Class 2

From The Web Difference
Jump to navigation Jump to search

David Weinberger took us on a tour of thorny definitional problems in Class 2. The point of the class, on one level, was to draw lines: what do we mean by the “web”? How does that differ from what we mean by “the Internet”?

The readings are all classics: Reed/Saltzer/Clark on End-to-End Arguments in System Design; Isenberg’s Rise of the Stupid Network; Searls/Weinberger on World of Ends; and Tim Berners-Lee on Weaving the Web (chs. 1 and 4; not online; you have to buy it).

The web, David argued, is a standard. It’s a linking standard. It allows for hypertext links, which are highly decentralized. It mediates humans and technology.

What is the web, if not a standard? Some responses from the class:

- It’s a cultural phenomenon; a thing that people do.

- It’s a sum of pages that appear in the web format.

- It’s a sphere (DW wanted to know: is there an implication of incompleteness?).

- It’s a communications medium.

- It’s a marketplace.

DW also talked us through issues related to HTML 5. He surprised himself (and all of us, who know enough to be surprised, I suppose) by saying he actually likes the proposed HTML 5. The people who write standards like to make it easy for computers to understand what the standard has allowed the user to include. But that is not the same as making a system easy for human beings to use. HTML 5 adds a bunch of new tags, anticipating what people want to do with it. But what does this have to do with end-to-end? One of the points of e2e is that you don’t anticipate what people are going to want in future. David still kind of likes it.

David got some hard questions back from the class. One line of reasoning: when we talk a lot about “centers” of the network, which one apparently wishes to be stupid, how do we know if a part of the complex network is in fact “center” or “edge”? And, separately, in World of Ends — a polemical piece — it appears that he and Doc prefer a network with minimal control. As one student contended, from time to time, control would be required on the network — if not at the center of the network, then at the ends. Who is to say (cf. Zittrain) that control at the ends is better than control at the center? And how do we know that minimal control is a good idea anyway?

We ended up getting into regional differences in control in Class 3, on Monday, when we took up the work of the OpenNet Initiative.