Intranet Journal
The online resource for intranet professionals
Y'all know the short subject "Powers of Ten"? If not, you must have been sleeping during the science class when the AV Squad rolled in the projector. It's a great film, depicting the universe seen at 10x increments, starting with you standing in your backyard in New Jersey. It goes all the way from the Solar System to the Galaxy to Larry Ellison's Ego to Super Gigantic Meta-Galaxial Clusters with Caramel and Nougat ... and then goes into reverse through molecules, atoms, sub-atomic particles, and footnotes to a senior thesis in logical positivism. Looking at maps of the Web has something of the same effect, except the scale runs according to the level of abstraction, not distance.
Mapping the Web is a huge field that falls into two main pieces: maps that show us something interesting about the Web, and maps that help us navigate the Web. These two need not be essentially connected.
We begin with Bill Cheswick's work:
www.cs.bell-labs.com/who/ches/map/index.html. He's a security guru at Bell Labs and a wacky guy. He and a colleague, Hal Burch, put together software that draws a beautiful map of the routers that are (as he says) the tin cans and string of the Web. It puts these 100,000 nodes together using colors to indicate congestion and arcs instead of boring old straight lines. The result looks like a fireworks display. While this map has a geographic correlate -- you could position the routers relative to a geographic map -- Ches declines to follow it. Rather, Ches is fond of pointing to a dense cluster of nodes, saying, "That's UUNET" or whatever. (PS: Try out Ches's non-optical illusion page on the McCollough Effect: www.cs.bell-labs.com/who/ches/me.)
John Quarterman
(http://order.mids.org/~jsq/index.html) has been mapping the Internet in various ways for over ten years. For example, he plots the growth of Net hosts in the U.S. on a geographic map of the U.S.(www.mids.org/mmq/604/pub/us.i.gr.c.html) or the presence of hosts plotted on a world map (www.mids.org/mapsale/world/). Quarterman's group also produces the Internet Weather Report (www.mids.org/weather/index.html), a set of maps that display the speed of the Internet ("latency"). Every four hours, the IWR "pings" several thousand servers world wide and measures how long it takes the packet to do the round trip.
If you want to see the geography of a hyperlink, you can use a utility such as neotrace (www.neotrace.com/). Give it an URL and it will plot on a world map exactly how your request to see that page has been routed. While it's intended to help you diagnose problems reaching sites, it also gives you a warm feeling about just how global -- physically and literally -- the Web is, and also, on occasion, just how bone-headed stupid it is. (You can see a dynamic map based on traceroutes at http://home.online.no/~ggunners/NetBird.html.)
But the Web isn't merely hardware. Upside recently ran an article by Robert Buderi on Ravi Kumar's charting of 200 million pages and 1.5 billion links to and from them. (He's backed by Altavista, Compaq and IBM.) His initial finding is that there are four main regions on the Web, each with about 50M pages. Sites in the Strongly Connected Core are no more than 7 clicks from one another. Sites in the In region point to the Core but the Core doesn't have the good graces to point back. Similarly, the Out region is pointed to by the Core but doesn't point back. Finally, there are the Tendrils which run off of the In and Out regions but can't be reached by the Core, presumably because they're behind firewalls. According to the article, "Kumar says, despite its ad hoc creation and constant evolution, it seems the Web is actually highly organized." This is a weird type of map, sort of like clustering streets not by how they're connected but by whether they're one-way or two-way.
Valdis Krebs (www.orgnet.com) is an expert in organizational network mapping. Rather than doing simple org charts, he aims at creating dynamic maps of the various types of structures of an organization, including the flow of information. He has created a map of the Internet industry, showing the relationship of the various Net players: www.orgnet.com/netindustry.html. This map obviously is hugely dynamic, and would need to be updated with every press release.
So far, none of these maps is intended to help you find your way on the Web. If you want help navigating, you need to decide among the Web's three basic information structures: random hyperlinks, clustered hyperlinks, and hyperlinks organized into browsable hierarchies, a la Yahoo. Each can be presented as a map.
You can see an example of Yahoo's attempt to render its hierarchical directory as a geographical map at www.cybergeography.org/atlas/yahoo3d_large.jpg. It's no longer available, possibly because the geographical view of a hierarchical directory defeats the purpose of such a directory, which is to show a whole lot of information in a very small space.
There's a mixed mode of clustered hyperlinking that's extremely common: every Web site can be presented as a set of pages all linked (eventually) to a home page. Frequently, maps of this structure show the home page in the center with the other pages linked to it and to one another. Many Web utilities show this to you in your role as webmaster.
There have been lots of attempts to make random hyperlinks viewable and navigable. This isn't just a Web problem, however, so work has been done in lots of fields, from semantic mapping to document management. The Brain, for example, -- a utility that some people swear by but which is to me of no appeal -- helps organize scraps of information; Mappi.Mundi (http://mappa.mundi.net/map/), a site about mapping, has an example of the brain linked on its home page that maps the ideas on the site. ThinkMap (www.thinkmap.com) doesn't confine itself to the Web but it does provide a Web utility which you can see at www.bacardi.com. Be sure to click on the ThinkMap button on the bottom left. You will then see a dynamically updated set of maps, presented as circles and lines and mysterious symbols, that are supposed to help you visualize where you are. For me, they only simulate one too many pina coladas. Inxight (www.inxight.com/), a spin-off of Xerox "Kiss of Death" PARC, for years has been trying to convince people that they want to think about information as a spinning ice cream cone (oops, I mean "hyperbolic tree") and other oddities. The demos are very cool (www.inxight.com/products_wb/tree_studio/ tree_studio_demos.html [UNWRAP]), but are they useful? Not to me, but, then, the right half of my brain was removed after a series of unfortunate homework assignments in my freshman Drawing the Nude course in college.
One of the hottest areas is clustering sites based upon an analysis of their content. The wildly overhyped Autonomy engine does this (www.autonomy.com), as do others, some commercially available (e.g., www.fulcrum.com) and some in research labs around the world. Putting related sites together visually even if they are not linked is a way of manufacturing serendipity. The hard part is figuring out what's related, but you shouldn't underestimate the difficulty of coming up with a visual metaphor that works. There are already 2-D maps, browsable hierarchies, and some mappings into 3-D space. (Won't someone please adapt the Quake III engine for this purpose? Thanks.) [Note: I'm on the board of a company still in stealth mode that's involved in one of these areas.]
There will be many solutions to this problem, and which ones we like will have everything to do with our personal way of thinking and the type of problem we're trying to solve at the moment. But navigable maps of Web sites clustered by relevancy to our interests are of unique important to the Web because the Web space is itself organized not by uniform units of distance but by *interest* itself. Distance, on the Web, is measured by irrelevance. Navigable maps capture this essential fact of our new world, and thus not only map Web distance but conquer it.
---
Here are two sources which were helpful to me: http://mappa.mundi.net/maps/
http://www.cybergeography.org/
David Weinberger writes JOHO and is one of the Ringleaders of cluetrain.com,
a manifesto of web ethics. He also provides strategic marketing
consulting to high-tech companies, writes for several magazines
(including Wired)
and is a commentator on NPR's "All Things Considered."
He was, as VP of Strategic Marketing, one of the shapers of Open
Text's intranet strategy. David sits on several conference boards
and is a member of AIIM's Emerging Technology Advisory Group. Reach
him at self@evident.comThe Author
