Comment by cmurf | Hacker News Reader

Comment by cmurf | Hacker News Reader

I put Ford's 2021 and BMW's 2025 and so on, into the autonomy is novelty category; just like the Roomba (yes the robot vacuum cleaner). There is no possible way in that time frame that either the software, or public infrastructure, or laws and regulations, and insurance are all going to magically go "full autonomy" in major cities, let alone outside of them, in that time frame. We've been trying to do this with trains and planes for a long time and while highly automated to a degree that might surprise some people, they are not fully autonomous. And yet they don't have to share infrastructure with significantly sized objects that not only don't use the same software, but not even the same spec including fault tolerance.

And another reason why I'm gonna be a bit negative Nancy on this, is because the entire software industry sucks when it comes to software updates. About the best at it has been Apple and iOS. The worst is a toss up between the automotive industry's own proprietary navigation systems (crap UI/Ux, crap nearly non-existent ship it and forget it update) and mobile device manufacturers. There will need to be software updates for more than a decade for cars, to account for all the unknown unknowns as infrastructure and rules all change, to account for interactions with other proprietary systems, as well as humans. But this industry sucks at software updates.

Anyway, good luck with that BMW.

> But this industry sucks at software updates.

Except Tesla.

I think we're going to see a race from two ends of the spectrum: Can Tesla ramp up production faster than traditional manufacturers learn how to build software?

I think you've hit the nail on the head here. It is an interesting race to the middle and either way consumers stand to win from it.

This is exactly the kind of situation that I think typifies the positive aspects of capitalism. You have two companies that each have one half the piece of a puzzle and are furiously trying to compete against each other to produce the other piece.

We will see a lot of different techniques tried and that could potentially lead to a lot of tangential benefits to the economy along the way.

It's more complicated than that. Tesla needs to increase their production much faster since the other manufacturers combined creates the long tail that Tesla will be competing with. Also, traditional manufacturers may rely on some industry software gold standard that lets them outsource the software development if they believe the core competency is unrealistic to develop in-house without 1) software mindset 2) time 3) money 4) organizational structure.

I'd bet on the legacy manufacturers.

Musk has a reputation for firing senior managers that give him answers he doesn't like, or that fail to meet his unreasonable demands. That level of management meddling doesn't bode well for Tesla's manufacturing future. Nor does "putting parts in the car knowing [they] are bad", if the accusation (by a departed employee) is true. (http://www.wsj.com/articles/elon-musk-sets-ambitious-goals-a...) I have previously questioned, and still do, Tesla's PLM.

Legacy manufacturers do know how to do software, despite the general lousy state of in-car entertainment systems arguing otherwise. They also know how to manufacture a quality product at scale. It's an easier task for them to bolster their software than for Tesla to learn what they already know about manufacturing and ramp up manufacturing.

youre making the assumption that the software problems (not yet solved) have easier solutions than the manufacturing problems (solved multiple times).

Legacy manufacturers do know how to do software

Err ... no they don't:

   Toyotas Unintended Acceleration
      and the Big Bowl of “Spaghetti” Code 
   https://news.ycombinator.com/item?id=9643204

OK well I'm 100% confident Ford and BMW will be in existence in 20 years. I'm maybe 80% confident Tesla will be, in any case it's definitely not 100%. So even setting aside the software updates for 10-20 years question, as if that'll even be a thing with these cars, I have less than complete certainty Tesla will be around in its present form making cars at all.

I would imagine intercity travel is immensely easier to solve than intracity travel.

On limited access highways, there are no pedestrians, no bicycles (mostly and not in traffic lanes), no turns, separation of travel in different directions, generally well-defined lanes, etc. Non-limited access highways aren't quite as straightforward but they still seem immeasurably easier than a busy city.

Yes, speeds are higher but dealing with mostly well-defined problems like sudden braking at speed is something it seems as if computers should actually be pretty good at as opposed to all the random events that happen on city streets.

[EDIT: Yes, I misread. I'm in violent agreement with the parent. :-) ]

I think you read the parent comment wrong

Indeed I did.

Reply on news.ycombinator.com