I can play the conjecture game too.

The average Tesla driver is probably a worse driver than the average gas-powered car driver. Why? Because they have the disposable income to purchase a Tesla, which means they're probably less careful with their possessions (as they can presumably afford to replace them), plus the Model S is a sports car so they're probably driving faster and more carefree than your average person.

Not that I really believe this, but come on, your argument is completely made up.

Musk is speaking definitively, and has literally a billion dollars at stake with this post.

My point wasn't necessarily that I'm right -- much like you, I'm just arguing that the logic isn't as simple as it's made out to be.

So if I drove that 100 mile/workday commute for 4,000 years, something like this might happen approximately once.

Incidentally, I have seen, during said commute, the remains of an ordinary car that caught on fire. There was nothing left but a burnt skeleton of a car that appeared to be made of ashes. I'd rather be in the Tesla while it was still on fire than the burnt wreck I saw.

But risks must be compared to the relevant alternatives. Certainly, those who arrange their life such that they do not drive at all are removing a significant risk, as I can personally attest to having encountered several drivers who were, frankly, suicidal. I speak specifically of the person who pulled into a turning lane to pass (when I was already in it, stopped), and gunned it driving straight at me. By no means should you think that I am diminishing the very real risk we engage in every day while driving.

But most people measure risk by how easily they can imagine a scenario. As such, I offer this for you to contrast with the photos in the article: https://www.google.com/search?q=burnt+out+car&tbm=isch Such things are not uncommon and I have, personally, witnessed such. While measuring risk according to one's memory is a useful cognitive shortcut in many situations, but with the way our news sensationalizes things by highlighting unusual events and under-reporting common ones, it is not a useful metric.

Just like Elon did it, it sounds nice but doesnt make a lot of sense.

Note that there have been a lot of fires in those 3 trillion miles driven by ordinary cars (not just 1), as even a quick search shows: https://www.google.com/search?q=burnt+out+car&tbm=isch

On an individual level, I am far more worried about identifying and avoiding the disturbingly large number of idiotically suicidal drivers I have met. And I mean that very seriously: as soon as I identify someone driving erratically, I find some way to get away from them, whether it means forcing them to pass me, taking an unplanned exit, or what have you, I will make sure I put distance between us.

But I remain unconcerned about a single report of a small car fire when there are many worse ones every single year (if not every day...)

I do use the Prius in the city though, I easily get 75+ mpg when supermiling.

[0] http://www.caranddriver.com/features/how-to-adjust-your-mirr...

I love my 1 year old Son. That makes me more terrified to drop him.

Do not assume just because someone is able to get enough money to buy something expensive,that they have tons of money. By that same logic, anyone who is able to buy a home is clearly flush with cash and able to throw it all around. Completely ridiculous.

As for the house argument, you and I both know that the two are totally incomparable. Putting a roof on your head is a much bigger priority than buying an electric supercar. People need to pay for shelter. They generally can't afford to pay a lot for luxury cars. If you argument is true, why isn't everyone buying $63,000 cars? (Not to mention the millions of people who can't even afford homes.)

Your posts demonstrates everything that is wrong with class stratification in America. I don't think you or your friends who elected to choose Teslas over Porches know what it means to be "average."

That means this SF bay area laborer must give up a Saturday or Sunday afternoon every week, but it's hardly impossible "to afford" or a "massive tradeoff in quality of life." On the other hand, it's challenging to haul a few thousand pounds of mulch in a Tesla, I expect.

Oh, and the $600/mo assumes a $7100 down payment, which needless to say is far more than the average down payment on a car and is really more like what the average american puts down on a HOUSE. In other words: c'mon.

Not to mention the opportunity cost of that money. If you're making $25/hr in SF you do not have a high standard of living, especially when compared to your neighbors which, studies have shown, is exactly how we measure our own financial happiness.

Though you're right about the down payment, so let's assume it will be a used Tesla and therefore cheaper. And you're right about the relative comparison point. But these were rough numbers, remember, and not intended to be a detailed argument as much as a refutation of the impossible "to afford" claim above.

I'm really confused by "average, middleish class people who normally would have bought a porsche or a tesla" - is that a typo? Did you mean "would not normally"? If not, your idea of "average middleish class people" is really oddly out of whack.

