Comment by 001sky | Hacker News Reader

Comment by 001sky | Hacker News Reader

Parent Full thread

Reading the headlines, it is therefore easy to assume that the Tesla Model S and perhaps electric cars in general have a greater propensity to catch fire than gasoline cars when nothing could be further from the truth.

Actually, the Data do not support Him. He is starting to tread on shallow ground, and perhaps should re-evaluate and stratify his data analysis to fires 'on public highways'.[1,2] That is where all of the Tesla Fires are happening. Unfortunately, those data do not support him.

By this metric, you are more than four and a half times more likely to experience a fire in a gasoline car than a Model S! Considering the odds in the absolute, you are more likely to be struck by lightning in your lifetime than experience even a non-injurious fire in a Tesla.

And, Bayesian statistics might give you a different analysis.

From a PR perspective, I wonder if this is really doing him any favours? The note he wrote after the october crash was flawed, but perhaps excusable. This is too pre-meditated and now must just be considered somewhere on the spectrum of 'wrong' to 'misguided' to 'misleading'.

[1] In data terms, it excludes intentional fires like vandalism, and parked cars being struck, and accidents in motor repair facilities, etc.

[2] http://legal-dictionary.thefreedictionary.com/Public+highway

So where are the data for gasoline car fires on public highways? You're accusing someone of not being supported by their data, and offering absolutely no data in counterpoint.

see, for example: https://news.ycombinator.com/item?id=6691541

You have pointed to another HN comment in which you make assertions about car fires without providing any actual evidence.

I've had a look and entirely failed to find any evidence supporting your claim that 80% of car fires occur while the car is not being driven on roads.

It does appear that most car fires are not the result of crashes or collisions. I don't see that that's relevant. If I were deciding between two cars on the basis of fire risk, the question I'd care about is "how likely is this car to catch fire and injure me?", not "how likely is this car to catch fire and injure me as a result of a collision?".

What does look fishy to me about Musk's argument here is that most vehicle fires seem to be the result of mechanical or electrical faults, which become more likely with age as things wear out and break and corrode. All Tesla cars currently on the roads are rather new, so comparing them against the population of cars at large could be quite misleading.

[EDITED to add: aha, following a longer chain of links from your earlier comment I found some statistics. The terms used in the NFPA report in question are ambiguous, but what's clear is that the statistics you yourself cited plainly refute your 80% figure. Here are the figures: 22% "residential street, road, or residential driveway"; 18% "vehicle parking area"; 17% "highway or divided highway"; 8% "unclassified street"; 6% "street or road in commercial area"; 5% "mercantile or office"; 4% "open land, beach or campsite". So, at a bare minimum, the lowest fraction of fires that could be on roads would be 17% + 8% +% 6% = 31%, and that assumes that none of the "residential street, road, or residential highway" and "mercantile or office" figures are for cars being driven on the road -- which seems monstrously improbable.]

The data comes from Figure 1 of the Report that Tesla cited in October. I provided the above merely as an example of another way to read the report, but feel free to do your own analysis. Clearly, Telsa fires were 100% on public highways (by any definition). To which the report reads:

Where do automobile fires occur? Roughly three-quarters (71%) of automobile fires and associated deaths (76%) occurred on highways, streets or parking areas. While only 17% of the fires occurred on highways or divided highways...

So, taken strictly the analyis seems apples to apples. In any event there is clearly some adjustment that needs to occur rather than a blind-usage, IMHO. Whether the haircut is 33% at a minimum or a higher amount.

[PS I prvly addressed the :Edit: points in another comment below> https://news.ycombinator.com/item?id=6760562]

The loss he cannot recoup is this, the average person out there would never have associated CAR FIRE with an electric car. Hence, you take a propulsion technology which people know even less about than they think they know about their gasoline car and their fears can take flight.

Look, they made a compromise on pack layout to facilitate swappable packs.

Lastly, Must should not be comparing his car to gasoline only cars, how many Leaf or Volts have caught fire after a collision? I do recall a Volt doing it once, days after in a junk yard. So Elon, why doesn't the Volt or Leaf have this issue? Is there battery chemistry less volatile? How about their pack designs?

