Why Do Google Cars Have Such A Dismal Record Of Rear-End Collisions?

Geez AutoSpies, aren't you supposed to be the "journalists"? Aren't you guys supposed to do your "investigative journalism" and then reveal your conclusions to us?

I hold the opinion that the public does not possess sufficient data to form an informed opinion because so far, Google has not been meticulous at documenting each incident and/or they have not made that information public. Without that information, we're just guessing and arguing about how my guess is right and yours is wrong. We don't even know the number of rear-enders as compared to all incidents, so we can't reasonably conclude that they actually have "Such A Dismal Record Of Rear-End Collisions," let alone speculate as to why that is.

At least that's the way *I* see it.

Out of 12 accidents Google has reported 8 have been rear end collisions (75%).

The national average for get rear ended is about 25%. So Google rear end accident rate is 3 times higher than that.

With that given you can see there is a problem, however it does appear they know how to avoid them, with 2 sideswipes leaving only 2 by frontal collision. That in itself is remarkable.

'009: Do you realize that a small sample size can dramatically skew the data? For example, a single die has 6 sides, and the odds of getting any specific number from 1 to 6 is exactly 1 in 6. However, throw that die 12 times, the same as the number of Google collisions, and you might get the number 3 appearing four times instead of the expects two, and you might not see the number 1 appear at all instead of the expected two.

So what would an observer conclude from this? Well, one observer might (wrongfully) conclude that the die is defective because it shows 3 at twice the normal rate and the number 1 must be completely broken because it never appeared at all. Meanwhile, another observer could (correctly) conclude that with such a small sample size, 12 tosses of a die (or 12 Google car incidents), you can't get any statistically meaningful data.

So are you saying if all of the test cars had zero accidents the reality could be they are death traps waiting to happen? After all it is such a small sample rate.

'009: All we need to do is look to a recent, real-world GM scenario in which a huge sample of vehicles appeared not to be "death traps," and then in a fairly short time, it became widely known that an ignition switch flaw has resulted in close to 100 deaths. Note that only a tiny, tiny percentage of all affected vehicles actually resulted in injury or death.

This just goes to demonstrate that even with a very large sample pool that appears to be perfectly fine, a very dangerous issue can be discovered! A small sample pool would not exempt from this.

The answer is simple...Google has programmed their automated Lexus/Toyota's to behave as actual Toyota/Lexus Owners...

Slamming on the brakes every 30 feet for no apparent reason is a part of Toyota/Lexus ownership...ahhh the pursuit of imperfection...