There is quite a bit wrong with your comment. The odds don’t change whether you give them in km or billion kms. However “the odds of dying on each of these” is wrong: Those are not the odds.
As you wrote what you wrote it would mean that only 2 people in the whole US population of 300 million would die on a car.
(annualized) death rate was 1.66 per 10,000 vehicles
The 17times more likely is telling the truth. Of course you could do look at miles consumed per mode of transport, but the point will remain that trains are much more safer (and some people die on them rather by old age, than accidents).
In addition the way you present the numbers with leading zeroes means you have no academic experience in the field of data presentation. Which shows.
1 in 1000 is not really unlikely if it is in regards to your life ending.
However even taking your number it would lead to devastating pictures.
2 married people with 4 grandparents 1 aunt 3 friends and 3 kids means every year 22±5 out of a small 1000 family neighbourhood will be affected by car deaths.
Assuming that a relevant time period is from the birth of a child till it is 30 and therefore might have had a child of their own, so 30 years we get that around 66%±10% of families will be affected. Instead of only 1-3 families during that time.
You did not lift the veil of ignorance, you created a new veil of diffusion.
It would mean that two out of three families would lose a close person within a 30 year generation due to cars, instead of only a small percentage. This is the power of the 17 times!
The alternative of train rides would mean that within a generation virtually no family is affected by car deaths.