Featured Tesla Autopilot recall probed by safety regulator following new crashes

Where did you get 1.8? My sources show the national average hasn't been that high in over 30 years, since 1991.

My reference point for what humans can do, is the 1.08 rate achieved in 2014, and almost replicated at 1.11 in 2019. With the arrival of the Pandemic and post-George-Floyd de-policing and whatever else, the rate increased to 1.34 in 2020, and 1.37 in 2021.

But remember that these are total fleet averages: all weather, all conditions, all roads, all drivers, and all vehicles. Including large numbers of much older model vehicles whose physical crash safety is far behind that of the newer models representative of virtually all FSD-equipped models. E.g. the average age of vehicles on the road is over 12 years, which is several crash-safety design cycles ago. I don't believe any FSD-equipped vehicles are even close to the old.

To make a proper AP or FSD vs without comparison, much more data is needed to allow a sifting out of the data selection biases caused by different mixes of model years, demographics, and usage.

 

I think in Mike's situation my car slows down and stops all the time. If I want to make the light I hit the accelerator and it stays in fsd beta. If I think its dangerous like you I hit the brakes and it leaves fsd beta. If I want to take over I can and should in this situation.

 

My data comes from the IIHS and it looks like yours too, but my 1.8 is off, I might have took it from some other sources or confused two data set. It's about 1.3 since the 2020, which I'm surprised, because I would have thought that with less people on the roads due to the pandemic, there would have been less accidents, not more. And it's not just the per 100 million miles travelled, but across all data sets.

There are obviously more travelled miles in the nicer months, but it's also where there is the highest percentage of deaths too.

Yes, more data is needed and there is no other vehicle manufacturer than Tesla that has a an almost 100% connected fleet (I don't think the Roadster is connected). Tesla has released early last April that they reached one billion miles driven in FSD.



It has increased exponentially since it was opened to the public and with the free trial going on since April first (second batch of cars started to receive it late last week), it will be interesting to see what it did to the miles driven in FSD. So far, no deaths have been recorded since the trial started.

 

Any 'surprise' element should have faded a long time ago. The increased road misbehavior has been in the news almost 4 years. The increased death rate statistics, news for more than 3 years.

 

I wasn't aware because for us in Canada, the number of fatality rate as fluctuated around the same numbers
2014 (like for you) with a slight increase in 2022 back to 2018/2019 numbers.

Fatality Rate (per 10,000 motor vehicles registered)

Year Fatality Rate
2003 1.47
2004 1.43
2005 1.50
2006 1.43
2007 1.34
2008 1.15
2009 1.04
2010 1.02
2011 0.91
2012 0.93
2013 0.85
2014 0.78
2015 0.79
2016 0.78
2017 0.76
2018 0.77
2019 0.69
2020 0.66
2021 0.69
2022 0.73

We only have "per billion of kilometers" for 2022 and that value is 4.7, which is 0.75 per 100 million miles if my math is right.

 

In the U.S., the Pandemic brought a first big step up in misbehavior shortly after the stay home' and 'work from home' orders went out, traffic dropped, enforcement contacts plummeted as police didn't want to catch the virus either, some drivers took advantage to drive faster, and stories of extreme speeds that police were still issuing tickets for, increased.

Then came the George Floyd murder, protests against broader patterns of discriminatory police use of force, even more de-enforcement (some of it by new law) and road misbehavior. Anecdotal evidence at first, but the preliminary quarterly traffic fatality stats were not far behind, confirming it.

Your Canadian rate is certainly much better than our U.S. rate, and seems to not have an equivalent Pandemic bump. A number of European countries much better still.

 

Thinking about this, using the chart posted by Tesla, from the start of the year to the end of March, there have been around 500 millions miles driven with FSD and no death was reported. Using the year 2014, which is the year with least amount of death per 100 million miles driven since 1975 (1.08), FSD, just to be "as safe as" that best year, it should have had over 5 deaths before the end of March. And furthermore, since the launch of the trial in early April, a lot more miles have racked up using FSD and there are still no reported deaths using it. This, by itself is an accomplishment shared by both the technology used and the people using it staying engaged with their driving. Kuddos to all of them and shame on the first one that will taint these immaculate results.