Many people by homes because of a calculation that owning a home is cheaper in the long run than paying rent over that same period of time. Many people buy homes not based on such a calculation and simply because they have disposable income and owning a home is nice. Many people also buy homes based on neither financial decision-making nor disposable income, and I seem to recall this having some negative impacts on our economy a few years back, which may make it a poor analogy to why it makes sense to buy a Tesla (or a porsche?!) without disposable income.

Then go look up the real numbers.

I suspect you're in for a surprise.

You're either being disingenuous or your perception of reality and normality is really skewed.

You'll probably see the posts saying this to you as condescending, but if you can even think of purchasing a Model S you're in a very exclusive group of people.

When one tiny sliver of society (web engineers) simultaneously wields so much influence (as web companies do) and is so cut off from the reality of America (as you apparently are), it doesn't bode well.

Outside of a couple critical mass areas in the US -- I'm thinking SV, greater Boston, NYC (just because everyone there is in their own little bubble ;) ) -- it's just another job.

And if you're not in the US, I wouldn't even know where this sort of SV halo exists.

I grew up in a ~30k household in the 90's. Something I would call lower middle class. Upper middle class is probably ~70k-100k.

$100k does not go very far in San Francisco and Manhattan. Nobody making $100k in those places could reasonably afford a Tesla without severe over-spending (which they're welcome to do if they want to--whatever floats your boat). If you view the middle class as owning detached single family homes with yards, you would be lower middle class at best in those cities at $100k. If your spouse also made $100k, you could afford the crappiest single family home that was still livable.

But it's a really good salary in Denver or Houston; you could definitely find a nice house to buy and afford comfortably, though it might have a bit of a commute. If your spouse also made $100k, you could afford a nice home in any neighborhood with no commute.

And it might nearly qualify you as wealthy in a small rural town of 15k people total, where you would never need to spend more than $150k on any house.

You know what else? Travel. Domestic and international travel is a LOT more affordable when you're making $100k in SF.

And a big one: Retirement. You sock away your SF salary for 40 years and then retire in a cheaper COL area with no state income tax.

The truth is, even here, $100k is a lot of money. It really is. It's not rich by any stretch, but it's certainly not average middle class living. It is frustrating that a couple making over $200k can hardly buy 1000sq ft house while staying inside 30% (of gross income) housing budget, a metric widely seen as "affordable housing." But still, on balance, that couple making $200k is doing very, very well.

And the truth is, there are a lot of engineers here that, with liquidity events (which might just mean vesting your RSU's in an already public company), are earning $200k a year. That literally puts you in the top 1% of wage earners. (But not in THE "1%" of course which is usually a term referring to net worth and not gross income. But still.)

Edit: The math is different of course if/when you add children to the mix. But I deliberately left them out, because in reality, it's just not a common lifestyle choice in San Francisco.

Small correction -- state income tax in CA is a big deal even if you're saving for retirement. And once you retire, you presumably won't care what the state income tax is.

Retirement savings goes into tax advantaged accounts like 401ks and iras. This is pre tax income. You pay the taxes at the end, when you're drawing from your accounts during retirement.

But for those people, taxes are immediate rather than deferred.

Actually, Roth IRAs are limited to $5000 a year per person, and once your income reaches ~$110k you can't deposit into a Roth at all. The vast majority of private retirement dollars are saved in 401k accounts that have have a cap much higher than in a Roth. Three times higher.

And actually, if you do the math (or just go read the math) you'll see that there is very little difference in your total retirement funds whether you use a pre-tax or post-tax account. This is because in a pre tax account you have the advantage of earning capital gains on the IRS's dime. The wisest choice is of course to use both if you can. Though many many people living in this area, due to income restrictions, cannot.

Anyway, I intended my reply to you as just honestly informative. I'm happy that you now are informed, but there was no reason to say "uhh, i mean a roth. yeah. a roth."

My reply was in the context of the $100k earner mentioned earlier, who would not be restricted from contributing to a Roth. Our hypothetical young tech worker is going to expect his or her real salary (and tax rate) to go nowhere but up as the years roll on. It's not bizarre to imagine this person might choose Roth rather than traditional for individual savings, independent of any 401k they might be contributing to.

There is only one way to read your comment: "Small correction -- state income tax in CA is a big deal even if you're saving for retirement. And once you retire, you presumably won't care what the state income tax is."

You didn't qualify that! You added a "small correction" that was anything but. And yes, you got me, I didn't realize that for 2013 (possibly 2012?) the Roth limit was increased to $5500. Still not much stacked against the $24,000 I saved in my tax deferred accounts last year.

