How tiny can a data visualization be and still be meaningful? Edward Tufte pushed the limit of infographic density with his "sparklines ," and now Dan Zarella of Hubspot does something similar -- with a micro-sized heatmap of a similarly micro-sized form of communication: the 140-character tweet . According to his visual analysis, links inserted into tweets are more likely to be clicked if they appear somewhere between the start and the middle of the tweet, not the end (as most of us would assume and do in practice).
That data-driven insight is intriguing enough, but so is the way Zarella visualized it. If you're like me, when you hear "heat map" you think of a two-dimensional map of a website color-coded in something like Predator-vision, which visualizes where users focus their attention onscreen. Zarella's one-dimensional heat map (it only maps points on a line) elegantly matches the parsimoniousness of Twitter itself -- in fact, it could probably be squashed into the physical space that a tweet takes up on a smartphone screen and lose none of its informative power.
Heck, if a sparkline is designed to be meaningful at the size of a pencil eraser, Zarella's micro-heatmap could probably fit into the size of a favicon and still get its point across: human beings are pretty excellent at distinguishing edges and color-contrasts in fine detail, and the areas in Zarella's graph corresponding to "high clickthrough rate"--ie, the parts of his graph that actually carry the useful meaning in contrast the rest, which essentially functions as "background"--would jump right out even if it were shrink-rayed. It would be neat to see Zarella's graph translated into ASCII, as well--a tiny string that could be tweeted along with some interesting commentary and still fit into 140 characters.
OK, maybe that's pushing it. But experimenting with the perceptual limits of information density in our displays isn't just idle fiddling. Twitter itself is a (very successful) experiment in this regard: five years ago, would anyone have believed that a network of SMS-length mind-burps would be capable of transmitting world-changing information? What if Twitter (or some enterprising startup) could offer a suite of visual analytics whose gestalt state could be grokked by flicking your eyes at a square of screen-space no bigger than a few dozen pixels? Alone, such a nano-visualization might be more cute than useful -- but what if such visually dense graphics were aggregated, so that en masse they harnessed the formidable power of the human visual system  to reveal useful patterns across many different channels, states, or media? Sounds like a job for the next Edward Tufte.