I really need to get to Paris again. It's remarkable to me how a technical article analyzing how people think about Paris can make me nostalgic for it. Especially since I've only been there about 48 hours.
I've played the mental map game with people before. It's a great way to get a group of international folks to talk together: ask people to draw cocktail napkin maps of their homes, or places they love, and listen to them as they describe what really matters to them in a city.
So all that being said, I like this article a lot. It's a really clever set of investigations, interesting methodology as to what questions they think to ask. Their own visualzations are fun, too, particularly figure 8.7 (the "molecules") which overlay geographic and psychological connection in a very clear way.
A lovely little article. Note the date, October 1995. Fairly early for this kind of writing, when the ideas were fresh. I think he mostly got it right, except....
This article makes me question the value of "map" as applied to the Web. Maps are fundamentally pictures of two dimensional structures; land surface. The Web is non-dimensional. Trying to force a 2d structure onto the Web is doomed to fail. For instance, take a look at the awful map he presents as an analog of a tribal map for the current Internet as a whole. Ick! Does this tell you anything other than that there's "the Internet", and then a few specific pieces of it based on particular protocols? This doesn't help anyone conceptualize things.
His article ends with some hopeful notes about "rationalizing" addressing and maybe building maps out of that. No way. It hasn't happened and it isn't happening now. Furthermore, to do that would require centrally structuring the Internet much more than it is now, and to do that is anathema. The Internet only works because it is decentralized. Map-makers want one nice bird's eye, centralized view of things, but it's just not going to happen.
On the other hand, I do like the idea of conceptualization of the structure of Web or the Internet as a whole, expressing people's personal views of how they navigate the whole mess. But in my experience this conceptualization is organized into bookmark files, or hierarchical structures like Yahoo, or free-text search engines. Flat visualizations just haven't proven themselves as being valuable. Either the problem is really hard, or the metaphor is just wrong.
This article is undated. However, the HTTP transaction shows it hasn't been modified since 29-Jan-96, so it's pretty much from the same era as Staple's. Interesting how all this reading is fairly old, when the web was new. Is there less current work about mapping the net, or is it just harder to find? I think it's telling that none of these ideas have played out.
"Getting lost in hypertext" was a fashionable concern a few years ago. Is it one anymore? I don't feel lost on the web, just like I don't feel lost in a library. I do have a hard time finding what I'm looking for sometimes, but it is easier on the web than a library. Thanks entirely to search engines..
The metric his work is based on is flawed. It presumes the only way to get from one page to another is to follow links. In practice, people don't navigate the web this way. They go to search engines, where everything is right next door.
Um, what do these maps mean? I appreciate trying to algorithmically crunch his measured topology into 2.5 dimensions, but when I look at the pictures I see nothing.
I don't mean to be too harsh on this paper, and the previous one. If I had more enterprise and time a few years ago, I would have produced similar things. But looking at them now, I'm just overwhelmed with how hard the problem really is. Trying to find an objective way to map the Web doesn't help, because the objective shape is not what's interesting about it. We need some sort of fuzzy, subjective, conceptual mapping.
Hah, so now you've been nasty, where's your sketch of a web site, big man?
I tried to draw a conceptual map of how I think of the Media Lab site. Black hollow circles figure most heavily - these are the group pages for things like "Agents" and "Necsys". Blue filled circles are secondary; these represent individual user's home pages, the users I know best. Black lines show links between two major groups. These are directional, and tend to flow from the middle outwards. Blue lines show links between people and projects, and orange lines show links between people. Some lines are dashed, indicating a tenuous connection. All links are conceptual first, and might or might not actually have hypertext links.
The map as a whole is quite amoprhous. Some structure is visible - Necsys is cut off from the rest, the main site is the octopus in the middle, and people are scattered all over the place, with me in the middle.
I had originally intended to geographically structure this according to the physical layout of the building - put each groups pages where they are in the building. But that really requires a 3d map to do right, and anyway it makes it hard to group things together conceptually. It's also a sort of negative map, since I think the rigid structuring of where people are in the building isn't healthy for our research community.
| Nelson Minar | Created: November 18, 1998 |
| <nelson@media.mit.edu> | Updated: December 15, 1998 |