If every person counted in during the 2010 U.S. Census got a dot representing their location, what would a map of the United States look like? Â
Brandon Martin-Anderson has answered that question with his Census Dotmap (no longer active).  Martin-Anderson took all 308,450,225 persons counted and assigned each one a dot, creating a map of the 2010 Census. Â
He did so by using a Python script to generate points from US Census block-level counts.  With each census block, the dots representing each person counted within that block were uniformly distributed.
Martin-Anderson has a writeup describing the steps he took (Wayback Machine link) to convert a shapefile containing census counts for all of the census blocks within the lower 48 states into a CSV file over 17GB in size.
The writeup includes links to the python script used.
Martin-Anderson’s motivation for creating this census map?
I wanted an image of human settlement patterns unmediated by proxies like city boundaries, arterial roads, state lines, &c. Also, it was an interesting challenge.
As with other projects that mapped out a widespread feature using only one data source (see the map of  the United States when mapping Starbucks or McDonald locations, or the world at night), the map easily forms a recognizable view of the United States.
Martin-Anderson explains, “I’m hosting the tiles on Amazon S3, and showing them using the Google Maps API v3, using the ImageMapType” Â
Users can zoom into areas of the map to see more detail about the distribution of dots. Â If you get lost when zoomed in, use the menu that shows up in the upper right hand corner to toggle back and forth between a Google base map showing streets and the Census dot map.
The map below shows a zoomed in view of Seattle, Washington, with a comparison to Google’s street base map on the left. Major thoroughfares such as the 5 freeway, lakes, and parks all show up as white zones in a sea of black dots.
Is it possible to download a digital version of a certain area? One that I can use in our GIS?
There’s a lot to like about this project and a lot to not like. I’ll start with the likes:
Like:
1. A gargantuan effort to collate all the Census Block geography into a usable, streaming interface. And that’s just the geography.
2. On top of that, writing the coding to be able to handle and output the data is a unique effort. This is a supreme amount of data we’re talking about here.
3. Having the dot *be* the symbolization, just black – very well done. There is no need to include any other layers, which would take away from the message.
4. The representation here truly shows the depth of the work of the Census Geography division and their partners. Census Blocks with no people show up without dots, and act as a perfect exclusion zone. The efforts of the Census to attempt to capture all the non-habitable areas is light years better than the last decade’s version.
5. The duality of simplicity and complexity increase the aesthetic appeal of the data.
However.
Do not like:
1. Ugh. Dot Density without actual exclusion zones is still a random splattering of dots in a shape. Consider a hi-rise apartment building in a densely packed city: in a 2D world, those stacked respondents do not accurately translate into a true representation of the space. Additionally, if you have the same hi-rise apartment building with a footprint that covers a fraction of a block area, its respondents are scattered amongst the entire block. Randomly.
2. While the Census Block is the smallest of the Census geographies, there are still errors within the data. These errors show up as illogical blocks within rural areas and subdivisions, i.e., several blocks cut out of a larger subdivision in a seemingly random pattern. (it happens)
3. Given the prior two points, you will not be able to see the shapes of the geographies on a good dot density map. As you drill down closer to any specific area in this map, you can make out the the areas of the blocks used. At this point, the dot density concept is not showing density, but rather shading.
4. The way this map is being presented on various blogs – “Every person in the US shown on one map” “Everybody in the US as a dot in a map” or, my current favorite: “This Crazy Map Has One Dot for Every Person in the United States”. *sigh* Please. I imagine the creator of this map is facepalming every time this misrepresentation of his work is thrown out there.
5. The worst case of this misrepresentation is thus: Every Person in the US. Nope, every person *counted* at the 2010 US Census. Given the return rate of 72%, this map represents a little under two-thirds of the actual population. Turn in your Census forms, people.
All in all, I applaud the maker of this map with taking the time to visualize something few people realize – where we are concentrated.
@Leanne – if you click on the link in which the map methods are described, you will find a link to the Census ftp site where the shape files are located.