Comment by ethagknight | Hacker News Reader

Comment by ethagknight | Hacker News Reader

Parent Full thread

I’ve had a Yukon with supercruise for 18 months. It is fantastic for highway driving, particularly, we’ve driven it west from Tennessee to Colorado and Florida. Other comments claiming it’s the same as other standard driver-assist packages are incorrect. Totally hands free, no “ping pong” with the lanes.

That said, when trying to BUY a supercruise vehicle, the sales guys were clueless, I had to review the stat sheets of each car to see which ones did or did not have it. GM is treating this technology as “surveys say 2% of the market wants a self driving vehicle” (incorrect question) compared to “install on every vehicle as an available subscription, let the sales team earn their way.” I’m sure the tech is expensive, but it can’t be that much more at the OEM level.

I also have a Tesla with FSD. FSD is truly amazing but still struggles with edge cases like, my office has two entrances, one is blocked by barrels, it tries to turn into the barrels every time.

From my understanding, super cruise only works on pre-approved roads/road sections, whereas FSD works just about anywhere. When I last checked my commute to work, the main highway I take becomes unsupported half way to work, despite it being the same bland typical highway layout.

I really don't want to buy a Tesla, but from what I can tell, nobody has anything close to what FSD offers.

> nobody has anything close to what FSD offers.

From what I read about it, you're absolutely right. Down to the point of unwanted veering towards road dividers and other vehicles.

> Down to the point of unwanted veering towards road dividers and other vehicles.

Our 2024 Kia does this every 10-20 miles on average. You don't even have to enable lane assist. It also randomly decides to weaken the regenerative braking, or switch out of single pedal mode.

If you disable enough of the safety features it stops doing these things, but, for instance, that means you can't have blind spot detection that works when you're going over 5mph and a working steering column at the same time.

Yeah, people act like some Tesla problems are bad, but I've certainly used systems that are worse. Just in a few miles, I've had a Nissan accelerate towards stopped traffic aggressively enough to trigger its own collision warning. It'll also fly try to drive off the road in relatively shallow curves, triggering its own lane departure warnings.

The Tesla system isn't perfect, but it is far from unique in being imperfect.

That's true. I took a jab specifically at Tesla, because it's generally framed as the "leader of the pack and problem free to the point that you can sleep in it".

FSD is a very hard problem, and the systems deployed are at best alpha level. These things should be "hyped responsibly".

Not paying attention while FSD drives for you ended with a motorcyclist crushed to death under a Tesla, and a manslaughter charge for the driver near me this summer.

SuperCruise and FSD are not comparable.

SuperCruise does what it claims to do, keep it up the middle on listed corridors. Most highways and interstates are eligible. I love both systems.

FSD is truly almost there for unsupervised, point to point driving. People want to virtue signal and complain about Tesla but it is an amazing piece of hardware being mass produced today. My car is 3 years old.

Any thoughts on OpenPilot? I’ve been hearing good things but would be cool to do a comparison.

I worked in this industry. No, FSD is not almost there. Not even close. What matters is the long tails of events.

You might "feel" it is almost there because it gets it right 99.9% of the time but that is still way too many accidents and injuries in the long run. And the work to go from 99.9% to 99.9999% is 1000x more complicated.

We need to compare against the real human error rate (including drunk drivers, sleepy drivers etc). What is that error rate?

Also, fault tolerance error rates don’t work that way - difficulty increases exponentially as you increase fault. In other words, it’s much more difficult than three orders of magnitude to go from 3 9s to 6 9s - it’s easily 5-6 orders of magnitude.

When you're playing candy crush and FSD kills someone, whether or not it's drives better than the average driver is not something the judge and prosecutor are going to consider.

First, that’s because FSD is legally only L2 and thus legally you are required to pay attention. It has nothing to do with the engineering realities of the impact on improving safety.

When L5 becomes available then this becomes a different calculus. And it’s honestly questionable about how good the characterization actually is since Tesla’s L2 FSD seems to outperform Mercedes’s L3 and it’s more a matter of the liability the manufacturer is willing to take on vs objective measurements of quality.

yes, the long tail of difficult event is exponentially more difficult to handle. That's why I said above that people "feel" it is ready but it is nowhere to be even close to ready.

The average crash rate for human is one every 500k miles.

I disagree with you. Glad you worked in the industry, I am using FSD daily, it is really excellent.

Drive 100 miles in your sleep.

You'd trust your partner or parent to do that for you ...

Got it. differing expectations. Im not expecting 'unsupervised' out of the system, I dont think I would trust any system to transport me or loved ones un-piloted. Even when being driven in a taxi, I am still supervising despite not having much recourse beyond barking at the driver.

That said, based on my experience with FSD, I'd be tempted to take a nap if it let me, reinforcing my initial statement that it's very darn close.

If you drove with a very very bad driver that crashes every 50k miles (average is every 500k miles), you would have exactly the same feeling and you would also be tempted to take a nap.

100 Miles literally don't matter. Even 1000. On average accidents happens every 500k miles.

Are you ready to have your tesla drive FSD with you sleeping in it for 500k? You are letting your feeling dictate that FSD is ready. The math is more complicated.

It’s wild to me anyone would think turning on barrels is an edge case.

This i what is so tricky with Self Driving. People "feel" it is almost there because most of their rides are mostly ok. However to make a system truly driverless you need to master the long tails of difficult events and FSD is nowhere even close to do that.

Going from 99.9% to 99.99999% is what makes a system truly driverless and where most of the work is. Waymo is way way ahead of FSD for this.

I know essentially nothing about machine learning, nor about what approaches to ML that Tesla or Waymo are using. Is there a Metcalfe-type effect possible here? Where the better FSD becomes, the more people use it, finding more edge-cases, so the driving get better, so more people start to use it. With the end result that the learning starts to get better/faster with time, in a positive-feedback like mechanism?

The answer is very strictly no.

Your comment/question seems to assume that somehow accurate classification of performance is possible, which is demonstrably false assumption.

First, if somehow system-internal watchdog was able to detect erroneous outputs for training, it would also be able to stop those outputs propagating to control system live, leading to zero errors. Such a watchdog requires self driving to be quantified analytically, leaving implementation entirely testable and therefore control system not passing the testsuite (i.e. exhibiting erroneous behavior, i.e. still having "edge cases") not deployable in public. Many words to say that that in practice with driving being somewhat loosely defined you need humans in the training loop.

Second, vehicle operators are not only unqualified to accurately monitor behavior of these autonomous systems, in part due to not being formally trained on safety systems in general and the system they are monitoring in particular, but on top of that incentivized to underreport erroneous output.

My prediction has been that we are going to see increased number of accidents with ADAS deployments, not less, before the number can start dwindling.

Reply on news.ycombinator.com