Hacker News Reader

Comment by contingencies | Hacker News Reader

1. Maybe partly, but 1B USD seems a lot to pay for that.

2. Self-driving is not going to happen in China the way it happens in the US. The roads are full of pedestrians, e-bikes, people breaking any sort of nominal rule or law... it's too accident prone. Plus you have the extra challenges of the politically accuracy-challenged navigation systems. No, I don't think this reason floats.

3. Maybe partly. Locking down stable long term CNY income via a one time USD investment seems like it could even be seen as long term corporate tax planning.

View on HN

> Maybe partly, but 1B USD seems a lot to pay for that.

Apple has at least $216B cash on hand as of 2016 [1] and majority of them are held overseas. $1B strategic investment with all said benefits sound like a VERY GOOD deal.

> Self-driving is not going to happen in China the way it happens in the US. The roads are full of pedestrians, e-bikes, people breaking any sort of nominal rule or law... it's too accident prone.

Probably not for a long time, yes. But I would argue that's exactly why Didi's data is MORE valuable for autonomous driving modeling to be based off. As Elon famously said[2] to George Hotz about his comma.ai demo, 99% of the self-driving is easy to do. The hard part is the 1% edge cases and billions and billions of data points for engineering validation. Apple has a very aggressive timeline for their self-driving EV, buying that massive amount data from a harsh environment serves that purpose REALLY REALLY well and it's an efficient use of capital resources to achieve their goal in time.

> Maybe partly.

And lastly, it's hard to argue a ban on Apple when Apple can say to the government that hey we've investing $1B to grow millions of driver jobs and the local tech economy. It's about jobs!

[1]: http://www.marketwatch.com/story/apple-isnt-really-sitting-o...

[2]: https://www.teslamotors.com/support/correction-article-first...

> But I would argue that's exactly why Didi's data is MORE valuable for autonomous driving modeling to be based off.

We are assuming Didi/China will allow Apple access to the data ...

Is rule of law in traffic important for self-driving cars? If self-driving cars are based on neural networks rather than on rule-based systems, do we have any kind of intuition about under what kinds of traffic circumstances they will do well?

I would imagine a (truly) self driving car wound not assert itself aggressively - but carefully - in a crowded square and would therefore not make any progress at all. A steady stream of pedestrians jay-walking, knowing that they had nothing to fear. (Or at least, believing they had nothing to fear.)

See this gif: https://i.imgur.com/loW46yj.gif

Humans have the luxury of being able to charge into an uncontrolled intersection and trusting other drivers to make room. Sure 1% of the time it's going to cause a fender-bender, but that's just what you have to do in order to get through the intersection.

There's no way that taking chances like that would fly with a self-driving car. Even though the human passenger would have done the same thing, they're still going to feel like the car company owes them for "the car's mistake" when it's involved in a collision.

On what grounds do we have that trust?

You can just as easily teach a computer to trust people to make way as you can teach it to assume people won't make way.

Yes, you could. Like I said. And like I said, good luck marketing a product that drives recklessly into a busy intersection hoping that people will move.

Re: 2;

Perhaps Apple is thinking that this is the hard part. Nailing down the most difficult areas of self driving cars will both give them the market in China/India or other areas with similar traffic, and also leaving the "easy stuff" for last, which will only need to be supplemental.

#2 sounds like San Francisco ...

You need to drive outside of the US. Try New Dehli.

SF is busy, has tourists, and asshole taxis, but is just a normal city.

Delhi has 6 lanes if fluid traffic that mixes like a stream, cows walking the wrong way, elephants, tuk tuks that have heard of this thing you call lanes and find it a silly idea, and pedestrians that show no fear and believe they are made of iron.

China can be very similar.

China traffic (even Beijing or such) is absolutely crazy compared to anywhere in the US, and I hear that India is an order of magnitude crazier.

It's not even really comparable. Anywhere in the US you at least have lanes of traffic, people driving on the right side of the road, and don't have a gazillion 1930s-looking tricycle carts weaving in and out everywhere... Scariest part to me about China traffic, everyone covers up their seatbelts so you can't use them.

Driving in China is absolutely not comparable to driving in India

Reply on news.ycombinator.com