Listen man, i'm not one for silly debates on the internet, so go ahead and have the last word. But If you're going to add a "small correction" that is not really a correction, you should expect some disagreement.

Perhaps what you really meant was "That's true, unless you chose a Roth IRA as your only form of retirement savings. In that case, current tax rates matter and future tax rates don't."

That still glosses over the fact that I'm talking generally and you're talking about a specific, somewhat rare scenario. But at least it's accurate. As it was, had I not replied to you, a casual reader would've read my post, then yours, and walked away assuming that retirement savings is taxed at your current tax rate, in your current state of residence, when in reality it's just not so.

And, yes, of course I'm aware that a Model S is the same price no matter where you buy it, to a first approximation. The point is that housing is so expensive in the Bay Area that you can actually end up having less disposable income left over for your car than someone making $20k less than you on paper in a much cheaper city and in a state with lower (or no) income taxes (Washington, Texas, maybe a few others?). If you don't believe me, you need to just sit down with a calculator that will do taxes and everything, look up comparable rents, and do the math. You sound like you might be surprised.

Consumer goods like cars and laptops and TVs will cost the same, to a first approximation, anywhere in the continental U.S. But in some coastal areas, you have real estate costs that are 20x that of cheaper areas with salaries/compensation packages that are generally not 20x higher. In central Pennsylvania towns the median household income, for instance, tends to be around $26K and the median house cost is $64K, meaning houses are 2.5X median income.

In Palo Alto, on the other hand, the median house cost is around $2.25M, and the median family income is around $160K, meaning houses are 13.75X median income. Part of this are older PA families who bought when land was cheap many decades ago and are grandfathered in with lower tax rates, but part of this is also people who are stretching their finances dangerously to be able to afford there. (Even though those $2.25M median homes are often smaller than the national average, measured by square footage.)

It's possible to make $100K in the SF bay area and be, in real terms, poorer than someone making half that in central Pennsylvania.

Population of San Francisco: 0.8 Million.

Population of Manhattan: 1.6 Million.

Population of London: 8.1 million.

Percentage of the world's population living in either San Francisco, Manhattan or London: ~0.0015%

Exactly what do you think an outlier is? If you live in one of those three urban areas, you're already a member of an elite class.

But I don't really understand why living in an expensive urban area automatically puts one in an elite class. The standard of living is not much better than other places for one thing--what you gain in diversity is lost in housing conditions. Also, it's not like $100k wages there are super rare. You don't need to be a 4.0 Ivy League graduate to score a livable wage in San Francisco. That's the starting salary for a lot of professions in cities like that, not just software developers. Police officers in the SFPD can make as much as a Twitter engineer fresh out of Stanford: http://www.sf-police.org/index.aspx?page=1655

Edit: I can't edit my prior comment. Time window?

I'm not crazy about percent notation. I find it confusing. I might have written it thusly:

> 0.0015 of Earth's population are living in San Francisco, Manhattan or London

I wouldn't buy a Tesla on $100k in North Dakota either. Maybe, if I felt like blowing through money, I'd get a BMW 3 series or Audi A4, both of which are still considerably cheaper than a Model S.

I was just pointing out that besides housing (especially owning) the cost of almost everything in CA is the same if you move somewhere where the cost of living is lower yet the salaries elsewhere are significantly less. So there is an opportunity to save a lot of money if you are willing to compromise somewhat on your living conditions.

I suppose it depends on whether you think it's important to own your home and/or you mind having roommates. When I first moved to CA I had severe sticker shock re: the rents so I rented a room in nice houses with other young professionals for the first 5 or so years I lived here.

If you can afford to spend the average household's entire annual income on a car coolness upgrade, then yes, you have disposable income to throw around, and no, you're not middle class.

Or sense. Or street smarts. That was the OP's point.

That said, I'm not really sure why this is such big news. Accident happened due to debris on the road. No one was hurt. The car was severely damaged, but got the owner to safety.

AND THEN a small fire happened...

Then I read things like this: http://www.dailynews.com/general-news/20130928/5-killed-1-in...

Let's consider this for a second - Tesla Model S had a no-injury accident and made national news. 5 kids burned alive in a Sentra, and this barely stayed on local news for 3 days.

