Apple Drops Hints About Working on Self-Driving Cars | Hacker News Reader

Apple Drops Hints About Working on Self-Driving Cars

wsj.com

24 pointsuptown46 comments

Let's just pretend for a moment that there's no ambiguity and they're actually doing this. Would you drive...or rather...ride in a car powered by Apple software?

I would not.

If there's one thing the current iteration of Apple has shown, it's that it's not great at writing software. And with this in mind, I'd be terrified to ride in a vehicle powered by carOS.

This isn't coming from a place of anti-Apple sentiment...I use quite a few Apple products...I just accept that they're pretty bad at this stuff these days.

Edit: Trying to understand the downvotes here. Care to follow up with a comment instead? I'm genuinely interested in discussing this.

This is a pretty silly argument.

Obviously you apply different standards and methods to develop life-or-death regulating software than for customer grade game and web surfing platforms.

I agree that my Apple products aren't as solid as they used to be, but this would (I assume) be a very different team run in very different ways.

Sometimes I think we "anthropomorphize" companies and brand names. They're not monolithic things with one personality.

That's fair, and I agree that we tend to think of companies/brands in a monolithic way, but there's also an organizational component. If you look at a company and seriously question the leadership from the top down for decisions they're making related to their core business, I think we should still question whether or not they can properly assemble a "very different team".

I've mentioned this in other comments: I'm not saying they can't - but they have some convincing to do.

I'm with you in that I've lost some confidence in Apple's level of quality.

But not nearly to the level you're expressing.

> Obviously you apply different standards and methods to develop life-or-death regulating software than for customer grade game and web surfing platforms.

And that's exactly why it will go badly if the average web dev is assigned to do critical software. They don't have the culture, they actually have a culture and training against it.

Imagine for a second a company doing both.

- In one office, there are the critical developments, where people are taught that quality matters and bug kill people. Also, you can't go back when shit is shipped, it's too late.

To remind you of what "critical" means. There are people in the team, including yourself, who are still waiting for the updates on the AB957 plane crash from last month where 157 people died, because they worked on the navigation system that was in THAT plane. It's pretty sure at this point that this is not linked to the crash, but we'll be more confident of what happened after we figure out what happened.

- In the next office, there are the web devs, where people are just pushed to ship more shit faster. Sometimes they say that they'll fix things later, even thought they know that they won't.

They've got a list of bugs, that is mostly ignored. They know they are shipping new bugs but they don't care, they don't even try to improve the situation.

Now, what's the difference between those two groups?

There isn't any! All the working developers could take time for a break, sit down and have one more review before they ship :D

This the hardest thing to teach to all developers: QUALITY MATTERS!

It's very challenging to teach that to the average web dev because everything they know and live-in is against that.

It's gonna be a problem to have an organization with tinder-like-web-development and critical development close to each other, they're too different socially and the web-culture is gonna dominate the critical-culture [Sadly, the path of mediocre-driven-development is easier than is-it-REALLY-good-enough-driven-development].

I'm not a down-voter, but I have to ask: which major company would you trust to develop this uncharted, potentially lethal software? All of the options seem pretty frightening, simply because they don't have a track record yet. I've got my money on Uber winning out in the long term, but only time will tell.

I would trust companies that have a culture of making fool-proof software such as major airplane and car manufacturers. I was talking to an engineer that worked on the software for the main computer of a fighter jet and he said writing code is a small part of it, writing the specs and tests for it is what takes time. I don't think any of the Silicon Valley companies are willing to match that, it's not in their culture.

There are plenty of medical device companies in the Silicon Valley, not to mention aerospace.

Uber intrigues me, but they also scare me given their complete willingness to act in less than ethical ways. Combine this with something as critical as a self-driving car and I worry.

I tend to agree with the sibling comment: Google and Tesla. It's easy to say Google since they've clearly been doing this for awhile, but they've always been on the leading edge of some remarkable software, and seem to fit the criteria of a company well-positioned to take this on.

And considering Tesla's SpaceX connection (some serious triumphs in hardware AND software), I think they're uniquely positioned to take this on.

Tesla and Google. Tesla has tons of data and already have proof of concept ready, and Google knows how to write software that makes fast decisions in an ambiguous reality to achieve good enough results.

So when Google's car runs you over, the only way they'll pay medical expenses is if you post on HN and an employee happens to see it?

Obviously kidding on the specifics because your insurance company should go after them, but let's not give Google a free pass on their systems all having "good enough" results.

I do have to say that having just switched to Google Fi, Google's support for customers that actually pay them is far beyond what their reputation would have you believe

and google has a pretty large fleet driving around the bay area. every time im in mountain view I see at least 3-4 a day when I'm just driving around

One indicator of a company's confidence in its product is how much money they are willing to bet that it works correctly. For typical consumer products such a bet comes in the form of a warranty. Although this doesn't answer your question directly, I will be looking to see how much liability a company is willing to assume.

Didn't downvote you but here is my guess: it should be save to assume that Apple will approach developing software that drives a car completely differently than the consumer software they typically produce where bugs in the very vast majority of cases do not result in physical harm to anybody. Any sane software developer/company would. It would be foolish to develop your music player like it's a car or airplane and it would be foolish to develop your car like it's a music player.

