Hacker News new | comments | show | ask | jobs | submit login
Mapping the Census: A Dot for Every Person (theatlanticcities.com)
92 points by japhyr 1754 days ago | hide | past | web | 26 comments | favorite

Neat. I've often thought about setting up an installation with one pixel per person...in the world.

With HDTV's it's not too extravagant either - about 3345 panels. A lot (of drivers too) but doable.

Any support for it? I was thinking that a circular dome ring of panels would be an amazing visualization tool.

Ideas: 1. current age of every person, represented by color 2. Language 3. Religion 4. State (health, nourishment, etc)

But then I run out - but I think it would be an incredible tool for a variety of data sets.

You could start smaller too (country level)

Has anyone done this yet? I'm aware there are no new ideas on the Internet.

"I've often thought about setting up an installation with one pixel per person"

The problem with this kind of simple scatterplot is that there is an maximum density that you can represent, when the dots start to overlap. And you can't tell what that maximum density is by looking at the map, so it in effect misrepresents the data. The example in the article compensates for this by allowing zooming, but the problem is still there at most zoom levels. To avoid this you need to use one of the techniques that tessellates the plane and colors each tile according to the average value in the tile.

I read bwooce's comment to mean one pixel per person literally (not in map projection). If you're not trying to represent geography at the same time you don't have that problem. You could still sort the data representing individuals by two variables if you want, e.g. latitude and longitude, and map them to x and y, but it wouldn't be a map.

You're right, I should have read the comment more carefully. An interesting idea! So please take my comment as applying to the article rather than bwooce's comment.

Yep. That's what I thought too. The more number on a pixel, the darker (or brighter) it gets. But I don't know how far that can be pushed.

Facing a similar problem, there is an art exhibit somewhere that shows the complete contents of a hard drive (1TB as I recall), every single 1 and 0 (yes, at bit level, written as 1s and 0s). Details escape me, but I did see it in the last year. It was surprisingly readable, printed on the walls.

I was hoping for a downloadable single-image version but he doesn't have one. I made one myself by compositing his tiles.

Image: http://frammish.org/dots.png

Code: http://pastie.org/5602534

Your first link crashed my Safari Browser.

It caused a spike in Firefox's memory usage of over 1 extra gigabyte.

Minneapolis looks a lot larger than I'd have guessed.

The pattern of dots connecting larger cities in the middle of the country is really impressive - and a bit suspicious how straight the lines are.

Have you traveled that part of the country? There are large areas where every road is east-west or north-south, and the towns are evenly spaced for very long distances. I don't see anything suspicious about it at all. Is there something I'm missing?

Sounds right to me. Here's an example:


There's some information on the origins of those remarkably regular (and recursive) grid roads in this article: http://en.wikipedia.org/wiki/Public_Land_Survey_System

I imagine these correspond to highways. I did a similar thing a while back with postal codes and their density. I don't know why I was so blown away, but I was. Take the West Coast for example: http://i.imgur.com/WBzWG.png

I have also heard that the regular spacing of towns east-west is how far a steam locomotive could go before it needed water, but have no idea as to the accuracy of that claim.

I've heard the same as authoritative from a postgrad studying U.S. history; I'm with you, though, that's just as anecdotal as far as the HN standard of truth goes.

I wish my own U.S. History course would have covered ground like this instead of starting in the early 20th century with the plight of immigrants. No independence, no civil war, nada. I felt cheated by that (high school) course and I've been meaning to fill in the gaps.

> No independence, no civil war, nada.

To be honest, most schools that do cover those do an even worse job than not teaching anything at all.

If you're interested in filling in the gaps on your own:

http://www.amazon.com/Dont-Know-About-History-Anniversary/dp... http://www.amazon.com/Lies-My-Teacher-Told-Everything/dp/074...

Thank you for the recommendations.

The increasingly magnified images remind me of playing with fractal-viewing programs, with one major difference: here the zooming in leads to distinct patterns at specific resolutions, rather than repetition of one at a different scale. Once you get to the clearly visible "blocks" of Detroit and LA you're learning something completely different. Fascinating.

I like how you can toggle between the normal Google map and the census dots. It's fun to try to guess what the large white spots are in areas I'm familiar with, such as the Crystal Springs Reservoir near San Mateo, CA.

I zoomed in on my neighborhood. My question is... I live on the edge of a national park, and I see a number of dots (roughly 50) within that national park. How is that possible?

From my understanding in a lot (at least a few) National Parks not all of the land is owned by the park. There are spots of land in the parks that are privately owned; and, some people have decided to build homes on that private land in the National Parks.

CBS Sunday morning had a piece [1] on the whole deal with people building homes in national parks.


Same, I live near a National Forest boundary and it's showing a substantial number of dots inside that boundary in places where there are most definitely no homes. There are occasional chunks of private land within the boundary, but not where these dots are showing—they are, in fact, often within a designated wilderness boundary which allows no permanent residents or active buildings.

The dots are not placed to exact addresses, but roughly evenly distributed within census blocks. The number inside the block should be correct, not the exact placement.

Hermits in treehouses?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact