All 4 books by Edward Tufte now in
paperback editions, $100 for all 4
Visual Display of Quantitative Information
Beautiful EvidencePaper/printing = original clothbound books.
Only available through ET's Graphics Press:
catalog + shopping cart
All 4 clothbound books, autographed by the author $180
catalog + shopping cart
Edward Tufte e-books
Immediate download to any computer:
Visual and Statistical Thinking $5
The Cognitive Style of Powerpoint $5
Seeing Around + Feynman Diagrams $5
Data Analysis for Politics and Policy $9catalog + shopping cart
Edward Tufte one-day course,
Presenting Data and Information
Among subjects that challenge the imagination for how to display it graphically, the Internet would certaintly be a top contender for one of the most difficult.
I was intrigued by this story in New Scientist about a project to create a visual map the Internet. You can see a recent map here (700 x 700 pixel graphic). The homepage for this project is http://www.opte.org.
I'm not sure quite yet of how this adds to my knowledge of the Internet. It does change my view of the Internet, however, by making it beautiful visually, not just conceptually.
I do look forward to the project developing the ability to pinpoint your own website, so I can say, "Hey! I can see my webpage from here!"
-- Zen Faulkes (email)
See, of course, An Atlas of Cyberspaces at
-- Edward Tufte
Or you can get your very own map of the internet here. http://www.thinkgeek.com/cubegoodies/posters/techie/ They have posters of internet maps from 1999 to 2002 and even a poster with first beginnings of the internet. In the "Beginnings of the Internet" poster, it's interesting to see the many different types of visualizations. http://www.thinkgeek.com/cubegoodies/posters/techie/581e/detail/
-- Sean Gerety (email)
These kind of maps are interesting, but most neglect that links on the internet are directed. Generlised maps which deal with the character of the links can be enlightening.
I'm thinking here of something like "the continents of a directed network" from the visual companion to Albert-Laszlo Barabasi's Linked, http://www.nd.edu/~networks/linked/newfile16.htm
In the book, Barabasi talks of sections of the web (continents) which have a majority of internal and inward-bound links (e.g. Amazon) and those which have a majority of outward-bound links (e.g. Search engines).
-- Tom Carden (email)
Maybe something like this? http://www.telegeography.com/products/map_internet/index.php?PHPSESSID=8f059a2e175f15a6427467ebab3bafb1
[link updated March 2005]
-- Sean Gerety (email)
An interesting map to ring in the new year (from a Web comic, no less):
-- Jack Johnson (email)
Looking at the line drawing at the bottom of the webcomic, I note that the IP numbering follows a Hilbert Curve. This pleases me unduly, as I happened to be reading about them very recently, and had filed them mentally as 'pretty but of limited utility'.
Can a kindly contributor enlighten this non-computer-scientist as to why having a string of consecutive IPs translate into a contiguous region on the address space is useful?
-- Jakob Whitfield (email)
I think he just used that pattern for laying out the "tiles" so that you could trace the addresses consecutively on the map in an interesting way, and a block of consecutively assigned addresses would form more compact regions rather than strips across the map.
If they were simply in order across the row, you'd have to jump from the end of one row to the beginning of the next. And using a back-and-forth row scan would cause the labeled regions of the map to be in long strips going across the whole map rather than squarish islands.
It's not shown in his map, but I think theoretically each tile (subnet) could have nested internal Hilbert curve "mazes"? I'm not sure.
-- Reed Hedges (email)
Another map....this time of underwater internet cables: Underwater Internet Cabling
I like the data representation used.
-- Tim Banker (email)
The term "mapping the internet" may lead some readers to assume that the internet is a single, real entity. In fact, it is a group of networks that exist within a mutually-agreed-upon address space.
The 'traditional' address space for the internet has featured 2^32 possible addresses; a newer "IPv6" address space is gradually replacing the original IPv4 specification. This new space has a size of 2^128 possible locations.
Barrett Lyons' graphic is lovely, but the spatial plotting of points does not appear to yield much information.
I believe that the real problem with internet mapping is not the number of points, nor even what kinds of points they may happen to be. Instead, it is the fact that each point may represent a gateway to any number of subnetworks, each of which could -- in theory -- be as large as the internet itself. And each of those subnetworks may, in turn, host its own subnetworks of wildly variable size. A typical home network is a good example. The internet service provider (ISP) typically assigns a temporary IP address to the router in your home. You may have a network of computers, printers, and file servers in your home, any of which may be able to access the your ISP through your router. The small network in your home talks to the bigger network run by the ISP, which in turn talks to the internet through its own routers.
I have toyed with the notion of using two- or three-dimensional fractals as a starting point. Using IPv4 for simplicity, divide the fractal into 254 points representing all theoretical Class A networks (1-254.xxx.xxx.xxx); note that networks 0 and 255 are reserved. Divide each of those points into the 254 subsequent Class B networks (1-254.1- 254.xxx.xxx), and each of those into further sub-networks. Dark space is exactly that: dark space.
The beauty of using a fractal projection is that it enables a class-sensitive view of the component networks that make up the internet AND their internal subnetworks, if such information is available.
Coupled with Lyons' mapping of connections, a fractal projection of a group of networks -- be it the internet or a portion thereof -- could help network engineers design and maintain their increasingly complex networks.
It would also look really, really cool.
-- Michael Friesen (email)
I attached a little snipped that I think went with the Guardian piece on the recent service interruption in the Middle East.
The Submarine Cable Map is printed on Yupo, a synthetic material that is much more durable than traditional paper. The map is available folded or flat (shipped in a tube) and is available for $250.
Yupo is pretty interesting, it is a synthetic paper made from polypropylene resin (i.e. oil/petroleum) rather than trees. I first felt/read a Yupo book via Melcher Media's waterproof book division Durabooks of National Design Award winner William McDonough's Cradle to the Cradle.
You can get last year's map (two copies) for $150 - maybe I can ask them to donate a couple copies of the 2007 map.
-- Rob Bergin (email)
I'm surprised no one has brought up CAIDA:
They're doing some very interesting visualization of the core Internet protocols and traffic patterns over time.
-- Derek Chen-Becker (email)
I recently developed a different map of the Internet. This map looks at the world's two-digit country code domains and sizes them according to the population of each country or territory.
You'll notice that .IN and .CN dominate the map. I actually had to size them down somewhat to accommodate the layout. Not surprisingly, .CN is now the world's second most popular domain, next to .COM.
-- John Yunker (email)
Today I learned about the Submarine Cable Map . I was dazzled and wanted to share my discovery with this community. I googled to find this thread and see that Rob Bergin has already mentioned it. However, his Telegeography links are stale. The Internet Exchange Map is at .
-- Tim Chambers 1E4AF729D5CEFFD0 (email)