This week, I was an author on an open letter to the climate science community, calling for and end to the use of the dreaded “Rainbow” colour palette for scientific visualisation (mirrored over at my data viz blog Better Figures). It was the busiest day ever at both CLB and BF, and we got lots of great feedback on twitter and in person (more on this later, I hope).
We used the hashtag #endrainbow, which I suggested we change at the last minute from #endtherainbow, because it was a bit shorter. The upshot of this was that if you type #end… into the twitter search engine, it auto-suggests #endebola as a completion. This got me to thinking.
Should I *really* be spending my time worrying about people using a non-optimal colour palette in their slightly obscure publications? Wouldn’t it be better to devote my energies to a bigger, world-shattering problem, like Ebola? Is “Rainbow”, cosmically speaking, important?
Having thought about it, I still think that somebody should sweat the smaller stuff. Why?
1) It’s important in its own right.
I’ve had colleagues come up to me, or email me this week, and say “thanks for highlighting this, I’m colour blind and had to put up with rubbish visualisation for years”. It’s easy to ignore or exclude people if they are a relatively small group. This isn’t an excuse in my book, and in this case the solutions are basically free, and have little to no consequence for the larger population.
2) It’s an indicator of a bigger thing.
Clear communication matters. Good data visualisation matters, and beyond climate science. Beyond science, even. For me, support for the “Rainbow” campaign is an indicator that communication is important, and worth taking care over. We’re hoping that it could interest people more widely in good visual communication.
3) You have to start somewhere.
This comic, from the brilliant Bill Watterson, stuck with me when I read it as a kid.
As Kate Marvel points out in her excellent piece on palaeontology and Royalty (I think that was what it was about), the existence of other Bad Things does not negate a Bad Thing. We’re in a pretty unique and privileged position of being able to push hard to sort a small, essentially solvable problem, that might have positive impacts beyond our immediate field. They don’t come along every day, especially not in our line of work. We should take the opportunity.
Good post Doug.
Related issue: I’m still seeing people use log scales on one axis only (typically the x axis), in a way that creates either an exaggerated, or outright false, impression of whatever point the authors are attempting to make (often, an increasing or decreasing rate of y as f(x)). I’m talking about people who should know better are likely aware of exactly what they’re doing. This is an absolute NO GO, as it can be highly misleading. Indeed, it’s worse than the abrupt color change issue IMO.