Wiki set points

Imagine putting the Feynman Lectures on Physics up for public editing on a wiki (Feynmanpedia). Would they get better or worse?

My immediate gut instinct is “worse”. However, when I posed this question to a colleague I greatly respect he asked me pointedly if I’d actually tried it. It’s a good question. There’s no doubt that with some wiki communities, perhaps most, the Feynman Lectures would rapidly deteriorate in quality. But maybe with the right community they’d improve.

For many wiki communities there’s a useful notion of a “set point”, a quality level that an article written by that community will converge to over time. For a poorly written article, most edits will tend to improve the article, and only a few will make it worse; thus, the article will improve over time. However, for a superb article, many of the edits, even well-intended ones, will make the article worse, and so the article will get worse over time. The set point is the quality level at which edits improving and worsening the article balance each other out.

For Wikipedia the level of the set point is moderately high. I’m pretty sure that if one took a section out of the Feynman lectures and put it up on Wikipedia, it would get worse. On the other hand, if the community started with a blank page on physics, it’d demonstrably get a lot better.

For other wiki communities the set point is different. I’m a fan of the TV show Lost, and there is an amazing fan-created wiki about the show called Lostpedia. The set point of Lostpedia is quite a bit higher than the Wikipedia set point.

The idea of a wiki set point is obviously imprecise. Indeed, any idea that deals with quality judgements and community action necessarily will be. The caveats that need to be applied include: the set point will be different for different articles; even for a given article it will vary over time as contributors change; what does it mean to speak about the quality of an article, anyway; surely it makes more sense to talk about a set quality range, rather than a single point; and so on.

Despite these caveats, I think the set point is a useful way of thinking about wikis, and stimulates many useful questions. What types of wiki community or wiki design increase the set point? What types decrease it? How high can the set point go? How could we design the wiki software and community so that the set point is above the level achievable by any single human being?


  1. Interesting.

    Putting the philosophy of “quality” aside, I’d say that the set point is still largely determined by by the article guidelines ( and how strictly they are enforced. Wikipedia honors the ideal article in the Feature Article of the Day in regards to their guidelines.

    For another example of a different set point, look at Simple English Wikipedia:

    Perhaps a will also one day exist and the set point which will be similar to that of Lostpedia (which I frequent regularly!)

    Another interesting aspect of the Feynmanpedia example would Wiki-Meiosis. A process I have just coined which refers to division of a large article into 2 smaller wikipedia articles. It would be interesting to see if, starting from a large body of text, the Feynmanpedia would naturally reduce to a state similar to that of the current Wikipedia.

  2. Interesting post. In addition to quality, one should consider the intended audience. If you look at the discussion pages for some of the more technical articles in Wikipedia, you can see there is much debate about the level of background knowledge that a writer can assume when writing an article. It may be difficult to separate these two elements. The articles that assume more background knowledge may appear to be higher quality. Wikipedia writers tend to converge on a consensus of relatively low background knowledge. If the Feynman Lectures were posted on Wikipedia, perhaps most of the changes would be adjustments for the assumed audience, and the quality would change relatively little (if one could separate the quality level from the assumed audience).

  3. Hi Peter – I don’t think it’s a question of assumed background, at least, not in the way you’re suggesting. My own experience is that the more technical articles are of lower quality, perhaps because the number of people willing and able to make meaningful contributions is much smaller. The article on “Britney Spears” is a lot better than the one on “qubit”, for example.

  4. The article on “Britney Spears” is a lot better than the one on “qubit”, for example.

    Maybe an expert on Britney Spears would judge that the Qubit article is much better than the Britney Spears article? As a Britney Spears layperson, perhaps you are not qualified to judge the quality of the Britney Spears article? As a Qubit expert, perhaps you are more critical than most readers?

    There are many interacting factors here: the expertise of the person who is judging the quality of the article, the assumptions that the writers make about the expertise of the readers, the willingness of amateurs to make edits when the readers are assumed to be laypeople versus experts, the difficulty that experts have when writing for laypeople, the size of the readership, the size of the pool of expert writers, etc.

    I think there are some lessons to learn from comparing Wikipedia to a more technical wiki, such as the wiki of the Association for Computational Linguistics. For a writer, it makes a big difference whether you imagine your reader as a layperson or an expert. (Of course, I can see that you know this, from your post about your book on The Future of Science.)

Comments are closed.