Tesla Model S is a phenomenal vehicle with an excellent, if short-ish, track record. It really annoys me that reporters go out of their way for sensationalist articles, when there are far worse tragedies to cover.

What does that mean? Taking a common definition of "marketing" to be "expressing the value of a product to potential customers," any article that is about safety or about a recent product incident is also marketing. It sounds like you're using "marketing" as an ill-defined term that basically just means "something bad about the way a company publishes information."

This is not a dry facts page, but a public relations communication that is intended to sell you on the fact that Tesla Model S is a safe vehicle. Nothing wrong with that and nothing "bad" as you seemed to imply.

"The nationwide driving statistics make this very clear: there are 150,000 car fires per year according to the National Fire Protection Association, and Americans drive about 3 trillion miles per year according to the Department of Transportation. That equates to 1 vehicle fire for every 20 million miles driven, compared to 1 fire in over 100 million miles for Tesla. This means you are 5 times more likely to experience a fire in a conventional gasoline car than a Tesla!"

Facts stated: * 150,000 car fires per year according to the National Fire Protection Association. * Americans drive about 3 trillion miles per year according to the Department of Transportation. * 1 fire in over 100 million miles for Tesla.

Sales pitch at the end: "This means you are 5 times more likely to experience a fire in a conventional gasoline car than a Tesla!".

The reason I call this a sales pitch or a marketing messaging, is that while the facts technically support this, as pointed out earlier, it's based on a data set of 1 occurrence.

Personally, I buy Musk's argument and if I was in the market for a new car, figuring out how to pay for Tesla would actually be at the top of the list for me. This story only re-enforces my own personal opinion that these are great cars.

Particularly for people such as Musk, who are doing so much to advance the expectations of consumers and raise the bar for what is possible with new technology (and here I'm speaking as an un-abashed Musk supporter), I can't help but support his rigorous defense of his projects. Whether it's the NYT "review" that attempted to slam the Model S mile range or the astonishingly good crash ratings for the same car, I think it serves us all to let Musk use his soapbox to be an advocate for his work at Tesla, SpaceX, etc.

The data on 100 million is probably enough to compare with the 3 trillion miles. The populations don't need to be equal to compare them, just big enough that they are random and distributed enough.

However, there was one fire over 100 million miles. The problem isn't the 100 million, it's the 1 fire. This wasn't a controlled experiment, either -- they just stopped the clock as soon as the first fire happened, and multiplied. A week ago, they could have said "You have exactly a 0% chance of your Tesla catching on fire" and have been right by this logic.

To think of it another way -- let's say you get lucky and get a hole-in-one your 10th time golfing. Does that mean you'll have 10 hole-in-ones if you golf 100 times? Doubtful.

EDIT: Also, don't forget that Elon is mixing numbers. There's no fire if someone doesn't run over something. All these numbers show is that the average driver is more likely to run over something. Of course Tesla drivers run over fewer things -- there are no 16 year old kids texting while driving a Tesla... yet.

For the purpose of our simple modeling, suppose that there is a constant risk per mile of the car catching fire, making an exponential model reasonable. Under this model, observing the first fire at 100 million miles would give a 95% confidence bound on the rate of fires of about one fire every 33 million miles.

If we're comfortable with the stated rate of about one fire every 20 million miles for other cars, then this would give a 95% confidence upper bound on the Tesla's rate of fires at about 60% of a normal car's rate. This isn't the 20% that Elon's statement would imply, but it does suggest a difference (which could just be due to other problems with the comparison).

That said, I also think the 1 fire is the problem here. Just think about how that relation changes with 2 fires.

If I had the cash I would still purchase a Tesla after the second fire as well.

But since the number of cars is increasing, it's not a Poisson distribution; if the chance per car per time is constant, you'd expect the time to the next fire to be shorter.

The median car in the US is ~11.6 years old[1], while Tesla's oldest vehicles were released in 2008, and the vast majority of their fleet was sold in the last couple years.

Obviously, the massive differences between electric powered cars and internal combustions engines means that they may never reach parity, but unless Musk has figured out a way to beat entropy, its pretty safe to assume that older cars will break down/suffer leaks/explode more than newer ones.

[1]https://www.polk.com/company/news/polk_finds_average_age_of_...

Please don't argue against statistics (mathematical information based on fact) when you don't understand them.

The Law of Large Numbers states that as more miles are traveled, the fires per mile will approach the expected value. It is entirely possible to have 10 fires in the next week.

