A Tragic Loss

The description of the accident surprised me as well, but your comment makes sense. Vehicles getting under trailers is something I've only seen in (US) movies.

Underrun guards are a good idea but not yet practical for trailers in the US. The reason is that loading dock approaches have far too much variance in ground height, and the empty space under the trailer is essential to have. If guards were in place, they would come into regular contact with the ground and either damage the trailer or the truck would be 'high centered' and stuck, resting on the guards. I suppose the guards could be made to fold up out of the way but this would add weight and complexity, and probably affect strength. Trucking companies in the US do things like use 'super single' tires that are extra wide to span the width of dual tires. This saves the weight of two unneeded sidewalls and save a few pounds weight. Because truck design is so influenced by large fleet purchases, even very tiny changes to weight are avoided. Lighter > safer.

Correct. The EU requires underrun guards on all vehicles except tipper trucks, which cause a disproportionate number of fatalities still.

There was an attempt to introduce legislation in the US in the 90's, I believe, but it was killed.

The guards are mostly just thin aluminium panels on steel supports, so if the car impacted them front first, going at highway speed(lets assume 50-60mph?) they would have made no difference whatsoever.

The thing I fear now, is fearmongering by people who dont drive Teslas, havent spent time with Autosteer/autopilot and ignore that this was a freak accident, that probably would have killed the driver regardless of using autopilot.

This is the future of roadsafety.

I'm not even sure it counts as level 3, it's more like a high-end Level 2 system.

Level 3 requires the car to detect "all" conditions when something is going wrong, and provide the driver "sufficiently comfortable transition time". If there is any class of event (such as a 1m overhang) which the vehicle can't safely cope with (perhaps coming to a complete stop before requiring the driver to take over), then the driver really has to be paying attention all the time, making it a Level 2 system.

And there is basically no way to deliver on that promise on today's highways without a suite of sensors and computers that are capable enough to do full Level 4.

Yes it has a better driving record than humans do already.

Really speaks to the promise of autopilot.

There's a 10% of abandonement for each additional paragraph in the post so you need to put the most important facts first to ensure your message is conveyed. Their priority is to restablish confidence and there is nothing that can be done to bring that person back to life. They are trying to save other people's lives by pushing their idea for a safer driving.

Well this is 9 days old so I doubt you or anyone will see it. But I really want to say that this is just how language evolves. Phrases appear, become common, and even change meaning, subtly or entirely, from the original.

I don't think it's a waste of space. I think it conveys some information. If you remove it from a paragraph, the paragraph feels different. The meaning slightly changes. In a literal sense it may not mean anything, but I feel it does convey some intention from the speaker.

I kind of agree but I think there's a few subtle changes as to how you phrase it.

I sometimes type "Please note <something important>" to make sure the recipient pays extra attention to this bit and actually notices it, because people have a tendency to read through email very quickly and skip vital parts. I know this for a fact based on stupid responses I've gotten.

"Please note" never implies it should actually be noted on a piece of paper, just notice it and ready it thoroughly.

"It should be noted" - however, could be interpreted as you want the recipient to actually write it on a note.

Edit: And yes, Tesla's use of "It is important to note" kind of falls in the gray zone for me... but I've never even considered this an issue

Not everything that is important enough to write down has the same importance. While the phrase may be over-used, it is merely one in a long range of methods we use to draw more attention to statements we believe have particular importance at a specific point.

If the statement is not genuinely something that needs particular attention, then I agree that such phrases should be avoided.

However Wikipedia's policies in are not particularly relevant here, as in the case of an encyclopaedia which pieces of information are important depends entirely on what the reader is looking for at any given moment - the purpose is to make the information available, not to direct the reader to pay special attention to particular bits of information.

In many other contexts, the purpose of the text is not to merely make information available, but to influence the reader, and in many cases to specifically draw their attention to specific pieces of information. E.g. consider safety instructions or documentation of something where there are serious caveats.

In those situations various forms of emphasis are important, and using specific phrases to convey that a given statement is more important can be one valid way of creating emphasis.

> By putting it into words, you have said "this is important enough to actually say it."