 

Post #243 cross posted a tweet that was created by Tesla AI... It claimed 1 billion miles driven on FSD (which version is not specified), half of that driven in Q1 (the first 3 months of 2024). I'm curious how they can make that claim when Tesla did not force FSD onto new buyers until the first few weeks of Q2. I did not see an announcement that made free FSD mandated for all existing 2,300,000 Teslas on the road in that month.

the NBC news report said:

It seems like the Tesla AI failed the first rule for posting graphs on the net. It does not produce sources for the data represented by the graph. It does not even label the axis for X and Y. Why can't Tesla ever do things normally?

 

Version 12 was released to the public March 26th so safe to assume it was version 11 (released in March 2023). Version 12 when released was only for vehicles manufactured before 2022. It's only been since May 6 that the newer vehicles started receiving version 12 and their free trial.

The numbers of people with FSD enabled car grew from a few thousands in late 2021 to over 300,000 in early 2023. FSD was opened to anyone in late 2022 and as the software progressed, so did its usage, and it wasn't "forced" onto new buyers, where did you get that idea? All NA owners with HW 3.0+ were given a free month of FSD when they updated to a version that contained FSD v12 (20024.3.x for older vehicles and 2024.14 for the newer one). Nobody was forced to use it, but of course, for many, curiosity took over and at least tried it. And some, like me, were impressed by its performance and decided to subscribe to it once my trial month was over.

As for the graph, I think that your bad faith is showing here. The source is them, Tesla. It's not investigating journalism here. It's a twitter post from them, which, shocking, have that data. It's showing the progression of FSD miles driven since it became available, not a graph from a science literature. With a graph titled "1B FSD Miles driven" and the Y axis going from 200,000 to 1,000,000 and the X axis showing the dates from the first release until now, do you really really need axis titles to determine what the graph is all about? Common man.

 

Independent from the whole discussion - this is similar to Elon Musk Posts. No company would post stuff that makes looks them bad - actually they cant really as a shareholder traded company.

Not sure why there is always so much trust in what Tesla or Musk posts - even if its not a straight lie they only show you what they want you to see (their best side) and leave everything else out. Especially if they know about people dying from their product.

Based on this people would still smoke like crazy and eat sugar for "health" reasons.

 

Actually, yes, I do want a 'graph from science'. And now that Anders points it out, the graph's wiggles and inflection points and non-linear x-axis don't quite look like 'science'.

Show me the actual data, not something that looks suspiciously like a generative AI invention. This one is even clearly labeled as from an AI handle.

 

Ok then, here's this graph from their Q1 2024 Update. 500,000 happened sooner than interpolated from the twitter post. Nevertheless, look at the sharp increase with the release of the trial. It looks like 200 millions miles were added in just 3 weeks. At that rate, they're probably at 1.5 billions by now (with the new batch of free trials that started May 6). I interpolated the data for January 2024 (that blue vertical line and ticks) and it's a bit over 800 millions and by April 21 2024, they were at around 1,3 billions. So, the 500 millions from January I had earlier still stands and the "zero death", unless proven otherwise, still stands. Sorry Dan, still no X/Y axis titles but this time, the X axis is linear

 

Once again, the charts and the PDF only gives a partial view. Did you even notice that nowhere in the entire PDF was there mention of the government probe? No mention of recalls? No mention of accident rates or injuries? No mention of the deaths?

There WAS a death of a driver in the last 12 months. April 17, 2023 — "The Tesla smashed into a fire truck in Walnut Creek, killing the driver and injuring four Contra Costa County firefighters." Tesla told police that they were uncertain whether or not the car was running under a version of FSD or other automated driving. They (Tesla) claimed that the severe damage interfered with them collecting the data.

