Chris Dent replied to my comments on purple numbers. I got caught up with other things last week, so the delay is mine, not his.

Essentially Chris casts the issue as a trade off between reading and writing.

That sounds about right to me. Ward Cunningham also noted that wiki optimizes writing at the cost of reading.

If you think about it, that's true of a lot of the "bottom-up" / "worse is better" / "democratic" technologies which have become prominent. The web itself, with its ease of creating pages and links, but the tolerance of broken links and initial lack of any inbuilt "findability" strategy. Weblogs which allow anyone to publish, without constraint, but leave the reader lost in a morass of un-fact-checked, "untrustworthy" opinion masquerading as fact. Wiki (and particularly wikipedia) which enable anyone to contribute information, but do little to prevent malicious misinformation.

Optimizing writing rather than reading is a good strategy for a platform (or technology), because if you get a lot of "writers" or builders on it who create some value, this in turn is an incentive for someone else to solve the problems of the readers : search engines solved some of the problems implicit in the web's architecture. Blogrolls and Technorati and tagging solve the problem of finding good and appropriate blog entries. WikiMinion solves the problem of wiki-spam (more or less).

On the other hand, if you optimize for reading, while making writing more difficult, you have a challenge. As Chris notes, there's a hope that eventually technology will rescue you and tools will appear. But in the meantime, the platform is built out more slowly. And there's a danger that it never gets built at all.

I'm not sure what it says about the purple numbers concept that it optimizes reading rather than writing. Or that it's a little enclave of reader optimization embedded in the larger writer-optimized world of wiki. That may be a strength or weakness. Are purple numbers like tags or search engines : a supplimentary technology to improve the readability of a medium that started out optimized for writing? Or are they more like the semantic web : something which would be wonderful for readers if applied, but may never be.