But you have not said this is important enough for you to read, right now, nor especially important enough to remember every detail. If everything written were important to read, we'd spend our lifetimes failing to work through the backlog - of every word ever written since the dawn of writing. There are key points to remember, and then supporting facts to be forgotten after they've done their job of arguing for those key points, and this language emphasizes their key points (in defense of Tesla, naturally, rather than the mundane statistics.)

> Further saying that it is important is a crutch.

Perhaps. But remember - crutches are legitimate medical devices built to serve a need, they have purpose! I tend to use the terser "IMPORTANT:", "NOTE WELL:", "N.B.:"[1], "NOTE:", etc. to call attention to high priority stuff. I'll especially use this style of emphasis in code, where I can't make my text bold, underlined, large-fonted, red, blinking for emphasis, embedded into a modal dialog you can't dismiss for the first 10 seconds, just to make sure you read the damn thing. These are the notes I deem to have e.g. a 100:1 time saved/spent ratio, where I need my fellow programmers to stop skimming as a thousand inane comments before have perhaps trained them to do, and to instead save us all a lot of time and energy by reading the important thing.

But being that terse in a press release like this would come across as incredibly tone deaf and insensitive. Try replacing "It is important to emphasize that" with "NOTE:" and "It is important to note that" with "IMPORTANT:". The paragraphs themselves might read fine, but in the broader context of the press release it'd be cringe worthy.

[1] https://en.wikipedia.org/wiki/Nota_bene

[4] https://www.mtholyoke.edu/acad/intrel/orwell46.htm

Because, Orwell was awesome.

On the other hand, dashing off a quick comment isn't exactly the pinnacle of writing. "It is important to note" is lazy writing. So many of my comments here have the terrible sins of passive voice. I think; in my opinion; just; very; i throw in useless phrases all the time.

Forgive us for writing a long letter, we didn't have the time to write a short one.

> The phrase is just filler and means nothing, yet it is used everywhere

I guess this is due to legal concerns (which also provided us with pages-long email footer legalese bullcrap).

It is important to note that there is some writing that sounds like a college graduate wrote it and some writing that is good.

There is no overlap.

Compare these two versions of the first 2 paragraphs. The second version actually emphasizes the point better, without saying "it is important to emphasize."

Original:

We learned yesterday evening that NHTSA is opening a preliminary evaluation into the performance of Autopilot during a recent fatal crash that occurred in a Model S. This is the first known fatality in just over 130 million miles where Autopilot was activated. Among all vehicles in the US, there is a fatality every 94 million miles. Worldwide, there is a fatality approximately every 60 million miles. It is important to emphasize that the NHTSA action is simply a preliminary evaluation to determine whether the system worked according to expectations.

Following our standard practice, Tesla informed NHTSA about the incident immediately after it occurred. What we know is that the vehicle was on a divided highway with Autopilot engaged when a tractor trailer drove across the highway perpendicular to the Model S. Neither Autopilot nor the driver noticed the white side of the tractor trailer against a brightly lit sky, so the brake was not applied. The high ride height of the trailer combined with its positioning across the road and the extremely rare circumstances of the impact caused the Model S to pass under the trailer, with the bottom of the trailer impacting the windshield of the Model S. Had the Model S impacted the front or rear of the trailer, even at high speed, its advanced crash safety system would likely have prevented serious injury as it has in numerous other similar incidents.

Simple edit to one sentence:

We learned yesterday evening that NHTSA is opening a preliminary evaluation into the performance of Autopilot during a recent fatal crash that occurred in a Model S. This is the first known fatality in just over 130 million miles where Autopilot was activated. Among all vehicles in the US, there is a fatality every 94 million miles. Worldwide, there is a fatality approximately every 60 million miles.

The NHTSA action is simply a preliminary evaluation to determine whether the system worked according to expectations.

Following our standard practice, Tesla informed NHTSA about the incident immediately after it occurred. What we know is that the vehicle was on a divided highway with Autopilot engaged when a tractor trailer drove across the highway perpendicular to the Model S. Neither Autopilot nor the driver noticed the white side of the tractor trailer against a brightly lit sky, so the brake was not applied. The high ride height of the trailer combined with its positioning across the road and the extremely rare circumstances of the impact caused the Model S to pass under the trailer, with the bottom of the trailer impacting the windshield of the Model S. Had the Model S impacted the front or rear of the trailer, even at high speed, its advanced crash safety system would likely have prevented serious injury as it has in numerous other similar incidents.

