Knol has a couple features not available in Wikipedia. The most significant of these, I think, is article ownership. Where Wikipedia specifically discourages people from maintaining their preferred version of an article against all efforts to change it, Knol allows this as an option in creating an article. Paired with this is that Knol articles are not unique; if you don't like an existing article on Pokemon, you can write your own rather than setting out to change the existing article. (The implications of this on edit warring are obvious.)
Another feature is article ratings and reviews. You can very easily praise or pan a Knol article by clicking on a 1-to-5 star bar, and a cumulative rating is kept for each article. The rating of an article is shown whenever it comes up as the result of a search, so you can use the information to help choose between different articles on the same subject. It's unclear whether the rating affects the order of search results.
Each article also has space for reviews and comments. There do not seem to be any limitations on what qualifies as a 'review' as opposed to a 'comment,' and most reviews that I have looked at are just people saying they liked the article. I'm not sure of the degree to which article owners control what show up as "reviews" as opposed to "comments."
I haven't much cared for navigating around Knol. There is not yet very much linking between articles (given that Knol is a Google undertaking, presumably such linking would affect search results, but who knows?), and a painful amount of what linking there is goes to Wikipedia articles. A lot of Knol articles at this point are fairly transparent porting-overs of Wikipedia articles. There are also no functioning equivalents in Knol, yet, to Wikipedia's "categories."
A lot of the implication of Knol remains to be seen. There are not that many articles in Knol yet, and more glaringly there is not a lot of interconnectivity between the articles that do exist. This may change as Knol grows.
It also remains to be seen how the differences between Knol and Wikipedia will play out. I have no doubt that if Knol takes off in popularity, efforts will be made to make sure that the "right" information gets promoted, as they have been in Wikipedia. It is unclear how this will work, though. Presumably, the ultimate goal will be to bury "bad" information in search results, given that there seems to be no way to monopolize content control. Whether this will be as easy as bombing articles with poor ratings or will require building and linking competing articles is unknown to me.
The writing in Knol seems, so far, to be better on average than that in Wikipedia. It may just be that the high school kids have not yet discovered Knol. Un-owned articles, such as all of Wikipedia, have the potential advantage that people can do drive-by grammar corrections and the like, which cannot generally be done in owned Knol articles. The effect of this advantage in Wikipedia is not very impressive, however.
In any case, I think Knol has a very distinct advantage in how it allows for edit warring to be diffused: if you don't get your way, you can create your own, competing article. If nothing else, this will tend to put an upper limit on the amount of energy that is spent edit warring, whereas edit warring probably consumes an amount of effort at Wikipedia that rivals and possibly exceeds the effort expended at creating and improving articles. For this reason alone, I think that Knol will eventually start eating into Wikipedia's current position.