Hacker News Reader

Comment by deaneckles | Hacker News Reader

I don't like word clouds either (and I'm one of the authors!). You'll notice there is no cloud in the actual paper, just in the blog post.

Do you have any references about human perception of blue-red versus monochrome gradients? Would be interested to have good recommendations for such cases.

View on HN

I'm not GP, but the problem, in my colorblind opinion, is not the hue of the gradient. The extremes of your gradient are way too close together. Nothing really stands out. At least with blue/red, you can make it one extreme or the other (as opposed to white or black...on a white background).

That said, once they got to purple, I'd barely be able to distinguish them, anyway.

Robert Simmon has a good intro to color perception for data visualization http://earthobservatory.nasa.gov/blogs/elegantfigures/2013/0... .

Cynthia Brewer's work is currently the gold standard

http://colorbrewer2.org/

https://canvas.uw.edu/courses/891359/files/25713244/download...

My first thought when they mentioned it was the optical illusion where the color looks different because of the colors around it[0]. Maybe having distinct colors (red/blue) would help with that.

[0] Bad at explaining so an image: http://i.imgur.com/oyWO4IP.jpg

Hijacking the thread, since you're around:

Regarding the last graph, the observed data looks like it matches a log-normal distribution, not a power law... Is this correct? Not sure how to interpret it, though. How did you compute your estimation?

I think log-normal distributions usually fit this kind of data as well as power laws do. Anyway, such a plot is not the best way to choose between those two models (see http://arxiv.org/abs/0706.1062). We don't actually do any such test in the paper though (or say that it is a power law or log-normal). You might want to check out the paper for more details. https://www.facebook.com/publications/244240069095667/

Reply on news.ycombinator.com