Summarizing my talk at Wikimania…
 |
0) In quantum mechanics, looking at something changes its nature. Media attention creates just such effects on Wikipedia pages, and elsewhere all over the two-way web. Call it the “Paris Hilton” quantum effect.
0.5) Wikipedia has good tools to deal with individual vandals, mostly based on searching for “bad” strings in text/username and then blocking IPs that create such edits. We also need tools for vandal waves and spin waves, problems we’ll face increasingly in the future as Wikipedia gets more media attention. 1) An vandal wave* occurs when a controversial topic gets hit by a lot of different editors, time-synchronized because they arrive from a media event. For example,
Such waves yield mostly bad edits because of the way the editing software fails when they occur. |
2) You can detect a vandal wave by two simple metrics. 1) The average time between edits by *different* users gets very short. 2) The ratio of edits by IP to edits by registed Wikipedians goes way up. (This isn’t because most IPs are vandals–it’s because a heavy influx of IPs to one page gives you warning that a lot of new users are suddenly seeing that page.) Putting numbers on that–depends on what kind of traffic your page gets normally. (Here I made some arm-waving mention of “derivatives” and even “second derivatives.”)
3) It’s important to respond fast–first because the media event is giving lots of people their first impression of Wikipedia, so you want that impression to be an accurate picture of Wikipedia at its best. Second, because editing software fails in a bad way under such heavy use–“edit conflicts” block people’s thoughtful contributions, while permitting less desirable but faster edits, such as blanking the page, adding an obscenity, or even just correcting one word without realizing a larger problem exists.
4) Wikipedia’s vprotect response should be re-thought as a way to welcome potential new editors at the same time as blocking quick bad edits. For example, the vprotect could include a link to recent page history to show why the vprotect has been added. Also, “failed edits” shouldn’t be dumped and lost–chances are people put thought and effort into creating them. There should be a backup page or two for each controversial article where such “lost” edits get archived so that they can become part of the discussion once the pace of bad edits slows down.
5) A “vandal wave” occurs in response to media attention. “Spin waves” occur in anticipation of media attention, as motivated and paid professional writers try to spin the content of pages. These writers will become increasingly good at hiding their motivation and their identities–we need better techniques to deal with them than outing the few inept ones who get caught.
6) Wikipedia is a resource not only of facts but also of coding solutions that other big interactive websites will be needing in the future. Wikipedia is full not only of words but also of numbers–for example, the timestamp on each edit makes it easy to compute time between edits. These matters will become very important with the growth of “two-way web.” Here at Wikipedia, we saw them first!
* People at the talk correctly pointed out that “vandal wave” was too narrow a description. So with a tip of my hat to Doc Searls, who talks about an “intention economy”, maybe we should call them intention waves. A wave of people arrives at your page–motivated not just to take a look, but to try to play with your software to make the page look more like what they want to see.
There you have the meat of it–my first PowerPoint assisted talk ever–sad that my summary has to leave out the funny parts and fine pictures….