Edit: fixing phone autocorrect issue.

I completely agree that one should expect a very different approach to developing this sort of software, but that also leads to a key question that should be asked about any company approaching this sort of thing: is there a good reason to believe they're capable of this different approach?

And if other facets of the company are showing signs of fundamental problems, do they deserve trust that "this will be different"?

None of this to say they're not capable of this, but I think there are good reasons to doubt that a hardware company with a spotty track record when it comes to software has the ability to put forth the software engineering effort required to build something of this caliber.

$0.02, just my opinion, etc.

Then again, I think its safe to say that traditional auto makers are developing electric vehicles exactly the same way they develop ICE vehicles. They market them the same, they build them the same, they sell them the same. I don't think its too much to say that when a company has gotten very large they tend to do the "safe" thing and build upon what has worked in the past. Given that, I would tend to expect Apple will develop its auto"mobile" products similarly to their iOS and macOS products.

I think this example and reasoning conflate the development and marketing of a fundamentally similar product (traditional vs. electric vehicles) with the development and marketing of type of product totally new to Apple.

Is it reasonable to say that car manufacturers stuck with what worked because they are essentially swapping out the internals of the car? Yes, much like Apple sticks with the same philosophies when swapping out the internals of their computer. I don't think it is reasonable to say that car companies would stick to the same ideals when developing an electronics branch of their company, so why would it be reasonable to say that Apple would stick to the same ideals and process when developing a car branch of their company?

And of course, the use of the common word "mobile" meaning "able to be moved" is totally disingenuous when arguing this point.

Apple sells several of the most popular consumer electronics devices in the whole world, and contrary to the opinion of the techo-chamber, those devices pretty much just work. We have an interesting cognitive bias when it comes to these kinds of products: we can obsess about how awful iTunes is, for instance, without appreciating the tens of millions of lines of code that pretty reliably do their job day-in and day-out for hundreds of millions of users.

This, by the way, is true of pretty much every major tech company. It's hard to think of any of them --- even Adobe --- where the weight of all the stuff they've gotten right doesn't confer pretty much automatic credibility to any kind of new software project.

Like any company, Apple might not succeed at an ambitious new product. But you'd be nuts to bet on some random startup over Apple (or Google or Microsoft or even Adobe), at least for a "big" product.

> Apple sells several of the most popular consumer electronics devices in the whole world

But does this qualify them to build mission critical software to run devices that can kill 1 or many people with one mistake? I'm not saying they can't succeed, but pointing at the success of consumer electronic devices isn't very convincing.

And speaking of those devices, I've been buying Apple laptops since the PowerBook G4 and have owned every iteration of the iPhone except for the 6S, and while they usually work, almost every device has had some failure at some point. Some hardware, some software, but my point is, "pretty much just work" doesn't cut it for a piece of software that drives a car.

I have to disagree - moving into the creation self-driving vehicles does not come with automatic credibility. Only hundreds of thousands of hours of testing will make me feel ok regardless of who's building it.

Edit: Missed your last paragraph. I'm not betting on any random startup nor did I mention startups anywhere. I also doubt a startup is where this innovation will come from.

> We have an interesting cognitive bias when it comes to these kinds of products: we can obsess about how awful iTunes is, for instance, without appreciating the tens of millions of lines of code that pretty reliably do their job day-in and day-out for hundreds of millions of users.

Corollary: You have an interesting cognitive bias when it comes to these kinds of products: You can forget about the numerous little bugs because the software does his job good enough most of the times.

Corollary-2: You're also a dev so you empathize on how difficult it is to do anything not dramatically buggy, you even think that they must be incredibly talented and should get credibility for doing anything.

We'll see how you think about bugs when any single "bug" in a software has the power to instantly put you off the road at 100 km/h... eventually straight into a tree if there happens to be some in the area :D

Corollary: I have mountains around me. Going off the road in there is equivalent to going off-a-cliff.

I would not and I am an up-voter. It's been three years of flawed OS releases from Apple. The worst was installing Mavericks. I read the release notes and it's just Apple's e-book store. Then the random crashes and inability to reboot begin...

Tim Cook: I don't have a very positive experience so far during his tenure.

Apple has one of, if not the, best OS kernel out there. People will disagree about this, but this is based on all of my using iPhone, macOS, tvOS and a similar amount of experience with Linux, and a bit of Windows.

I do believe the XNU kernel is extremely well built and solid. If Apple does this, they may need a different kernel for real time features, or they would make XNU a realtime kernel.

Anyway, my point here really is this. I trust the XNU kernel as one of the most stable best supported out there. And with possibly the exception of seL4, I'm not sure that there are others I would trust more.

I'll ask a question now, what other company providing a full OS stack do you trust more to deliver a stable and reliable kernel for a car, that has a better track record with power consumption, uptime, or memory utilization?

> This isn't coming from a place of anti-Apple sentiment...I use quite a few Apple products...I just accept that they're pretty bad at this stuff these days.