Within the last 6 months there was a driver who drowned in her Tesla when she accidentally drove into a pond. The doors would not open.

I'd interpret the lack of mention of the known deaths can only mean that Tesla is withholding data in the hopes that no-one will question it.

 

Yeah, fluff because it doesn't match your view, got it.

Ah common man, the probe was sent May 7th, several weeks after the Q1 report. Are they supposed to read the future now? And you seem to be stuck in the past yourself since that accident was on February 18, 2023 (not April 17) and that's over 12 months ago (and so is April 17 2023 BTW). Ok, let's account for it being in FSD, from Feb 2023 until now, 1,2 billions miles were driven in FSD, that should account for 13 deaths, according to the BEST years since 1975, no? Where are they?

That lady that drown was drunk, backed up into the pond and the door wouldn't open because of the water pressure. They do have manual release, exactly where you would find a normal handle on a car.

Haven't you noticed that anything related to Tesla is amplified in the news. Any death would have been mediatize, like these 6 kids in a Tesla where three survived from that high speed crash they were in earlier this week. Too bad a new feature coming up isn't yet available because it could have saved their lives. Anyone with teens will sure welcome this addition.

Tesla to Add Restricted Driver Profiles for Safer Teen Driving

You're grasping at straws to prove (and fail) your point.

 

While I find your attempt to be witty and offensive at the same time is amusing, you keep making assertions that are not backed up by anything. Is there a twitter account for TeslaAi?. If there is, what's it's objective, other than to cast out untruths that look official? Have you noticed that people who post negative information about Tesla are not common on Twitter? Their posts tend to get deleted early.

I can understand why a Tesla owner (stock or the automobile) is terribly committed to supporting Tesla. It must be scary to see a large investment that's just a few posts away from losing a lot of it's value at any particular moment. BTW... did you hear that there's going to be a huge tariff on cars brought in from China? Wasn't there something in that PDF that you shared that said Tesla was getting a lot of cars from their plant in China during the 1st quarter?

 

They are finally adding a feature that has been available for many cars (including Toyota's) for at least 6 years. Good for them. It's launched with the "Toyota" app on my phone. And it's free.

 

I gave you data from their quarterly report to investors and yet, you dismiss it as "fluff" and you make stupid remarks about not talking in their reports of a NHTSA probe that wasn't even out back then, hence why you're grasping at straws.

And now you deflect by talking about "investment", well, sorry to burst your bubble but no cars from China is sold in the USA. At most, they import some batteries, that's it. It's not the doom and gloom you were hoping for, sorry /s

 

True, but unlike Toyota, their cars evolve with time, adding features that others might already have and some that are new to the industry. My Prius Prime has been stuck in time since 2017. They can't even fixed that stupid kWh/100 km gauge display ffs, while my Model 3 have received many updates in the 2.5 years I've had it. Like the recent update that sends a video clip to your phone when an alarm has been triggered so you can see what has caused the alarm to trip and see if you need to take immediate action.

I don't have it on my Prius Prime and no matter I do, I'll never have it (shocker), but if it's like Lexus Enform Remote, you do have to subscribe to it and no, for the Lexus crowd at least, it's not free. There is a trial but you have to pay once it expires.

Once Tesla releases its version in an OTA update, even vehicles as old as 2014 will be getting it, while like I said, my Prius is "stuck in time", but we digress from the main topic.

 

I'm not worried about the doom and gloom. I have no interest in Tesla other than getting them to do all their data gathering on private property instead of at the hands of amateurs in and around my neighborhood.

The picture just looked good to me. These Superchargers have been in place for several weeks but the only part that's functional are the garbage cans.

 

Here goes the fear mongering again. I gave you data that shows the system is already safer than driving alone. In the latest Wham Bam Dash Cam weekly video, out of the 20 clips, 9 could have been prevented if the offender would have been driving a Tesla in FSD.

As for the Supercharger installation, not sure what that has to do with the current conversation.