this post was submitted on 09 Feb 2025
981 points (98.5% liked)
Technology
62012 readers
4693 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Are you telling me that 35,000 vehicles is not a sufficient sample size to assess safety? Are you for real?
No. Incidence is a measure of probability of events over time (or with cars alternatively over miles). If the number of events is low (and 4 is low), your confidence intervals are extremely wide (which is the statistical way to say, we have no idea what the real number may be). The comparison is striking, the pinto had 27 fires over 9 years in >3M vehicles. https://fuelarc.com/evs/its-official-the-cybertruck-is-more-explosive-than-the-ford-pinto/
Let's add that idiots buy cybertrucks who disproportionately think it's bulletproof...
Again, "analyses" like this make great clickbait but contribute very little to our understanding, and that will remain the case even regardless of you getting angry at me about it or not.
And the answer is"What is the Poisson Distribution" Alex.
There is literally a distribution that describes the occurences of low probability events in large populations. It was developed to study deaths by horse kick in the Prussian army. So confidence intervals never come into it. You're applying Stats for Communications Majors reasoning to an adult problem.
Well, the problem is, even if I take the single case where this one guy exploded himself with his truck and compare it to the Pinto data, the poisson distribution difference will probably be statistically significant, yet the measure would be absolutely useless from a real-world perspective, because it has nothing to do with the vehicle's design.
I'd also argue that many of these events might not even be entirely occurring independently from each other (i.e., some of the key assumptions of Poisson are incorrect here) when people do all sorts of stupid shit with these rolling garbage cans like shooting at them, submerging them, etc. in a meme-like fashion for Tiktok views. So 4 events might very well be influenced by non-design-based, non-random human factors, which applied to other vehicles could generate similar results, and if the analysis were serious, they would have individually reviewed how these whopping 4 events happened, accounted for reporting bias towards EV fires (especially Tesla) and compared it to the F150 or the Ford Lightning as an analogous vehicle.
And I know the internet tends to conflate condescension with competence, but seriously, you should understand the above-listed things as a stats teacher.
edits for clarity
edit 2: also, in the times of the prussian army they did not have to account for stuff like people suddenly starting to pull the horses' tails for social media views.
It's so great to be able to find comments such as yours, unfortunately it feels uncommon in Lemmy specially when certain names are mentioned, the bias and willfulness to shit on those are making people a bit blindsided and easy to guide through bad data usage. My first thought reading the title was about the statistical value of the numbers given, which doesn't detract from the actual quality or lack thereof of the vehicle. At the moment using elon musk or tesla in a title of an article will increase the traffic automatically. Which is why we constantly get every single shitty comment made by him reported with useless data.
Yeah it's part of the enshitification process. This is why Lemmy appears superior to reddit thus far. On reddit, the quintessential early "are you stupid?" response is enough to shut down the conversation. I'm glad it didn't happen here.
And it's not even that I disagree that Teslas have major safety design faults, you cannot put door opening mechanism on an electric actuator, because you'll get trapped. I'd never buy a car that doesn't have a mechanical door latch at hand (it's hidden on teslas). Interestingly Teslas used to be considered one of the safest vehicles, but I think a lot of it is, the early EV adopter demographic is simply characterized by much safer driving, and as this demographic shifted, more and more reckless drivers obtained Teslas. (I've been driving EVs since 2017 and around 2022 the demographic shift, at least for Teslas, became very obvious)