They don't even have to do a recall; All Teslas can be updated (or in this case, downgraded) over the air. All they need to do is either learn from this and fix the 'white vehicle against a white sky' problem, or remove the 'auto-pilot' feature completely, until they actually develop proper auto-pilot.

Paraphrasing from JFK: We choose to make auto-driving cars, not because it is easy, but because it is hard.

There will be casulties along the way. Deaths are not a reason to stifle development. If we'd done that, we'd never have spaceflight or Vaccines (or Nuclear Weapons for that matter...) .

Tesla actually use the correct term for it, the fundamental problem is a mass misunderstanding of what aircraft autopilot does.

Autopilot in an airliner is capable of flying the plane on a set course in ideal conditions. This means good weather, no other air traffic in the vicinity and full feedback from its various sensors. As soon as the plane enters a scenario outside of which it's designed to operate, control is reverted to the human pilots.

That's exactly how Tesla's system is designed. So long as the computers can account for the current scenario, the car drives itself, but it explicitly falls back on the human (who, like a pilot, must be ready to take command at any time).

A trucker pulling out on a driver and blind-siding them falls outside the definition of 'ideal condition' in my book. I live in Europe where underrun guards are mandatory on all trailers. Sounds like one wasn't fitted here. That's likely more to blame for the fatality than Tesla's software or a distracted driver - let's face it, anyone could suffer a lapse in concentration at a critical moment on the road.

>When it became clear from early youtube videos that people were not using the technology 'correctly' Tesla did nothing more than issuing a statement.

Is personal responsibility something that's completely taboo nowadays?

No, as far as I can read, Teslas don't have LIDARs yet (but they are reported to have them, soon: http://electrek.co/2016/03/18/tesla-model-s-test-vehicle-aut...).

It's a narrow beam radar (about 200m range) which checks the current lane is clear, it probably assumed the other parts of the truck were some obstacle on the side of the road (or in another lane)

I'm not even sure the radar has a height dimension, I suspect they just focused it at bumper height to catch most cars.

The car also has a single camera in the front to see the lane markers, and a short range ultrasonic system to give it a 360 degree view, out to a few meters.

For the amount of confidence drivers are putting into the system, Tesla don't have anywhere enough sensor data.

Musk: "Radar tunes out what looks like an overhead road sign to avoid false braking events". So there's that.

https://twitter.com/elonmusk/status/748625979271045121

This is what I was wondering too. Which is also why I predict (although basically speculating!) that an alert driver would have noticed the actual lorry itself and made the assumption that it carried a trailer, and at the very least slowed down.

But isn't Tesla autopilot working only in highways, which are much safer then average road? Then this statistic wouldn't be that good.

If all the world drove cars whose passive safety was as good as Tesla Model S, number of miles driven per fatality worldwide would be more than 130 million miles, I think. An average car around the globe is quite a junk. I think we should compare the Tesla Autopilot numbers only with non-Autopilot Tesla numbers.

Well, if you read those vehicle statistics as "the richer the people, the less fatalities", then those Tesla Autopilot scores may be worse than you might expect.

If the average is 1 per 90M, it's worth remembering that with just 130M tested, the variance is huge.

I would have found it much more appropriate to first offer condolences before stating numbers about how good their system is performing

According to the numbers there's a 25% probability it would have taken more than 130 million miles before the first fatality, assuming an average of 1 fatality every 94 million miles.

That's ignoring the fact that Tesla's autopilot might only be used in situations with a relatively low average number of fatalities per million miles.

It wouldn't necessarily be cause to disengage the autopilot. Plenty of things could cause negative acceleration besides a collision. A hill, a patch of mud, etc. And presumably they didn't anticipate this kind of situation, so didn't include moderate slowing due to collision as a code path.

You can't stop momentum like that on a dime.

The dupe detector needs to be improved so that submissions cannot be posted twice. It works when I try to submit something which just has been posted the same day, but there must be some logic that allows it to be reposted and also the URL check isn't sufficient. Alternatively a merge of posts by moderators would be okay, too.

There's already 400+ comments discussion, so mods tagged this one as dupe.

tagging duplicates, I assume