Hacker News Reader

Comment by jonchang | Hacker News Reader

There's an interesting link to an analysis an (unaffiliated?) economist that suggests the study's methodology is flawed, in that it doesn't properly account for the simultaneous drop in gasoline prices (known to increase driving and therefore traffic deaths) as well as being unable to explain rises in death rates in both urban and rural areas (because Uber/Lyft are primarily urban)

http://cityobservatory.org/unsafe-uber-lethal-lyft-were-skep...

View on HN

Every time you see a claim like “study links X to Y” or “X is correlated with Y”, you should always always always remember that there are three possibilities: X causes Y, Y causes X, or some third factor Z causes both X and Y. I can’t understate the importance of thinking this way and not jumping to unwarranted conclusions. You should also exercise your imagination by trying to imagine why each of these three possibilities might hold.

Edit: There's also the possibility of spurious correlations, as mentioned in the comments below.

I try to make this healthy skepticism / criticism a bit more succinct for laymen by linking to spurious correlations http://tylervigen.com/spurious-correlations

I see a lot of evidence that "thinking right" is astoundingly hard.

But I don't see very many statements from real humans of any field, background, or level of intelligence/education that acknowledge the enormity of the challenge.

I'm not sure it's "possible", but I think we're profoundly suffering for the lack of an open-ended, recursive knowledge-production process more efficient than what we've left up to the emergent behavior of a chaotic system.

The Scientific Method can't be something we just learn and file away. Method means it is being applied, we constantly need to practice and study in its application. It is a way of being, not a fact.

Also, as the number of random variables increases, so to does the probability that some will be correlated just by chance. Pure noise. Can’t post the reference right now, but the author Nassim Taleb has either a paper or blog post that shows this neatly using random matrices.

Reply on news.ycombinator.com