Mistakes, we’ve drawn a few (2019) | Hacker News Reader

Mistakes, we’ve drawn a few (2019)

medium.economist.com

80 pointsdavidbarker12 comments

I think the dog one ("Mistake: Forcing a relationship by cherry-picking scales") was actually fine for a very specific reason: assuming dogs are of roughly constant density, weight is proportional to volume, which has a cubic relationship to a linear dimension like collar size. Therefore, if you have an increase of, say, ~7% (42 to 45) in your linear variable, you want an increase of 1.07^3 in your volumetric variable... which works out as 18 to ~22.1, which would make the two lines overlap even more. You don't want a 7% increase on both scales.

Intuitively, there should be a fairly tight correlation between size and weight, so overlapping them isn't misleading.

> Intuitively, there should be a fairly tight correlation between size and weight, so overlapping them isn't misleading.

Or maybe not, and tiny, thick-necked pugs are becoming more popular and this trend should be reflected somehow!

Wouldn't that be reflected in the graph (as rendered originally) if so?

For the dog one I would argue that they are so tightly correlated that neck size adds nothing useful to the chart and is just noise. The article is about dogs getting smaller. The weight chart shows this perfectly. The fact that smaller dogs also have smaller necks is neither surprising nor useful.

Discussed at the time: https://news.ycombinator.com/item?id=19508729

Fantastic article. Explains in such great detail about what goes into making a good chart from data (and what mistakes even experienced people can make!)

Does anyone find The Economist mirroring it's Medium blog at medium.economist.com a little roundabout?

Is there an infrastructural or SEO reason to do this?

Mirroring would imply it is being served from somewhere other than Medium, right? Is that the case or is this just a custom domain (medium.economist.com) pointing to their Medium blog? I believe custom domains are a common practice for larger brands.

> is this just a custom domain (medium.economist.com) pointing to their Medium blog? I believe custom domains are a common practice for larger brands.

I wouldn't label it "common practice for larger brands".

I would label it "common SEO abuse by larger brands".

How is a custom domain "SEO abuse"? I'm really not understanding what nefarious intent you're accusing the Economist of here.

Its SEO abuse to use their domain?

I find it odd that a publication has a blog. Is this content lower quality than what they usually publish? More meta? From a content management point of view it seems silly to have a completely separate stream of written content on a written content site.

Reply on news.ycombinator.com