We won't know what the expected fires/mile is until a much larger sample is collected. It will take years to prove out.

Here's a question (for anyone in this thread arguing statistics) that has an actual numerical answer: given the information in the article, what is the probability that Tesla's indeed experience less fires per mile than other cars? If someone doesn't know how to calculate the answer to that question, he shouldn't be arguing here.

This seems a bit dodgy since I'm "designing the experiment" after the fact, but I'm not sure how to correct for that. Any Bayesian experts?

You need an exhauseted state space. You cannot empirically infer a legitimate probabliliy, eg n/100m miles) with only a single failure observation, if there are 100 possible ways to fail. At best you have data on (1) of (N) ways to fail, but surely in the case of car accidents N=large.

A total of 2,650 cars were delivered to retail customers in North America during 2012, 4,900 during the first quarter of 2013, and 5,150 during the second quarter of 2013

Assuming 13000 cars on the road, each car would have logged 9k miles to get 110m road miles, as quoted by Tesla. But we know from past industry experience, that road fires are proportionate also with fleet age.

So, if anything we the probability of a road fire is likely to go up as more failure modes are discovered (including by chance), and as the vehicles cycle through a normal working life.

Our data is the fact that we went 100 million miles before a fire, after which exactly one fire happened, so we want to find the distribution `P(a|t = 100 million)` which tell us everything we want to know about `a`.

Then use Bayes' theorem: `P(a|t) = P(t|a) P(a) / P(t) = aw exp(-a(t+w)) / P(t)`. The normalization factor `P(t)` involves an integral over `P(t|a) P(a) da` from 0 to ∞, which wolfram alpha tells me evaluates as w / (t+w)^2.

So our posterior probability is `P(a|t) = a (w+t)^2 exp(-a(t+w))`, but we can take the limit `w -> 0` at this point for a fully uninformative prior: `P(a|t) = a t^2 exp(-at)`.

So we can just set `t=100e6 miles`, and now calculate things like the expectation of the distribution: `E[a] = 2/t = 2e-8 per mile`. Or the probability that the hazard rate is less than other cars, which is the integral from 0 to 1/(20 million miles): `P(a < b) = 1 - exp(-bt) (bt + 1) = 0.96`.

This wasn't after driving 100 miles. Aren't you off by 6 orders of magnitude?

http://what-if.xkcd.com/65/

By all accounts it comes down to the old bayesian/frequentist battle lines.

Of course, I'm exaggerating in the other direction. What we really should be calculating is the odds that Teslas burst into flames less often than the average car, given that the average car does so every 20 million miles and the first such event in a Tesla was at the 100 million-mile mark. We're still failing to account for the fact that the average Tesla is newer and probably better-kept than the average car, but it would at least be a reasonable start.

I don't know enough statistics to perform this calculation, but I would like to see how it is done.

For an exaggerated example, if Tesla had driven a billion miles and had 0 fires, you shouldn't say that there's not enough data - you definitely would have enough data to say that the chance of fire is below the gas-car rate of 5 fires per 100 million miles.

I'm honestly curious how one models this type of thing statistically, and I am not convinced enough of its obviousness to just accept numbers that someone throws around.

You can sample less than 2000 people and get 99% accuracy with a 3% margin of error for a population of 325 million. Increasing the sample size to 10k simply reduces the margin of error to 1.29%, hardly worth the extra sampling of 8k people.

Which is not true here, as the GP correctly notes.

[1] Assuming that you're not measuring average age or measuring 'who is at home', but if you want to see, say, the average political opinion of total USA population, which tends to correlate with age.

Furthermore, how do you gather the data needed to correct the polling numbers without being able to accurately poll people in the first place? Seems like a complete chicken-and-egg problem.

Of course, the refrain I often hear is that as long as you pick the RIGHT 1,000 Americans, it's as good as polling all 319 million.

If it's a proper random sample, then it's far better than sampling all 319 million because it's 95% accurate with about a 3% margin of error and vastly cheaper and actually practical; you can't poll 319 million people.

Polling 10,000 Americans would be vast overkill, in any case.

If you read the reports from the National Fire Protection Association they do estimate the number of "highway vehicle fires" at around 150 000 per year [1], but they also explicitly say that 'the term “highway vehicle fires” is used to describe the type of vehicle, not the location of the fire' [2].

If you look at fires caused by collision or overturn (which I guess this would be classified as) that's only 4% of the total, or 6000 per year.