Honestly, I can't remember a time when Apple was particularly good at software. What sets them apart from their competitors is the integration of the whole package: style, above average hardware build quality, pretty good UX design, and most importantly a heavy dose of marketing.

> Honestly, I can't remember a time Apple was particularly good at software.

I'm not following this argument.

BnL, people use Macs because of the software, not the specs. It's hard to make an OS that works that well for both luddites (GUI) and hackers (BSD flavor CLI). Who else pulled off this software feat this well?

People were pissed at loss of Aperture, because of software. Pages, Numbers, Keynote, serve lots of people's needs well. Even Preview is nice.

People use the iPhone because of software. Every new Android, tech press is agog about its specs, but iPhone switchers stay more loyal because of software over specs.

Marketing doesn't keep you using a phone, or an OS. I actually cannot recall the last MacBook ad I saw. MacBooks on TV shows these days even tend to have the logo covered.

I suspect what's going on here is that aspects of software are more important to mass market than to you. You may be aggravated at technical details, while the mass market likes the overall experience of not having to think about it.

Much like the debate of convention over configuration, Apple's value system is different than the hacker value system.

They have historically particularly good at software that iterates towards those values.

I'm a lot more concerned with whether it has passed a rigorous permitting/licensing and driven a few million miles than I am with the branding.

iOS 6 was a peak, according to my experience as a user. After this version quality started to decline.

The actual article belies the headline, and supports instead the same notion that most recent reporting has supported, which is that Apple has given up on building entire cars, but still may be interested in building software for said cars.

Why play the role of BMW and manufacture vehicles when you can play Mastercard and collect 0.5% of every penny spent on ubers/lyft when asking Siri for directions and confirming with Apply Pay and Touch ID.

If anything, Apple is gearing themselves up to profit off the movement of people and not the vehicles themselves. Seems like a better strategy if you ask me, but I'm not a strategist of any sort...

Or how about I just cut Apple out of the picture, which is what is actually happening.

Come help integrate services with us! :)

Companies write to regulators or act as amici curiae all the time. So maybe this is another "hint" of Apple, but so far we had a whole bunch of hints that media have willingly eaten and the only thing to come out of that was a bunch of crap.

This may just be Apple's lobbying operation running on because nobody officially told them to stop.

There were previous rumors last summer of Apple testing at GoMentum Station, which is part of the old Concord Naval Weapons Station north of Oakland. But nothing much has been reported about that. It's not that secret; BART tracks go over it, you can see most of it from nearby roads, and a high school faces it. Honda has been testing there, using an Acura with the inevitable spinning Velodyne on top. Apple doesn't have a California autonomous vehicle testing license, although 19 other companies do.[1]

[1] https://www.dmv.ca.gov/portal/dmv/detail/vr/autonomous/testi...

I can't see it happen. The company is struggling to ship bread and butter products like laptops.

Can you explain what you mean by “struggling to ship”? Maybe you’re talking about their previous quarter results which were down a bit year-over-year, depressed by people waiting for laptop updates? I haven’t heard any news of notable production problems.

Supposedly they’ve had record online orders of their new pro laptops http://www.independent.co.uk/life-style/gadgets-and-tech/fea...

Oh really?

> The MacBook Pro made more money in 5 days than most laptops in the whole year, Slice added. "In only 5 days the new MacBook Pro's sales already equal 78 percent of all the revenue generated by the MacBook 12-inch since it became available in April 2015." (http://www.forbes.com/sites/brookecrothers/2016/11/08/new-20...)

That kinda makes my point.

You had an explosion of pent up demand for a neglected core product line.

There's no accounting for taste...

I wonder if Apple would launch a self-driving Uber competitor, and I wonder if they'd make it iphone-only and deeply integrated, like their other services.

Will Apple cars have access to more information about iphone users than Google/Uber cars? Enough to be anti-competive? It's really interesting to see their hardware-advantage-driven approach to services come to bear on different industries.

How much of an edge do you get from information? The average car fare wants to go from A to B, for some price, probably usually cheap. They are going to be pretty willing to share A and B.

There's probably some more sophisticated planning you can do to increase utilization by using further information, but I bet that ends up being a pretty small edge.

I don't drive so ask HN: I seem to recall apple saying years ago that they wanted to make the entertainment/ui for cars, did that ever happen, was it a success?

http://www.apple.com/ios/carplay/

Not sure. It's a tough sell imo, because carmakers have a maniacal obsession for shipping garbage entertainment systems.

I've only seen it in a few cars, as my social circle are mostly cheapskates with old cars. It's a pretty good system.

Used to be true, but now, ads for any CarPlay-enabled vehicle seem to almost exclusively feature the stereo as the key selling point.

A notable holdout: Toyota, who has arguably the garbagest entertainment system.

That's great news. I haven't been car shopping for awhile.

A friend has a 2016 Honda Pilot. The enerertaintment system on this thing is a pure shitshow... I think it's 3 clicks on some awful touchscreen to turn the radio on.

Will there be an "escape" button? ;-)

Reply on news.ycombinator.com