By Harry McCracken | Monday, September 1, 2008 at 12:53 am
You could argue that it’s unfair–or at least unrealistic–to review Google’s Knol in its current form. After all, the Wikipedia-like service just went public a little over a month ago. It takes time to build a build a repository of the world’s knowledge, even if it’s less than comprehensive: Wikipedia surely wasn’t really ready for prime time six weeks after it was launched in 2001. As a Google service, Knol could end up being in beta for years.
On the other hand, as I said back in July, I think Knol is a neat idea. When it launched, it sported an oddball collection of entries that skewed heavily towards covering diseases. I was curious to see how much progress it had made in the interim. So I checked in today…and was startled by what I found. Depressed, actually.
I started by searching for ten topics (selected off the top of my head) that have extensive entries on Wikipedia to see if Knol had anything to say about them:
Buffy the Vampire Slayer
World War I
Knol returned an entry for only one of these, Impressionism–and it was a perfunctory 500-word article with no examples of the art movement. (Wikipedia has four thousand words and more than two dozen illustrations.)
I couldn’t believe that Knol went one-for-ten on my little test. And so I tried again, thinking that I’d done something wrong or encountered some weird glitch. It was then that I noticed a message that appeared at the end of Knol searches:
The message didn’t explain exactly what matches Knol was withholding, or why. But I tried clicking on it–and discovered that Knol did have entries for Buffy the Vampire Slayer, Sarah Palin, and Steve Wozniak. It just doesn’t show them by default when you search for those subjects. Why? I can’t imagine. It’s the first time I’ve ever seen Google do search and do it amazingly badly.
The Buffy entry at Wikipedia is a classic example of Wikipedia being simultaneously incredibly useful and a little scary: It goes on and on and on, and has 119 footnotes. Knol has 123 words on the series, which don’t even mention that it was based on a movie: I know virtually nothing about Buffy, and even I knew that.
The entries on Palin and Woz, however, were much more detailed. Suspiciously detailed, in fact–and they carried the obsessive-but-generic earmarks of the Wikipedia prose style. Could they have been lifted from Wikipedia, at least in part?
Yes indeed. Here are the first two paragraphs of Knol’s entry on Woz:
…and here’s Wikipedia on Woz:
The Knol version of the Woz article was nearly identical to Wikipedia’s entry, but not entirely so: I noticed that it didn’t mention his August marriage, indicating that it drew from an out-of-date version of the Wikipedia piece.
Here’s a snippet of Knol’s article on Sarah Palin:
And the corresponding section in Wikipedia’s entry:
Not identical, but more alike than different.
But the most striking thing about Knol’s entry on Sarah Palin is this: It rightly identifies her as the current governor of the state of Alaska. But except for the cryptic note “Sarah Palin VP of John McCain” at the very top of the entry, it doesn’t mention that she’s also the presumptive Republican nominee for the vice presidency of the United States of America.
Actually, it’s worse than that–it says there had been rumors she might be McCain’s running mate, but that she wasn’t really a contender because of an ethics probe against her:
A tad inaccurate and out of date, no? Especially considering that the entry was added to Knol on August 29th, the day Palin became McCain’s running mate.
Much of the power of Wikipedia, of course, comes from its collaborative nature. And within moments of news breaking such as McCain picking Palin, you can be sure that someone will add it to the appropriate Wikipedia entries. When someone makes a mistake in a Wikipedia piece–and it happens all the time–there’s a good chance someone else will come along and fix it.
Knol is fundamentally different: It’s designed to hold entries written by individuals. “[No] one else can edit your knol (unless you permit it) or mandate how you write about a topic,” states the Knol entry about Knol. Which means that information that’s inaccurate or stale may stay so forever–you gotta think that if the person who added the Knol on Sarah Palin hasn’t gone back to update it by now, there’s a strong chance that he’s lost interest.
