Ford Mustang Mach-E has a mile of wires it doesn’t need. That’s a big deal

I mean that's a dumb metric... raw number of recalls? Ford's commercial line does enough volume to be its own manufacturer, and recalls aren't fungible, and different numbers of vehicles are affected...

I'm pretty sure someone literally opened the NHSTA site, did zero normalization, and took their headline: https://datahub.transportation.gov/Automobiles/NHTSA-Recalls...

-

Taking 3 examples (note recalls affect cars older than the current year):

- Ford, 68 recalls affecting 8.7 million units, 1.8 million vehicles sold in the US in 2022

- VAG, 46 recalls affecting 1 million units, ~300k vehicles sold in the US in 2022

- Tesla, 20 recalls affecting 3.7 million units, ~500k vehicles vehicles sold in the US in 2022

VAG has significantly more recalls compared to their sales, but few vehicles affected per recall.

Tesla has one of the highest vehicles affected per recall of all manufacturers despite the lowest number of recalls.

All in all this is way too complex to reduce to a catchy headline.

The article says that the wiring harness is longer than it needs to be. My guess would be that there are several hundred wires in the harness, each signal paired with a ground, so an extra few feet (an extra fraction of an inch per mounting point?) would really add up. That extra length might make it easier for the installers to wrangle; efficient design often comes at cost.

Edit: and, come to think of it, this is a Steiner Tree problem with some very interesting constraints. Optimizing very large instances like this is a Hard Problem. Even if everything is locally optimal (unlike my hypothesis above), it can still be far from globally optimal. Recomputing a solution, spending longer on the computation, can make a big difference!

Wiring is a very hard problem in vehicles. It’s why you see physical switches being switched out for a touchscreen, and some of the tech the big automakers are most excited for are, at the end of the day, just to reduce the amount of wiring required for each car.

To add to the other replies - the networking configuration of the car plays a huge role as well! Is it CAN bus? Is it Ethernet? Is it a mix? How many computers are in it? How are they connected? Have you centralized the microcontrollers to some degree? What about connections between microcontrollers?

Just like microservice oriented architectures and monoliths, you may choose the form of your destroyer, and an extraordinarily well-connected topology with no centralized points to bounce between can lead to ass loads of wires :)

I'm convinced of the quality of Tesla's internal networking - I don't have concrete data on this but the capabilities they showed in 2018 such as over-the-air updates told me they had to be more sane than their competitors. To be able to update any part of the car like that I'd imagine they centralized a TON of the random bits and bobs into a few Ethernet connected machines even back then. On top of that, the rumors of the telemetry they collect is the stuff of legend!

The problem with a lot of these car companies is that they're a little behind the tech curve. I left automotive in 2018 so I may be old news here but in-house out-sourced was still a large debate when it came to the design of such systems. It will take a while for them to bubble up leadership with relevant experience to make the right calls - maybe with the layoffs at big tech co's it'll happen sooner than later? Automotive companies I interacted with in 2018 definitely knew then what they wanted, the hard part was finding the right people to deliver and manage.

The article implies that wire weighs enough to require additional battery, which then costs additional weight.