Hacker News Reader

Comment by madamelic | Hacker News Reader

>If you read the NHTSA letter they go into detail that their crash testing is done against fixed barriers, which means that it doesn't say anything about the safety of a lighter vehicle crashing into a heavier one. In a crash between a big heavy SUV and a Model 3, you may be better off in the SUV even if it has a "worse" crash rating.

All I am reading is that the NHTSA has bad tests?

If they are so mad Tesla is saying "We have the best crash test rating of any car", why not change the test to include those issues?

If someone was mad that I said "My code passes all of the tests", but there were still bugs in the code, the issue isn't my statement. The issue is that there is incomplete coverage.

View on HN

They have incomplete tests, since crash testing is really expensive.

But they are upfront in their guidelines about what exactly their tests do and do not say. Their tests are good for comparing cars in the same weight class. They are not good for saying what is the world's safest car.

If the NHTSA doesn't want their stats in marketing, why not figure out how to make them completely worthless.

Stars is way too marketable and deceptive to begin with. Make it Pass / Fail. 4 stars to actually failing is "fail". 5 stars is "pass". Actually failing being a rejection of approval and such a car wouldn't get a rating.

"Our car passes all necessary safety tests" ... good?

The NHTSA stars are adequate for their intended purpose - showing how well a car holds up in a crash compared to equivalent cars from other manufacturers. You just can't use them to make any conclusion about the overall risk of injury (which depends on the likelihood of a crash happening in the first place), can't compare them between cars of different weight classes, and can't draw meaningful conclusions about the relative safety of different five-star rated cars. Tesla did all three.

The last one might need a little elaboration. The NHTSA battery of tests is relatively small and limited. The lower the risk of injury in those tests is, the more dependent the actual real-world risk becomes on all the scenarios they didn't test. This isn't a hypothetical problem. The Model S also got the best score ever in the NHTSA tests and Tesla did a similar press release about that, but it didn't get the IIHS's highest rating because it performed badly in small-overlap collisions which weren't part of the NHTSA tests.

>If someone was mad that I said "My code passes all of the tests", but there were still bugs in the code, the issue isn't my statement.

Elon is saying more than "My car passes all of the tests"

> why not change the test to include those issues?

The problem with making comprehensive tests would mean small cars would be disadvantaged. You wouldn't pair up a heavyweight MMA fighter with a strawweight to figure out who is best, and likewise, you wouldn't have a bus and a small car t-bone each other to determine which is safer. (If you think a bus is a bad comparison since they're not usually sold to the general public, just try a pickup truck vs a small car.)

>If they are so mad Tesla is saying "We have the best crash test rating of any car", why not change the test to include those issues?

Or, they can do what they have done and tell Tesla to stop exaggerating the claims, which is a pattern with this company.

> If they are so mad Tesla is saying "We have the best crash test rating of any car", why not change the test to include those issues?

Because testing cars is expensive, and why should the association be beholden to a manufacturer's desire for marketing?

> All I am reading is that the NHTSA has bad tests?

Seems like a testing policy designed to keep people from avoiding buying small cars ever, and/or to keep people from agitating to ban or otherwise legally discourage heavy road vehicles.

If you hardcoded the 5 things that are tested, your code is still no good.

Reply on news.ycombinator.com