1. https://www.nfpa.org/safety-information/for-consumers/vehicl...

2. https://www.nfpa.org/research/statistical-reports/vehicles/v...

Yes, this fire was caused by a collision, but fires caused by other things are relevant. If a Tesla car caught fire in somebody's driveway, people would be concerned about that too, right? So comparing 150,000 highway vehicle fires to 1 Tesla fire, regardless of cause, is pretty reasonable.

And finally, if you've bought an expensive car like the Tesla, you may be more likely to drive it more carefully than the average person does their Honda Accord.

I don't think Musk is any more dishonest than any other CEO, and he's probably more honest, by far, on average. However, it kind of pains me to see how easy it is for him to sway the hacker crowd with data-interpretations that would be questionable by any standard. If this is among the strongest empirical evidence he can provide, then I think we should maintain some skepticism.

http://www.bankrate.com/financing/cars/study-bmw-drivers-are...

For whatever reason BMW stood out most of all as the worst of the worst in this study, but 'expensive car drivers are better drivers' has been proven to not be an assertion you can make from common sense or intuition.

1. We can't assume that Tesla drivers are the same as traditional luxury car drivers. While cars are a statement of wealth, and many arrogant bastards like making that statement...A Tesla is ostensibly a statement of a few other things, such as concern for the environment, optimism about technology, etc. etc...in the same way, even though many programmers make high-end salaries (low-6 figures), I wouldn't say that we can expect these programmers to golf as much as the average non-programmer low-6-figure-earner.

2. Reading the link you posted, it said the study evaluated based on the frequency of accidents and the size of payout...I couldn't find the study on the website, but this statement from the article was ambiguous to me:

> According to data from IIHS website, the collision insurance losses on BMW 7 Series are more than twice the average for vehicles nationwide, and BMW 3 Series two-doors are more than three times.

Twice the average of what...? Payouts total? Payouts per claim? Payouts per claim normalized by number of total cars of that kind on the road?

3. The NHTS data concerns the number of fires. For the study you cited to be more relevant, we'd have to know what proportion of the insurance claims also involve vehicle fires. If you're thinking, "Well, if a car caught on fire, then that's likely because of an accident...so more accidents mean more fires"...well, a> That's a classic logical fallacy (All dogs are animals. Sara owns a lot of animals. Therefore, Sara owns a lot of dogs) and b> the kinds of accidents that make up a majority of insurance claims (my guess is low-to-medium speed fender benders) may not represent a significant portion of incidents that end up in a flaming wreck.

All in all, I wouldn't say that Musk is clearly wrong here...He could be right, we just don't have enough information either way...so it feels slightly dirty -- a slap in the face of analytical thinking -- for him to cite these statistics as being anywhere near conclusive.

I firmly believe that what's happening on the stockmarket and the fire are merely a coincidence, there isn't a causal relation between them. Tesla stock has been due to correct back for a while, and a downgrade came in as well (unrelated to the fire).

Edit: first rule of the HN club: only positive things about Tesla, otherwise you are going to be voted down. Neutral opinion is considered negative. :)

If the battery really was punctured by an unfortunately placed piece of metal in a sort of surprising scenario, we ought to stratify by collision type in order to really understand the problem. How do gas cars perform during a gas tank puncture? At similar speeds? And how common are such collisions? The relevance of this consideration is expressed by Simpson's paradox:

http://en.wikipedia.org/wiki/Simpson's_paradox

I.e. analyses of performance when conditions differ ought to take into account those conditions, at least when they are both a: known (true in this case) and b: not causally connected to the variable under consideration (also true in this case).

A good point. It would be interesting to compare the miles drive on "brand new expensive cars" not just all cars which include obviously older and/or poorly maintained vehicles.

Agreed, there's insufficient data for meaningful statistics, but his main points were amour-plating, firewalls and that batteries are less fire-prone than "highly flammable liquid".

The 30,000x larger sample in the aggregate means the calculated fire rate for non-Tesla cars is probably a more reliable number. Maybe those 100M Tesla miles were "lucky"? Possibly not, though - I don't know enough about cars to guess.

The 3 trillion aggregate miles includes a lot broader sample size of car age than those from a single company that's 10 years old. This, to me, seems important to control for.

Finally - I bet the usage profile of a Tesla is different from the median usage profile of every random car on the street and throughout the country.