Great that you are using Processing to show data graphically.
But, and please don't take this the wrong way as I am trying to provide constructive criticism, I think there are some significant problems with your design. You are not alone in this, as there are plenty of shockingly poor infographics published professionally. Some things to think about.
1. Why the sinuous curve? The layout of the bills creates a sort of perspective effect that makes it extra difficult to compare sizes of bills at the top and bottom. Layering the bills means that different bills have a different amount of exposure depending on where they are on the curve (and entirely independent of the data you are trying to show).
2. Symbol size should be proportional to data magnitude. But this is not the case in your graphic. You state that 'length of bill is log proportional to job density', but given that all bills are the same aspect ratio (as far as I can tell), it is the area that is varying - ie the square of the length. This gives a rather complex relation between data magnitude and symbol size. Given the problems of occlusion and perspect (see above), this becomes even more complicated. The largest value is only about 100x the smallest (but see comments on DC below) so a straightforwrd value to sqrt(area) mapping would be easier to interpret.
3. I'd question the value of the extra 'ink' used to show the states in all four corners, the 'FED JOB' label and the inner rectangle. This just adds to the complexity of the image without adding to its value. I realise you are trying to enrich the metaphor of the ATM spitting out money, but infographic metaphors should enhance understanding of the data, not make it more difficult. I think some of
Ben Fry's (Processing) examples show how you can have aestheticly pleasing, but still highly data rich graphics that are easy to interpret.
4. Consider whether you can use layout to enrich the information you are showing. An obvious way of doing this would be to incorporate some form of mapping the states. That way you can see if geography is playing a role in the patterns you are trying to convey.
5. DC, Alaska and anomalies. As anyone who has tried to map any human aspect of US geography knows, DC will almost always emerge as an outlier. In many cases this is simply due to it being a spatial unit that is almost entirely urban. In other words, it may not reflect anything unique about the phenomenon you are interested in. I'd be wary of scaling an entire graphic to account for DC as this often means that the graphic is effectively just describing two states - DC and not-DC.
I'd very strongly recommend having a look at
Edward Tufte's books, especially
The Visual Display of Quantitative Information and
Envisioning Information, which both provide lots of sound advice on sensible design of information graphics. Also recommended is the work of
Stephen Few who has lots of sound advice in this area.
Oh dear, that did sound a bit harsh, but I do hope you take it in the constructive spirit in which it was intended.
Jo
