Stunning Business Intelligence Visualizations… from 1870
Posted by Timo Elliott on Thursday, March 18, 2010 · 15 Comments
(2 votes, average: 5.00 out of 5)
Loading ...
FastCompany has a great article this week on the results of the 1870 census, and the hand-made graphics (“BI –2.0”?) that were made from the data (thanks to RadicalCartography.net and the Library of Congress). [NOTE -- the original version of this post erroneously said "1830" throughout]
Here’s a selection of my favorites – click on each one to get the full graphic.
US geology – the hand-shading is so much nicer to look at than computer-generated graphics:
Nationality data mapped:
Here’s a chart that would have looked good on a 19th century Recovery.org website (where, incidentally, data visualization guru Edward Tufte is now a Presidential Advisor). There’s a lot of history illustrated here:
- The light blue spike in the lower right is the US Civil War (1861-1865), which prompted the introduction of a “temporary” progressive income tax (the light pink blob in the lower left) that was originally supposed to expire in 1866!
- The pink blobs on the right are a record of expenditures on the US Indian wars
- The dark blue spike in the middle of the left-hand chart shows land sales in the West – which may have caused the collapse of the banking system in 1837
This is a beautiful rendering of church-going in the US. Interestingly, rather than show non-churchgoers as a separate bar, they are relegated to a grey box around the outside.
Here’s a nice example of showing data through proportions.
And here’s proof that men are bigger idiots than women! (of course, they meant what we’d call “mentally ill” these days)
Population charts showing that the “cowboy states” (e.g. Wyoming, Nevada, Montana) were populated by young male foreigners, while Utah had LOTS of children.
This chart is a good example of something that hasn’t changed much in the last couple of centuries – the notion that you should show data just because you have it. For example, Chart 1 below show deaths per month per state. Is this data actionable in any way?
Later charts aren’t nearly as attractive, like these, from the 2000 Census:
Here’s a PowerPoint file with some of the key charts already included:
If you liked this post, you might also like:
- Giant Brains: Did Edmund Berkeley Predict Email, the Internet, ERP Systems and the iPhone in 1949?
- The Real Pioneer of Business Intelligence (and BI 2.0)?
- Performance Excellence, Past and Future?
I was the eighth employee of 
Beautiful stuff Timo. Thanks for sharing.
Best,
Dave
Interesting stuff – shows that visualization of data is not anything new. I would take issue, however, with your statement regarding the notion that mortality information isn’t actionable? Lots of health service and infrastructure planning relies on this kind of information. Would indicate where I need to be sending doctors and medicine, building hospitals, etc. And almost every entity that tracks growth information, (sales, income, units), needs to net out costs/losses to really understand what’s going on.
John, I think you’re illustrating my point. In general, yes, mortality data is useful, of course. But just because data is useful in theory doesn’t mean that a specific data set is going to be useful in practice. In this case, it’s deaths by month, which isn’t going to help much with the number of hospitals, etc., and more importantly, given the differences between all the states, it looks like a snapshot of randomness, and that’s not going to help anybody with their planning… Actually, that’s another point — a lot of companies DO try to plan today using data without taking into account of random variation.
One of my pet themes is that BI has to take better account of “fuzzy” data, helping average business people determine what’s real and what’s just noise. Experts can do this today with the right tools, but we need the application to do more of the expertise for us — and this is probably only possible in the context of an application (another reason why BI and ERP go together well)…
Would be nice to have a ‘retro’ pack to create visualisations like this.
I’d like to see those paper wrinkels in my charts. If it’s old, it’s probably true :^).
I agree! Maybe a nice “theme” add-in for Xcelsius?
Hi Timo,
Great blog
The link to the ppt is not working
Can you check this
Thanks,
Richard
Thanks for the heads-up — now fixed.