AAPL - Apple Inc.

[Dynamic]

If, for example, this year produces great sales of iPhone X and the other models still being made it could be a very tough year-on-year comparison and make the market fearful of a more permanent sales decline next year.

 

I think to some extent this is what happened in early to mid 2016 giving a great buying opportunity about $95. And at the same time as iPhone sales were below the previous year's levels, Samsung and Google where making quite good gains on high end phone sales albeit at much smaller profit margins. I

 

It wouldn't have to get that cheap to be a screaming buy in my estimation, especially if stock buybacks continue when cheap.

 

Now is a time people seem fairly optimistic about Apple so it doesn't produce really cheap prices. I wouldn't say it's all that expensive though for the long term and I can see a case for buying (which is why I'm holding). Yes, Warren Buffett (WEB) has been buying, but actually I think he was buying in 2016 but not in such volume as to prop up the price.

 

My position size is quite large so I might consider lightening up on Apple eventually, possibly even when it's still a little undervalued, but most likely when something else looks really cheap and I switch at least part of my exposure to that.

[Liberty]

https://www.imore.com/iphone-x-review

 

https://techcrunch.com/2017/10/31/review-the-iphone-x-goes-to-disneyland/

[Liberty]

Testing FaceID with hats, glasses, in the dark, etc:

 

[KCLarkin]

Testing FaceID with hats, glasses, in the dark, etc:

 

 

Very interesting. These are the things that Apple bears (Bluegrass) don't get when they say that Apple isn't innovative anymore.

 

Apple does an enormous amount of work with FaceID, Apple Pay, etc. making sure they work in the real world.

 

Companies that invent at a faster pace (e.g. Google or Facebook) put much less effort into their innovations. So you often see Samsung or Android come out with an invention. But it seems like a gimmick rather than a real innovation. Apple usually makes the same technology work to such a high standard that it promotes widespread adoption.

[LongTermView]

Testing FaceID with hats, glasses, in the dark, etc:

 

 

This was interesting a little after the 6 minute mark:

The thing with face id is 1 in a million chance of someone you don't want to get into [the phone] compared to 1 in 50 thousand with touch id.

[Liberty]

Testing FaceID with hats, glasses, in the dark, etc:

 

 

This was interesting a little after the 6 minute mark:

The thing with face id is 1 in a million chance of someone you don't want to get into [the phone] compared to 1 in 50 thousand with touch id.

 

This was part of the keynote reveal of FaceID. It's safer than TouchID, 1 in a million rather than 1 in 50,000.

 

I've been reading lots of reviews of the X, and watching lots of videos, listening to podcasts. So far people seem to like FaceID at least as much as TouchID or better. A conclusion that I've heard a few times is that after a while, it feels like you are using your phone without authenticating anymore (like in the early days, back before people put PINs on their phones), even if in fact you are secure. It reduces friction even more than what existed before -- you just look at your alarm in the morning and it lower volume because it knows it has your attention, you look at your lock-screen notifications and the content of the messages pop up once it IDs you, every thing that worked with TouchID automatically works with FaceID because the APIs were built for biometric ID without being dependent on one method, etc.

[gfp]

It is great news.  TouchID never worked for me.  My iPhone ten doesn't arrive until November 14th but I am really looking forward to it.  On a side note, Berkshire's apple shares are looking like quite a chunk of change lately.

[Liberty]

This looks so much better than TouchID:

 

[gfp]

Just anecdotally, my iPhone X order that was supposed to arrive between the 14th and 26th is now shipped from China and arriving by November 8th - so they appear to be accelerating a lot of the projected deliveries and this may bode well for (whatever they call our Q4) deliveries, guidance, etc...  If they can get manufacturing capacity high enough they can have a pretty epic Christmas selling season.  Either way (whatever they call our Q1) will be robust.  Average selling price is about to get a pretty meaningful boost.

 

It is great news.  TouchID never worked for me.  My iPhone ten doesn't arrive until November 14th but I am really looking forward to it.  On a side note, Berkshire's apple shares are looking like quite a chunk of change lately.

[hardincap]

once again a great quarter. its amazing how misunderstood the largest co in the world has been over the past few years

[rkbabang]

Just anecdotally, my iPhone X order that was supposed to arrive between the 14th and 26th is now shipped from China and arriving by November 8th - so they appear to be accelerating a lot of the projected deliveries and this may bode well for (whatever they call our Q4) deliveries, guidance, etc...  If they can get manufacturing capacity high enough they can have a pretty epic Christmas selling season.  Either way (whatever they call our Q1) will be robust.  Average selling price is about to get a pretty meaningful boost.

 

I was able to pre-order my third one yesterday from Xfinity.  Yesterday I got the tracking numbers for the two that I ordered last Friday and they are shipping from Ft Worth, TX by FedEx and are scheduled to arrive today.  The one I ordered yesterday says expected shipping Nov 24th.

 

[johnny]

This was part of the keynote reveal of FaceID. It's safer than TouchID, 1 in a million rather than 1 in 50,000.

 

This is a garbage marketing stat. Fingerprints are essentially random, so a 1 in 50,000 false positive rate for fingerprints may actually be far more secure than a "1 in a million" rate, if the underlying factors for the process are non-random in the population. The average person you don't want getting into your phone is far more genetically similar to you than the average Earth Human. The real test will be what the false positive rate is among Chinese people, Black Americans, White People From Oklahoma, etc. Since part of the "security" of the feature depends on them being cagey about exactly how it works (how are the factors weighted, if you are in the 99.9th percentile in nose-broadness, how much will the pass/fail determination lean on that?) being too confident about the feature before it's been in the field for a few months is probably not warranted.

 

That said, what's more important is actually what the false negative rate and total process time are: these are the factors that will make real users form an opinion about the feature (and possibly de-activate it). So far it sounds like it feels around in between the first and second versions of TouchID, which is probably good enough.

 

It's a really cool tech to be pursuing though, and they're in a really good position to push it forward fast.

[LC]

The reason I still use my digit key code on my iPhone is that if you are arrested or whatnot, or if you record the police doing something illegal, the cops can open your phone with your fingerprint/face ID. But you are not legally obliged to give up your passcode.

 

I doubt it'll ever happen on my life but I just don't like the principle of somebody opening my phone up without my consent or a warrant

[tengen]

I’m not too keen on facial recognition after reading this article about fooling image recognition software: https://www.popsci.com/byzantine-science-deceiving-artificial-intelligence

[pau_]

FaceID is different than 2d image classification/recognition though. It's projecting a pattern of  IR dots

(consistent but irregular and varying from phone-to-phone) on the subject, and using that to create a 3d representation. This is why unlike previous attempts by other companies, FaceID can't be fooled by a picture of your face.*

 

You also have to think about your threat model. With a depth map based on IR, I guess maybe someone could record the pattern on your face while you're authenticating, maybe you could create a custom mask of IR LEDs that would fool FaceID. But at that point you're dealing with state-level attackers who could probably just compel you to produce your passcode (aka rubber hose cryptanalysis).

 

All that said, adversarial attacks on neural networks is a very interesting topic. Google just published a paper on making adversarial 3d objects that can be photographed from any angle and trick a deep neural net (though note that this is still a 2d projection/image recognition problem). In their example they make a 3d printed turtle that always registers as a rifle.

https://www.theverge.com/2017/11/2/16597276/google-ai-image-attacks-adversarial-turtle-rifle-3d-printed

 

[edit: * basically these are better machine learning "features" than images, which are just pixels of color/saturation data. the more I think about it, perhaps there are also adversarial attacks possible here, but I would guess they would be harder]

 

[edit 2: here's Apple's whitepaper on FaceID security if you want the nitty gritty]

[Liberty]

FaceID is different than 2d image classification/recognition though. It's projecting a pattern of  IR dots

(consistent but irregular and varying from phone-to-phone) on the subject, and using that to create a 3d representation. This is why unlike previous attempts by other companies, FaceID can't be fooled by a picture of your face.*

 

You also have to think about your threat model. With a depth map based on IR, I guess maybe someone could record the pattern on your face while you're authenticating, maybe you could create a custom mask of IR LEDs that would fool FaceID. But at that point you're dealing with state-level attackers who could probably just compel you to produce your passcode (aka rubber hose cryptanalysis).

 

All that said, adversarial attacks on neural networks is a very interesting topic. Google just published a paper on making adversarial 3d objects that can be photographed from any angle and trick a deep neural net (though note that this is still a 2d projection/image recognition problem). In their example they make a 3d printed turtle that always registers as a rifle.

https://www.theverge.com/2017/11/2/16597276/google-ai-image-attacks-adversarial-turtle-rifle-3d-printed

 

[edit: * basically these are better machine learning "features" than images, which are just pixels of color/saturation data. the more I think about it, perhaps there are also adversarial attacks possible here, but I would guess they would be harder]

 

[edit 2: here's Apple's whitepaper on FaceID security if you want the nitty gritty]

 

And at the keynote Apple explained how they worked with professional hollywood mask makers to try to make 3D masks to fool FaceID and then used machine-learning to train the system not to be fooled by even extremely realistic masks. Seems to me like FaceID is as secure as most people can ask for, especially since no info about your face leaves the secure enclave on the phone (doesn't go to cloud, can't be accessed by apps.. They just get a biometric "OK" or "not OK" signal back when asked to ID).

[rkbabang]

FaceID is different than 2d image classification/recognition though. It's projecting a pattern of  IR dots

(consistent but irregular and varying from phone-to-phone) on the subject, and using that to create a 3d representation. This is why unlike previous attempts by other companies, FaceID can't be fooled by a picture of your face.*

 

You also have to think about your threat model. With a depth map based on IR, I guess maybe someone could record the pattern on your face while you're authenticating, maybe you could create a custom mask of IR LEDs that would fool FaceID. But at that point you're dealing with state-level attackers who could probably just compel you to produce your passcode (aka rubber hose cryptanalysis).

