Hacker News Reader

Comment by an1sotropy | Hacker News Reader

In the tradition of "considered harmful" papers, I think this is one of the less convincing. Or rather, it doesn't meaningfully answer the question: if the rainbow colormap is so awful, why is it still so common? Is it really "due to inertia", as the authors suggest? Or is there some other virtue?

How about: I want to maximize the number of distinctions between values that are possible via comparisons of colors. Something that takes a long path through colorspace is best for this. Short simple paths do better enable making ordinal judgements between values (which the authors care a lot about), but they're not as good for maximizing distinctions.

Were data vis a more conceptually mature discipline, we wouldn't rely on this kind of simple prescriptive guidance.

View on HN

The problem is that people will naturally interpret colors on certain scales whether that was intended by the authors or not. The goal of visualization is to make the natural reaction the correct one.

https://jakevdp.github.io/blog/2014/10/16/how-bad-is-your-co...

The link above goes over some interesting plots that have very misleading color gradients.

The implicit assumption there is that all data visualizations should be fast or "natural" to comprehend. Some people think that all user interfaces should be simple and easy for novices. Those people are not airplane pilots, or train engineers, or art historians for that matter.

Not all visual things are simple, and not all simple visual things are effective for their intended purpose.

If non-monotonic luminance variations are an acceptable cost for a colormap that otherwise offers superior discriminability, than that's a colormap that someone may have a good reason to choose.

Theoretically sure, but in reality people who like jet don't even know what the issue with jet is, let alone they can correct the non-lineairity.

In what kind of plots do you think maximum discriminability, at the cost of discriminating things that don't exist, is more important than linearity?

For example, if you want to show the spatial extent of some Gibbs (ringing) phenomenon at the edge of a structure that itself takes on a wide range of values (this is descriptive of visualizations I've had to make). The basic shape of the structure, and its low-frequency value patterns are already known (part of the problem definition), so increasing discriminability will make the colormap more useful.

You're increasing the discriminability at the cost of high _local_ distortion, I don't see how that is a good tradeoff.

As a comparison, what would you rather have: a non-distorted plot or a plot that is on average about three times as big, but it's heavily distorted. At some places it's smaller than the original and on other places it's 5 times bigger. Would you really choose the bigger figure? That is what jet is.

Look at this plot https://youtu.be/xAoljeRJ3lU?t=791 (watch the whole video if you have the time), ideally you want the derivatives (the 2 plots on the left top) to be constant so not to distort your data. Jet is all over the place while viridis is basically constant (not shown in the slides, but the derivative stays all the way around 120).

We're not really disagreeing. I also like properties of Viridis, but it doesn't make it better for all tasks. It is a shorter path through color space, so it is not as good for some tasks. "Distortion" is not really a well-defined term, absent a particular visual analysis task. People can say "use viridis" with as much self-assured confidence as others say "use python", but they're still missing the point that different needs call for different tools.

I'm disagreeing with you, and I say "never use jet" (I'm more a fan of Magma) with a lot more confidence than I would say "use python". Different needs call for different tools but there is no need that calls for making distorted plots of your data. Jet adds information that is not there, you never want that.

Every data visualization is necessarily a contrivance because it is creating visual features for things that weren't already visual. Data visualization research has not been able to formalize a rigorous notion of visual "information" in a form that usefully predicts performance across the enormous variety of visualization tasks and settings for which colormaps might be applied. "Distortion" is also a common but somewhat meaningless term. You're referring to perceptual effects (related to color appearance) that are side-effects of using "jet", but again, for some tasks they may be an acceptable side-effect, given the benefits of increased discriminability.

Our different levels of certainty and conviction may arise from my being a data vis academic, rather than a tool user or tool builder.

I think it is just misuse.

Rainbow colormaps show banding, which is great when you want to have an idea of contour shapes. For example, the most common rainbow colormap, "jet" is great at showing jet engine exhaust plumes, hence the name. But if the value matter more than the shape, then it is terrible. The same banding that highlight shapes create artificial discontinuities in otherwise smooth data.

Maybe it is like the Comic Sans of color maps. It serves a purpose. In the case of Comic Sans, it was made to give a comic font to the comic characters of Microsoft Bob. And people liked it and used it everywhere, much to the dismay of typographers. "Jet" serves a purpose, but people misuse it because it looks cool.

I think the problem is that rainbow shows banding, but only in some arbitrary narrow range, and then pretty much ignores a large swath of values that all map to slightly different hues of green. So the "bands" you see are not representative of the underlying data, unless you're lucky.

If you want to see bands I think you're better served by just digitizing the value into, say, 10 ranges, and color each range in uniform color, like how they did it in old-school paper maps.

> if the rainbow colormap is so awful, why is it still so common?

Because it's the default for a lot of matplotlib visualizations. Per Goethe: "Misunderstandings and lethargy perhaps produce more wrong in the world than deceit and malice do."

When I was a student, I too used it for a relatively banal reason: it looked pretty.

I always thought that it was a quick metaphor for heat(red, white) and cold (blue, black). I have a hard time relating other colormaps in the same way.

Maybe the difference is not 'conceptual' but some atavic or cultural reaction?

Reply on news.ycombinator.com