Hacker News Reader

Comment by neura | Hacker News Reader

I can't be the only person looking at the top of the article, thinking "but if you just add the descender and ascender values together (the total height of the character, not the block space that it occupies), you get almost 32."

I'm guessing if you even slightly tweak the way the measurements are being taken, you'll get 32. even if you just assume "always round up", you get 32.

This is one of the strongest cases of manipulating data by providing a bunch of data points to obscure the one point, enabling belief in "well, they provide all this data and none of it is 32, so they must have a solid point here!"

This misleading opening really just makes me want to avoid reading the rest of the article. I hope there wasn't any real information hiding behind that facade. :x

View on HN

Here’s a picture with more samples:

https://s.tonsky.me/imgs/font_size.png

There’s nothing to hide: some fonts’ height adds up to font-size, some don’t. Because it’s not a rule, you can’t rely on it, so it doesn’t change a narrative.

Fira Code is one of the fonts that don’t add up, though. 31.19 != 32, no matter how hard you massage it.

P.S. for other examples, I also chose the extreme examples to better illustrate the point. But I only chose relatively well-known fonts, so it’s not all random or manipulated.

>This misleading opening really just makes me want to avoid reading the rest of the article. I hope there wasn't any real information hiding behind that facade. :x

I think you've missed the whole point of the article (which is that this is font-specific as opposed as universal), and you've made a conspiracy theory to boot!

"This is one of the strongest cases of manipulating data" - lol.

That’s an absurd take... the article’s first paragraph is quite clear he’s going to tell you exactly where that 32 ran off to, and despite your random arithmetic that happens to find the number with no double-checking whether it works on any other font-size or font (equivalent to finding your birthday in license plates, by arbitrary manipulations of the numbers available), it turns out the 32 is in fact missing from any visible artifact of the font.

The author could have also mentioned, among the other not-32 heights, the distance from ascender to descender being 31.19 and explained how that's not it either. Instead, he just didn't mention it, as if he was trying to hide it because it was inconvenient to his narrative.

He could have also mentioned that floor(cap-height + x-height + descender) also equals 32, but isn’t the answer either. And probably a few other arbitrary arithmetic combinations would work as well.

But it seems it me it’d be much more efficient to just explain the real answer instead of telling you all the ways you could misinterpret it... which is exactly what he goes on to do.

Reply on news.ycombinator.com