Exactly this.

I don't understand why people can be so hypocritical and do the electric car cause a huge disservice by this blind fanaticism. Gasoline cars have had decades of experience with countless recalls to perfect certain processes and implementations. Why are people expecting electric cars to not go through the same ? Are people really expecting Tesla to be the pinnacle of perfection on their first few tries ?

People are far more accommodating when you say "hey we aren't perfect but it IS safer than gasoline cars" rather than this "we are perfect and if you disagree with us you are wrong" attitude.

If you're going to claim the numbers don't support the author's thesis, it would be helpful to provide accurate figures (or at least link to an article that does so).

Also: (from the NFPA data) "Highway vehicles include any vehicle designed to operate normally on highways, such as automobiles, motorcycles, buses, trucks, and trailers"

So in fact this could include electric cars as well.

I looked at your prior posts and was unconvinced. From figure 1, 17% occur on highway or divided highways. However, most of the remaining categories are also types of roads.

Happy for peer review. The important point is for people to read the data and use that estimate what they perceive as the risk. At a minimum, the 33% look to be easily excluded (parking 18, residential property 6, mercantile/office property 5, open land 4). Unlcasified street is also 8%, which if it were housing association or forest roads, would be an exclusion bringing you up to 41%. Alternatively, one could just simply tighten the comps to "55mph highway", and use the 17% number as the most relevant subset. However one wants to do it, the increase in fires form (1) to (3) and the reduction of the comparison space by 1/3 to 4/5 has a pretty material change in the implied probability of a fire/event.

These would be the changes vs how they appeared in October after the first PR:

Observed Events (1)/Index Population%(100)

October 1/100=1x Baseline

November 3/100=3x

Nov Min corrected data 3/67=4.5x

Nov Med corrected data 3/59=5.1x

Nov comps only 3/17=17.6x

Why do you want to limit by location when the the data is already separated by cause? All of the Tesla fires were caused by collision.

If you compare (Fires caused by collision)/Vehicles on the road, then I agree Tesla does not compare favorably. In principle, you should compare Fires/collision in case Tesla drivers have an abnormally high collision rate. And statistically speaking, it's pretty hard to draw a conclusion when your sample size is 3 fires.

But the fact is there are still no fires attributed to electrical or mechanical problems.

And the points on both sides about injuries/deaths from fires are pretty much red herrings since the numbers are already extremely low. There are only 2 deaths per 100 fires caused by collisions.

I'm also not convinced that selecting the data based on where Tesla's three fires occurred is the right why to analyze the statistics. Nor is it clear that the fires occuring in parking lots and similar areas involved solely stationary and unoccupied vehicles. In fact, all categories caused deaths at a rate similar to or greater than the fire rate except for one category, parking lots (which still caused 6% of deaths). Perhaps these fires involve someone driving too fast through a parking lot?

The fact that Tesla was not willing to trust the Insurance companies to price insurace going forward on the same basis should be a clue that there are people revisiting their priors. And I doubt they are reading PR pieces, they may even have less ambiguous data than what we looked at here. In any event its clear tha the range from the 'rosy' projection to the 'strict' comparison is an order of magnitude in variation. It seems implausible to argue that there is 'no material change' in risk from these new data. Regardless if the issue is just a design one (and nothing really inherent with EV vs ICE).

Your ideas are interesting. Unfortunately you've exhausted my interest in the topic. But, I still think you overstate the absolute certainty of your interpretation. I do agree that Elon is presenting the rosier interpretation of the data.

People need to give Tesla and Musk a break. It's big auto and big oil that want to see him fail, and they've got enough politicians and media in their pockets to have a fake world's fair with them. They want a slow, controlled transition to the next world energy source, and they're not ready for today's Henry Ford to come along and screw up their plan.

Also, when have safety issues not been a part of disruptive transport technology? For example:

http://www.popularmechanics.com/cars/news/vintage-speed/13-m...

We need to be less influenced by this sort of FUD if we are going to have a company that can actually compete with big auto and big oil. If Musk gave my family 4 Model S's, I'd feel safe letting each of my family members drive them, once a few were old enough.

Reply on news.ycombinator.com