Wikieverything

You all know good ol' Wikipedia, but there are also Wikibooks. As you browse through the books at that site, you'll notice two things: they aren't very complete (most are half an essay at best), and they aren't very good.

I'm not going to talk about the reliability and authority issue which seems to dominate most discussion of wikimedia. Personally, if I'm reading to get a lite intro to a subject of which I'm ignorant, I'll take Wikipedia as gospel, because it doesn't matter; if I'm working on an academic topic, then I'm not going to cite an encyclopædia of any sort, but will have my own external sources providing detail. You no doubt have your own sense of what is or is not reliable.

Instead, I'm going to talk here about why the deck is stacked against wikibooks and other attempts to apply the open source idea to every field of endeavor.

Narrative vs reference

The average entry on Wikipedia is between a single line and a few pages long. They have limited narrative depth at best, and generally just cover a simple list of facts. Although wikipedia would be thousands of pages if printed out in its entirety, nobody is expected to have edited anything beyond a sliver, and nobody expects it to have any structure beyond alphabetical order.

Computer code is much like this: a person working on one subpart of a program doesn't have to know anything about how the other subparts work. To write a translator, Jane can work on text parsing, Joe can work on a set of dictionaries, and Jess can work on the clicky interface, and all can work with little regard to what the other parties are doing.

Narrative works don't have such wonderful compartmentalization. Sure, there are chapters, but if the chapters don't tightly come together, we won't like the darn story much.

You ever play the Exquisite Corpse? You fold a paper in thirds, and draw a head, and then refold so the head isn't visible and hand the paper to a pal who draws the torso, and then your pal hides the torso and another pal draws the waist and the legs. Then you unfold it all and laugh about how delightful such a disjointed figure could be. If you were lucky enough to be one of the founders of the Surrealist movement, then your drawing will wind up on the wall of the Art Institute of Chicago (see figure). But for the rest of us, the game is a fun brainstorm but ain't a final work.

Anthologies are common enough, but we often call them edited volumes and put the editor's name on the cover to remind the reader that somebody sat down and made sure that the elements somehow cohered.

We're used to other aggregate works directed by one individual: pop songs that have a single producer and movies with one director. I leave to the reader the debate over the quality of songs written via jamming with the band versus songs written by a single composer.

OK, there's your survey of media. Painting, sculpture, movies, music, novels, all involve one or a small number of people directing a final product, which may have been touched by dozens of hands. This is not surprising, and I don't think anybody seriously expects wikipainting to truly surpass the old method.

But textbooks. There seems to be a serious belief that a textbook can be collaboratively written by a committee. This is not a new wikiconcept. In elementary school, we all had many a textbook with no author or editor on the cover, and a list of committee members on the title page.

We often refer to a subject like math or biology as a field. Picture a big expanse of plain, in which you could take any direction. When we go to school, we take courses--carefully guided paths through an open expanse. In other words, a good textbook goes somewhere. It is a narrative.

Conversely, some textbooks attempt to survey the entire field at once. Such books are frankly no longer textbooks, but are rightly called references. They have their place, but it ain't teaching. I can see the appeal for the textbook writers, who want to maximize their market share. They provide as much material as possible in the hopes that the teacher will select a course through the material; some teachers do, covering only chapters 1.3, 3.8, 8.1, and 16.4, while others wind up ploughing through the entire field, column by column. [If you are reading this in book form, I've put effort in to cohering the essays into something of a few narrative threads. Really.]

The wikimethod is good for writing references but bad for writing narratives, so the deck is stacked against wikitextbooks. Again, like the encyclopædic texts, they have a valid and valuable place on the e-bookshelf, but they can't replace narrative works, just as (conversely) we wouldn't read a single narrative and claim that we understand the entire field.

I built it and nobody came

Of course, it's not so simple. The real success stories in open source come from a single good idea, some good coding, and lots of good advertising and self-promotion. Of course, it also helps if your program is about porn.

Out of a thousand readers, over 990 won't fix so much as a typo, and a handful will make little ten-second fixes on a single equation or such. If you're lucky, maybe a single reader out of every few thousand will contribute the significant time investment to contribute a narrative. Open source provides a new alternative to finding and coordinating coauthors, but it's not particularly a revolution over existing coauthoring tools (diff, revision control, those cute little change tracking features in word processors). The fact remains that a narrative is best written by a small number of people in close communication.