This was far and away the most ambitious thing I’ve done so far – it actually required some designing, for which my colleague Hannah helped out. There’s a few technical things I learned, too, not least the cursor/ pointer thing which actually makes a huge difference when you’re hovering over the clicks.
Oddly enough the back end of this is both really tidy and really awful. It’s got loads of variables in it for locating all the boxes and graphs and text, which is all Good Practice. But the map is practically drawn by hand, not linked to any data. And adding the text in in jpegs cannot be a long term solution (though there really seems little other option at the moment. I must be posing my questions badly on Stack Overflow).
The technical challenge of doing this was probably less than the design challenge, which itself was less than the question of what it was that I/ we wanted to show. The time series graphs are fair enough – they’re pretty core to any understanding of the topic – but they aren’t exactly surprising. Potentially, though, there’s a lot you could do here. I’m reminded of the d3 tutorial about transitioning between types of graph which could be amazing on this kind of thing – lines to bars to steam graphs. One for later.