I've been re-reading Ted Nelson's great books, ordered copies for myself and Nicco, who gave his to EchoDitto. They say it's such an eye-opener (knew it would be) that they want to scan it and upload it. I understand the sentiment. I think everyone who uses computers should have Nelson's books and refer to them often.

One of the big ideas in Computer Lib is the bootstrap. Maybe it's the biggest idea of them all. A bootstrap is a process which, in order to create an X, you need one or more X's. The chicken-and-egg conundrum is an example of a bootstrap. Humanity is a bootstrap -- can't make a human without a human (two in fact).

One of the spookiest bootstraps is the process of writing a Pascal compiler (or any language, nothing special about Pascal). You start by writing a very simple compiler in assembly language. Get it working with some sample programs, then start writing a new Pascal compiler, in Pascal, and compile it with your compiler written assembly language. Keep working on it until you have enough features to comple the compiler with itself. Then you can throw out the assembly language one. That story really spooks people, but swear to god, that's how compilers are built.

That little story explains the trick of how bootstraps work. You start with a very simple subset that you know you can use. Then you use it. Then refine it, and use it some more. Repeat until you're ready to use the tool to create itself. In groupware apps, which is what blogging tools are, the bootstrap happens at step one. You always use the tool to organize the work of developing the tool. Same with outliners and BBSes. Basically everything I've ever worked on works that way.

Since my work and Doug Engelbart's have huge amounts of overlap, it should come as no surprise that Engelbart was the guy who focused our attention on the bootstrap concept. And it was Nelson, writing about Engelbart, that made them famous.

Some tools are built without the benefit of a bootstrap. Twitter for example -- but that's just how it appears. They used the results of a lot of research done by others who were bootstrapping.

Podcasting is a form of blogging, think of it as a weblog for the ears, that couldn't have happened so quickly if blogs didn't already exist. Same with almost all the new tools of Web 2.0.

Go back to the web, of course it's a bootstrap. But if you had listened to Ted Nelson, he would have told you that TBL's web is wrong, it's missing the features that Nelson felt the web must have. There's an irony. The teacher doesn't follow his own lesson! What TBL was doing was what he knew how to do, not waiting for the grand vision to appear. What Nelson was doing, with his Xanadu project, was waiting for perfection to emerge from the labs before offering it to the world (and presumably using it himself). He's waited forever. He should read his own book on that subject.

Anyway, to answer the question posed by this piece -- you bootstrap the federated 140-character loosely coupled network the same way you boostrap anything. Let's start with something very very simple, like two tin cans separated with a string. Something that is useful, perhaps minimally so, and can be federated. Then we use it. Then we think. Then we improve it. Start with working code. Start tomorrow, not next month.

So many people want to start by boiling the ocean.

 8/10/2010; 7:12:16 AM.   .