Hacker News Reader

Comment by torginus | Hacker News Reader

I'm afraid of the future with regards to electric cars. The simplicity and efficiency of current designs means that any further improvements in car quality are pretty unlikely, save for battery improvements. This suggests that electric cars of the future will be like the smartphones of today: containing largerly the same, or equivalent components with very little to differentiate between them, which means the competition largely boils down to price. And low prices can be only achieved through economies of scale, leading to a winner takes all situation.

View on HN

If little technological innovation happens and components become commoditized, it becomes easier for new players to enter the market. See for example the fashion industry. There haven't been any improvements to T-shirts in the last decades, but there is more than one company making them.

A major point of difference between car manufacturers will be software. Yes software will eat the car industry! Tesla and Silicon Valley have a huge advantage right now. But don't bet against Detroit, they will catch up because if they don't it's the ball game.

Also I've been predicting for some time that there will be a renaissance in car manufacturing. There's already a company in Italy selling a 'sled' complete with batteries. You get six people in a garage in Detroit that are good with carbon fiber and they can become a car manufacturer. I bet there are lots of under served niches not being filed.

> There's already a company in Italy selling a 'sled' complete with batteries.

Volkswagen will also license their MEB electric platform to anyone who wants it.

e.Go will be using the MEB platform for their small city cars: https://electrek.co/2019/03/07/e-go-electric-car-vw-meb-plat...

Ford will be using the MEB platform for at least one model: https://www.cnet.com/roadshow/news/vw-argo-ai-investment-for...

My professional opinion is that Tesla seems to have an advantage because other manufactures do not invest in software properly. The moment one of the big manufactures embrace proper best practice software development and security (i.e. pay for it), that manufacture will gain a significant advantage.

They are massively investing now. I am working for a service provider and we are rejecting several contracts due to lack of man power ( and we are massively expanding)

The key is not just invest and expand, but embrace best practices. It matters how your structure the entire thing and basically something like NASAs programming guide lines, and a CI pipeline with rigorous automated test. The budget should reflect significantly higher allocation to QA than actual development.

Of course the infotainment software do not have to adhere to such stringent requirement, but if they don't push OTA updates for that, then they might as well to avoid crappy software.

True, however you understimate the role of software in current automotive design (ICE).

Car manufactures have been doing ECUs for years. That is software that needs to work without bugs or it will destroy the engine. You never hear of bugs in them (it happens, but you don't hear of it) because they have figured out best practices that are unknown outside of cars and aviation.

VW emissions was not a bug, it was intentional.

He's not talking about ECU firmware, he's talking about infotainment software. You can't use the best practices for ECU firmware development as you use for other types of software development: it would never get done! Safety-critical software running on bare-metal or an RTOS (like for ECUs, ABS controllers, avionics, etc.) isn't much like software for doing GPS mapping and other infotainment functions, and the development process is entirely different. The latter has orders of magnitude more lines of code and is far simpler in scope.

The OP is right; currently, automakers basically outsource all their infotainment software, and generally do a poor job with it, though it's getting better. Considering how much drivers now interact with these or similar systems (usually on their phones), to do things like stream music and especially navigate, automakers should be doing a better job integrating this stuff into the cockpits for ease of use and safety.

I don't agree with you. There are several tools used in the creation of ECU firmware (and writing it). The software used for detection of car on rollers or road (diesel scandal) was one such tool for example.

What the heck are you talking about? Are you responding to a different post or something? You're not addressing anything I wrote there at all, you're talking about something completely different. My main point was that you can't use the practices used for ECU firmware for other types of software (anything that uses a general-purpose OS), and it's true, it's an entirely different kind of software development.

I was not really addressing ECU programming - that is an entirely different beast. But regardless, Toyota was involved in a scandal where they tried to cover up an error in the ECU that resulted in at least one death.

A subsequent review of their ECU code was quite worrying - it was to put it mildly, a cluster fuck that would result in deaths sooner or later.

The only real difference is the drivetrain, and only enthusiasts really care about the technical differences in ICE drivetrains. Interiors, cargo capacity, reliability, fit and finish, ride quality, etc. will all still differ between BEVs.

E.g. NVH has been getting better pretty much every year despite the fact that car companies have been working on it for decades

Reply on news.ycombinator.com