Hacker News Reader

When "cheating" is defined as preventing deaths in (i) the most common and (ii) the most deadly crash scenarios, I don't think it counts as cheating.

In this case, "cheating" is defined as designing only to pass whatever the NHTSA happens to be testing now, and ignoring all other areas where things could go wrong.

edit: I'm not saying I agree with GP, just explaining what I think they meant by stating that they were "cheating."

That doesn't really seem to be the case, though. The NHTSA looks at highway and road data to determine what types of crashes result in the highest number of fatalities and then creates tests around those. If "cheating" implies that they're only designed to pass what the NHTSA happens to be testing then, by extension, the vehicle would protect the occupants against the types of crashes that are most likely to result in a fatality. That seems like huge win to me.

I'm not saying I agree with OP, just explaining what I think they meant by stating that they were "cheating."

> ignoring all other areas where things could go wrong

Doesn't this just say that NHSTA is ignoring other areas where things could go wrong?

You are presenting a fatalistic theory that Tesla cars have some hidden safety flaws not found in ICE cars without any backing proof whatsoever. In fact, the NHTSA results show the opposite, where the top 3 cars with lowest probability of injury are Tesla cars. That is what I call FUD and it is happening daily on Twitter.

It's a reasonable thought experiment. Though it seems unlikely to be true it logically follows from the rest of the comment. This doesn't look like FUD to me.

You're not being very charitable here; you're assuming the parent has bad faith. Why can't we just be a bunch of people having fun together on the internet?

I strongly suspect the parent was arguing in bad faith. The word choice of "cheating" and the baseless allegation that outside of the specific test crashes that there were likely to be fatalities is indicative of bad faith.

A good faith argument could be made to discuss how the crash tests like any sort of benchmarking are likely to lead to optimization for the tests - you get more of what you can measure - so one needs to be careful that the results are broadly applicable. This of course would apply to all makers, not just Tesla. But that is not the argument the parent made.

To call this "cheating", you have to assume that car manufacturers are more knowledgeable about car safety in general than NHTSA, to be able to spot the loopholes. It may be true, they have more resources to invest in research and to hire experts in this domain, after all.

Still, I prefer to follow Hanlon's razor, and not assume malice. This kind of mistake always seem obvious in hindsight, but if it was, why was there no one warning about it at the time?

yes they are. nhtsa and others require so much preparation and analysis for each test that for the cost of one run the makers can run upwards of 1000 chasis compresion test which will cover every angle imaginable. so yes, the manufacturers know a lot more. but it is practical knowldge not cerfied and traceable and repeatable.

Ah yes, those crash tests that are not tested by anybody because we all know or assume ICE cars will do well in them, THOSE will be Tesla's achilles' heel!

It seems more likely that crash tests represent our best understanding of common crashes in the real world, in order to build vehicles that protect their occupants best.

I highly doubt it's likely there are other possible tests that are "extremely fatal" only with Teslas.

Well that battery could and does catch fire during collisions, however rare that might be.

True, and that never happens with cars with gasoline in the tanks.

(/sarcasm)

To add substance:

From the nhsta site (https://www.nhtsa.gov/sites/nhtsa.dot.gov/files/documents/12...)

Observations on various aspects:

* "Because the lower the flash point, the more hazardous the fire risk, flammable electrolyte solvents in Li-ion batteries are more or less as hazardous as conventional vehicular fuels."

* "Because the narrower the flammability range, the less hazardous the fire risk, flammable electrolytic solvents in Li-ion batteries are about as hazardous as conventional vehicular fuels, and certainly less hazardous than hydrogen."

* "Because the lower the minimum ignition temperature, the more hazardous the fire risk, flammable electrolytic solvents are not as hazardous as conventional vehicular fuels."

* "Because the lower the minimum spark energy, the more hazardous the fire risk, flammable electrolytic solvents are probably about as hazardous as conventional vehicular fuels."

* "Because the lower the maximum flame temperature the lower the exposure risk (skin burns) to fire, electrolytic solvents are expected to pose no more of a severe risk to burn injuries than conventional vehicular fuels, gasoline and diesel."

* "Because of the probable similarity in the magnitude of the maximum flame temperatures, if the combustion of the electrolytic solvents were confined, the resulting overpressure that could build up would maximize at about the same levels as for the other flammable fuels, meaning that the burst damage that would result from Li-ion explosions would also be comparable to that from conventional vehicular fuels."

* "The thermal energy expected to be released upon initiation and combustion of the electrolytic solvents would be less than that of gasoline and diesel fuel."

Also don't forget that the millions of miles and hundreds of thousands of teslas gives enough size to have statistically comparable results to ice engines, and it looks great for tesla. Edit - here's an article on this from a few years ago https://www.businessinsider.com/elon-musk-blog-post-on-tesla...

That is far safer for the driver and passengers than a gasoline fire. People only fear it because it is unfamiliar.

>it is very likely that other kind of crash tests, that are not even tested by anybody because internal combustion engine cars always did so well on those, are extremely fatal on teslas.

You assume there are such crash tests, but don't describe any specific ones. I am guessing that is because none exist, but perhaps I am mistaken.