 

All that said, adversarial attacks on neural networks is a very interesting topic. Google just published a paper on making adversarial 3d objects that can be photographed from any angle and trick a deep neural net (though note that this is still a 2d projection/image recognition problem). In their example they make a 3d printed turtle that always registers as a rifle.

https://www.theverge.com/2017/11/2/16597276/google-ai-image-attacks-adversarial-turtle-rifle-3d-printed

 

[edit: * basically these are better machine learning "features" than images, which are just pixels of color/saturation data. the more I think about it, perhaps there are also adversarial attacks possible here, but I would guess they would be harder]

 

[edit 2: here's Apple's whitepaper on FaceID security if you want the nitty gritty]

 

And at the keynote Apple explained how they worked with professional hollywood mask makers to try to make 3D masks to fool FaceID and then used machine-learning to train the system not to be fooled by even extremely realistic masks. Seems to me like FaceID is as secure as most people can ask for, especially since no info about your face leaves the secure enclave on the phone (doesn't go to cloud, can't be accessed by apps.. They just get a biometric "OK" or "not OK" signal back when asked to ID).

 

I've been using my phone since Friday night and I love FaceID.  I obviously can't attest to how safe it is, but I can say it works to unlock the phone with my face every time, with or without my glasses, and even in a dark room.  I tried Apple Pay at the grocery store yesterday and it was quicker and easier than using touch ID.  Also I thought I would miss the home button, but I don't.  The new swipe up and swipe up and hold gestures are just as easy, take just a few minutes to get used to, and all the extra screen real estate makes it worth it.  It feels roughly the same size as my 6S did in my hand, yet the screen is so much larger.  After less than 3 days of use I'm definitely glad I upgraded.

 

 

[Liberty]

The new swipe up and swipe up and hold gestures are just as easy, take just a few minutes to get used to, and all the extra screen real estate makes it worth it.

 

Another killer gesture is this one:

 

 

You swipe right or left on the "home bar". You don't have to do the exaggerated circular motion that is in this video, much faster to just do it in a straight line. Smooth as butter.

[rkbabang]

The new swipe up and swipe up and hold gestures are just as easy, take just a few minutes to get used to, and all the extra screen real estate makes it worth it.

 

Another killer gesture is this one:

 

 

You swipe right or left on the "home bar". You don't have to do the exaggerated circular motion that is in this video, much faster to just do it in a straight line. Smooth as butter.

 

Nice! Thanks.  I didn't know about that one.

 

Another one that is nice is go to Settings->General->Accessibility and enable Reachability.  Swiping down on the home bar brings the top of the screen down so that you can reach it with your thumb. Then swipe up to put it back.

 

[Liberty]

Nice! Thanks.  I didn't know about that one.

 

Another one that is nice is go to Settings->General->Accessibility and enable Reachability.  Swiping down on the home bar brings the top of the screen down so that you can reach it with your thumb. Then swipe up to put it back.

 

Yep, I did try that one. Another one if you want to kill an app quickly for whatever reason, you go in the switcher (slide up and hold), and then hold on the app until the red circles appear, but then you don't even have to tap the circle, you can just swipe the cards up.

[Liberty]

Ben Thompson on Apple and the iPhone X:

 

https://stratechery.com/2017/apple-at-its-best/

 

Also, just received my protective case for my X. The UPS guy remembered delivering my iPhone X last week and asked about it, said he went to the store to play with it, so I showed him mine. He asked about the home button a bit and when I showed how it worked, he was like "oh, that makes sense". He said the next time I see him he'll probably have bought one too. FWIW.

[gfp]

I've had the X for a couple days now and I like it.  My wife will buy one since it didn't seem too large for her - that's what she was worried about, ladies pockets are comically small.  It seems to me that there is room for Apple to introduce a "+" model of the X, call it Xs+ or whatever on the next half cycle update.  If Apple adds a larger version of the X, which to my hands there seems to be room to do, it could add a follow-on effect just like their first plus-sized phones did, and further maintain demand without reinventing the wheel.  Just seems like they have a lot of runway in front of them.  At this point the only thing I can see messing them up would be exploding / bulging batteries on a big scale.  So far so good on that front, but its early.

[rkbabang]

There is a serious calculator bug in IOS 11.1  Open your calculator and do: 1 + 2 + 3 =

 

 

[Liberty]

There is a serious calculator bug in IOS 11.1  Open your calculator and do: 1 + 2 + 3 =

 

Saw that. I read somewhere it's fixed in 11.2 beta.

 

It's more about the speed of input than about the numbers. If you do it slowly it works, but if you do it fast the + doesn't register in time and you get the wrong number.

[rkbabang]

There is a serious calculator bug in IOS 11.1  Open your calculator and do: 1 + 2 + 3 =

 

Saw that. I read somewhere it's fixed in 11.2 beta.

 

It's more about the speed of input than about the numbers. If you do it slowly it works, but if you do it fast the + doesn't register in time and you get the wrong number.

 

You are correct.  It misses the + sign if you type it too quickly after hitting the number, so it is getting 1 + 23 = 24.  You have to make sure that you pause a little after every number input if you want to use the calculator.