6 ms·

Interesting parallel (I know the events are not equivalent, but still): Here's a report of a Mercedes crashed at Nurburgring by a test driver: http://www.autom

Interesting parallel (I know the events are not equivalent, but still): Here's a report of a Mercedes crashed at Nurburgring by a test driver: http://www.automotto.com/2014-mercedes-benz-sls-amg-black-series-wrecked-nurburgring-test.html http://www.automotto.com/2014-mercedes-benz-sls-amg-black-se... Both the test driver and the passenger walked away unharmed. Do note that this was on the track, where the cars are supposedly driven at much higher speeds than on regular roads.

mikeash 13y ago
People are making a typical statistical error here. It is unlikely for a car to catch fire in an accident, certainly. But it's wrong to take that and conclude that this fire was therefore likely due to foul play. Foul play is still a much less likely explanation for the facts as currently known.
- fnordfnordfnord 13y ago
  When would statistics support a suspicion of foul play? The scarcity of some occurrence isn't proof against it.
  - mikeash 13y ago
    Figure out a prior probability of foul play, figure out the probability of a car bursting into flames on impact, plug the result into Bayes' theorem, and you get the relative likelihoods. No, scarcity isn't proof against, but it is evidence. A lot of people are talking nonstop about the rarity of cars bursting into flames as evidence against that interpretation, while ignoring the even greater rarity of journalists being murdered by the US government.
- coldtea 13y ago
  That's only the case if in your world theory (and based on the input you have) you trust the government to not do such stuff. If an investigative journalist "mysteriously vanished" in a Latin American dictatorship, for example, one would not even raise an eyebrow if told it was government work. Even if lots of people also vanished without foul play there.
shawndumas 13y ago
5 point harness and a helmet can really make a big difference.
dredmorbius 13y ago
It's one thing to run off a track onto grass, even if you're flipping and rolling at 180 mph. It's another to make friendly with a tree at a 100 mph lower speed. It's not speed that kills, it's the sudden deceleration at the end.

mikeash 13y ago

People are making a typical statistical error here. It is unlikely for a car to catch fire in an accident, certainly. But it's wrong to take that and conclude that this fire was therefore likely due to foul play. Foul play is still a much less likely explanation for the facts as currently known.

fnordfnordfnord 13y ago
When would statistics support a suspicion of foul play? The scarcity of some occurrence isn't proof against it.
- mikeash 13y ago
  Figure out a prior probability of foul play, figure out the probability of a car bursting into flames on impact, plug the result into Bayes' theorem, and you get the relative likelihoods. No, scarcity isn't proof against, but it is evidence. A lot of people are talking nonstop about the rarity of cars bursting into flames as evidence against that interpretation, while ignoring the even greater rarity of journalists being murdered by the US government.
coldtea 13y ago
That's only the case if in your world theory (and based on the input you have) you trust the government to not do such stuff. If an investigative journalist "mysteriously vanished" in a Latin American dictatorship, for example, one would not even raise an eyebrow if told it was government work. Even if lots of people also vanished without foul play there.

shawndumas 13y ago

5 point harness and a helmet can really make a big difference.

dredmorbius 13y ago

It's one thing to run off a track onto grass, even if you're flipping and rolling at 180 mph. It's another to make friendly with a tree at a 100 mph lower speed. It's not speed that kills, it's the sudden deceleration at the end.