User:Smack/A rewrite a day keeps the critics at bay

This is a personal essay on a topic of general relevance to Wikipedia.

It was recently brought to the attention of those who read the Village pump (news) that the Tech Central Station has published a criticism of Wikipedia, written by none other than Robert McHenry, former editor-in-chief of the Encyclopædia Britannica. The article is presently available here. It's rather interesting reading, at least for anyone who has given much thought to the benefits and pitfalls of the wiki way.

The reason why I bring this up is that McHenry raises, among other issues, one that I've been contemplating myself. If you do not choose to consider this issue important just on the strength of my opinion, you may find McHenry's opinion to be rather more weighty.

The mediocre article disguised as an excellent one

The issue is what the former editor-in-chief calls "editing into mediocrity". It is the process by which a perfectly good article, by means of piecemeal revision, is turned into a jumbled, muddle-headed, possibly factually incorrect, and generally poor article. If this style of editing can ruin a good article (as it did to the one that McHenry cites), we certainly can't expect it to improve one that isn't any good to begin with.

The Style and How-to Directory advises: "Integrate, don't append". Anyone who has used the diff ("Compare selected versions") feature enough has seen plenty of anecdotes, trivia, and otherwise important facts of limited relevance tacked onto the end of an article, as though by way of aside, rather than set carefully into the appropriate section. Granted, most Wikipedists do not write this way, but I would claim that most piecemeal additions are little better. They often disrupt the flow of the article and the chain of reasoning intended by the original author of whatever passage they are stuck into. They're also often inserted into an article after only a cursory examination, which is not enough to determine the proper place to put them in. Furthermore, they may needlessly duplicate facts presented elsewhere.

Some most unfortunate articles are built entirely this way. Granted, they contain all of the constituent parts of an article, but they are not really articles. The situation is like what would occur at a barn-raising where one good farmer has brought plywood, another lumber, another shingles, another nails, and another a large tub of red paint, but none stayed on hand to actually build the thing.

What to do?

A single piecemeal change is only a minor threat, but when committed en masse, they can severely undermine the effectiveness of an article, even if they increase its information content. The least painstaking way to rectify such a damaged article is to rip everything out and rewrite it from scratch. If you've never done this, fear not – it's not as hard as it may sound. Consider, for instance, the article on exponential decay. As of November 19, 2004, I was responsible for virtually every word on that page. However, during my recreation of the article, I added naught but a shiny derivation coded in TeX. I merely took the incohesive ramblings that were already there, compared the information they contained with my own limited knowledge of the subject, and produced something that (I flatter myself to think) makes a bit of sense and is logically arranged.

It would be wonderful if every article could be overhauled in this manner once in a while, and then carefully monitored with the aid of a watchlist to keep new additions in line. Unfortunately, this task can only be performed by someone with a bit of expertise. Expertise is a very unevenly distributed wiki-commodity. Consider history, for example. It comprises a vast part of human knowledge and experience, yet it is for the most part the arcane preserve of a few researchers. While the Wikipedia is likely, in due time, to amass a number of worthy articles on some topics of common interest, as well as (sigh) topics of geeky interest, it will be impressive indeed if it ever becmomes a comprehensive resource.