm Mxx chip, the tree map of which will be even more complex

PS: iOS apps can run on Mac with new M1 chip or iOS device with ARM chip like Apple A14 Bionic

Would be interesting to plot this trend normalized to transistor efficiency.

e.g. if we stall at the 3 nm node or whatever, would we expect the feature set to level off?

Do we really expect the phone to continue to improve at this pace the next 100 years?

There are other ways besides brute force. For example, AMD got 20% IPC improvement on the same node, and apple's M1 SOC has built in X86 address translation to improve cross compilation, it also has the 'neural engine' and whatnot. As for 100 years, nobody knows. Every step of the way so far, we accomplished stuff that we deemed impossible.
> the whole iPhone OS 1.0 is basically taking the size of the iOS 14 wallpapers

I forgot iPhones came with wallpapers

When in 2008 I think Apple was announcing launch of "icon language" aka emojis. I was thinking at the time on reading that arstechnica article "wow.... I would be going to have to learn an entirely new language based on graphics.... Oooo.... like chinese? How would I remember it...."

All those years later, I still don't use emojis in my daily life but I would love to read that article again