Candidates' Finances by State (NYTimes.com)

Take a look at this graphic from the NYTimes.

In general, I have little bad to say about this. After you click on a candidate's name you can see where (geographically) their finances have come from. You can even change the time period on the bottom, or click "play" to see them change over time (note that this view is not cumulative -- it might be nice to be able to change that feature). Hovering over a bubble shows the name of that region, the exact amount coming from there, and the zip codes covered. On the whole, it is incredibly intuitive to understand what is going on, and to use the more advanced features, which are two of the primary things to look at.

That being said, we can make some general comments on the overall nature of graphs like this, since we have a specimen which is generally free of flaws.

First, we have the problem of the placement of the bubbles, and the areas which they represent. The Times has apparently chosen areas spanning 1000 zip codes (or more accurately, the numbers span 1000, but many probably are not used), which seems like a mechanism that could be / is easily standardized. However, it leads to areas which do not include equal populations. At the same time, it leadings to a preponderance of bubbles in the northeast and California and a dearth in middle America, which causes both wasted space and significant overlap. The overlap problem has been cleverly solved by making the bubbles partially transparent, but it is still not the optimal form for readability.

Second, we must wonder "compared to what?" question. Bubbles here represent absolute amounts, but one wonders if we should have some rule like Tufte's "always convert to absolute instead of nominal dollar values". Should these be relative to population? Relative to income? Relative to the cost of living? All of these factors would be good explanations for why some areas have higher donations, but the visualization simply dismisses them.

Finally, something which I might like to see would be the ability to select multiple candidates either to aggregate or compare. For example, if we add up all the donations, do all the bubbles equalize in size? Similarly, perhaps it would be possible to show the margin between two candidates (or sets of candidates) by having bubbles be the absolute value of the diffenence and using different colors to show who is ahead.

What do other people thing? I feel the NYTimes has created an excellent version of a fairly traditional type of graph, but can we create a new form which is better?