Which brings up the “authors” of the two entries which crib from Wikipedia: The Palin one is credited to Sam Goldfarb and the Woz one to Jean Jacques Frapsauce (actually “jean jacques frapsauce”.) It’s possible that these gents contributed to the Wikipedia articles that they appear to have cut-and-pasted into Knol, but they certainly weren’t solely responsible for them. They’re taking credit for the work of others, and because their Knol versions of the entries aren’t editable, they’re not just copies of Wikipedia’s entries–they’re fundamentally flawed copies.
Goldfarb and Frapsauce, in other words, would seem not to be the “experts who know their stuff” who Google wants to use Knol, as explained in Steve Levy’s story on the site. Knol has a mechanism for verifying the identities of those who publish articles in it, but it can’t really identify them as experts. It can’t even stop them from pasting in content from Wikipedia or elsewhere. (Knol has a “Flag inappropriate content” link, but there’s no indication that it occurred to Google that plagiarism might be a problem: You can say that an article contains sexually explicity content or hate speech or spam, but not that it isn’t original.)
Knol also lets users rate the articles, which is presumably one of the “multiple cues that help you evaluate the quality and veracity of information” mentioned in the Knol documentation. But the Palin entry has five stars, and while the Woz one has no stars, it’s not clear whether that means it’s lousy, or just that nobody has voted yet. In fact, Knol doesn’t indicate how many people voted at all, which is a major omission: There’s a big difference between one person giving the Palin entry five stars and fifty people doing so.
Okay, so much for the ten subjects I tried to research on Knol. My luck was so poor that I went back to the Knol homepage in search of better entries. That homepage has five “Featured Knols” up top, the most prominent of which is about diabetic dog food. They’re pretty good. Then there are a few dozen more Knols in a section with the cryptic title “Plain ol’ bag o’ knols.” Is that section made up of popular Knols? Ones selected by Google as good examples? Utterly random ones? I dunno.
I browsed through the Plain ol’ bag o’ knols and noted the following:
–A link to a Knol on The Simpsons leads to a page that says “The requested knol is currently not published,” as did a link for something called “Planning Beyond the Numbers”;
–Knols on “Allah, Does It Mean God?” and “A Reasonable Faith” are religious tracts; I have nothing against ‘em, but they wouldn’t seem to be the “units of knowledge” that Google says are Knol’s purpose;
–”How to Be Successful” is actually meandering thoughts by a recent high school graduate on his career plans;
–”Liposuction and Metabolism: Does the body change after Liposuction?” is mostly what seems to be the outline for a long article in teeny-tiny type, with some promotional copy for a plastic surgery center in Texas at the end;
–Some of the Knols, such as “Step by Step Guide to Employee Satisfaction Surveys” are, at least, not bad. Not fantastic, but not bad.
I really hope that the bag o’ knols is not a spotlight on the best the site has to offer–but such a spotlight would be really helpful.
Oh, and one other note: I searched for a Knol I was actively looking forward to reading: The one that Steve Levy’s article says that Knol founder Udi Manber has written on New Yorker cartoonist Peter Arno. It’s not there yet. Wikipedia’s Arno entry remains inadquate, but at least it exists.
When I first wrote about Knol back in December of last year, I said the following:
“…Google is better at getting things started than finishing them. Services like Google Base and Google Page Creator remain rough drafts at best, eons after they debuted. Even a company with resources as vast as Google can’t do everything and do everything well.”
Knol’s content will surely grow exponentially in the months to come, but quantity is only one issue. Quality needs to get better, too–a Knol that’s filled with swill would be pretty dismaying, and the site in its current form shows that the emphasis on individual authors creates problems that Wikipedia doesn’t have. Basic functionality needs to get better, too: The Knol search engine in its current form seems to be broken, and I think it needs better features for separating wheat from chaff. And I’d give the Knol homepage a major overhaul that helps people find the best Knols rather than featuring some really bad ones.
I still think Knol is a cool idea. If real experts show up and fill it up with the authoritative articles it’s designed to hold, it could be a wonderful resource. But for now, I’m not making any predictions about when or if that’ll happen…