An article in yesterday's NYTimes describes YouTube as the next great search engine. As its primary evidence, it presents a 9 year old boy who used YouTube to do research for a paper on the platypus insetad of Google or Yahoo. Key quote:
“When they don’t have really good results on YouTube, then I use Google,” said Tyler, who is 9 and lives in Alameda. Calif.
YouTube's explosion of popularity as an information retrieval mechanism among young users is not particularly surprising - this is a demographic that lacks literary skills, after all. In The Vanishing Newspaper, Phillip Meyer gives a table of the average reading level of various newspapers. The vast majority are at a ninth-grade level. Wikipedia articles generally require a college-level education. (Unfortunately I can't find the source of that statistic, which I read some time ago, but the amazing random wikipedia article flesch-kincaid grade level guessing game provides strong anecdotal evidence.) Given that, of course young researchers are turning to visual media!
The first result I get when searching Google video for "platypus" is an excerpt of a National Geographic program -- heavily dramatized, with sound effects and a strong "show, don't tell" dynamic. The narration is relatively slow-paced and easy to digest, since it doesn't have to delve into obscure descriptions better conveyed by images.
It is not clear that YouTube's rise as a research tool is due to the quality of its information, but rather its availability (or accessibility). Wikipedia experienced a similar phenomenon in its infancy, as many people doubted the accuracy of it's articles, but were more than willing to check them out because it served as a one-stop shop for all research needs. Now, a more mature Wikipedia has gained acclaim as being more accurate than Encyclopedia Britannica in addition to having articles spanning nearly every conceivable topic.
As YouTube - or its eventual successor - similarly matures, video on the web will become even more standardized and useful to researchers who might otherwise go to Wikipedia (or a text based source like a journal). Text had a natural advantage in that each word embodies its own meaning, making hyperlinking and semantic connections relatively easy (i.e. the basis of Wikipedia's hyperlinked dictionary). Video is somewhat more difficult, since any semantic linking requires a fair bit of statistical analysis and latent semantic processing. Also, the cost of choosing the wrong source is much higher with video. With text, it is easy to scan a page and determine whether it has the information desired, since a page can be read (somewhat) achronologically. Video is much more difficult to scan, since it depends heavily on linear storytelling.
The next revolution in video search will be the ability to evaluate videos without that linear constraint. Until then, Wikipedia articles will still be able to embody many different reading levels at once (despite their high average, an article might start simply before delving into obscurities) whereas video is more likely to be set at the lowest common denominator, or whatever comprehension level "fulcrum" garners the most views.