What’s in a knol?

by Eszter Hargittai on December 14, 2007

Henry points us to a new Google initiative and was wondering what I might think about it. I started writing a comment, but thinking that a comment shouldn’t be three times as long as the original post (and because I can), I decided to post my response as a separate entry.

First, I think Kieran is right, knol is way too close to troll, I would’ve picked a different name. (That said, most people out there probably have no idea what a troll is so in that sense it’s just as well although I still don’t like the name.)

I address three issues concerning this new service of trying to create something Wikipedialike within Google’s domain: First, will it gain popularity? Second, what might we expect in terms of quality? Third, what’s in it for Google beyond the potential to showcase more ads?

First, those wondering whether this will gain traction should realize the power of being ranked first on a SERP (“search engine results page”). I know from my own research how often users tend to click on the top result (often!). (Many don’t even know that the topmost link is often a sponsored link even though it says so above to the right of the link.) Of course, one could argue that people click on the top result, because it is the most relevant, which is possible. But check out the results of an experiment with a tweaked condition. Users trust Google (and probably other search engines, pointers to research on others are welcomed!) and it sounds like Google plans to post links to these knols on SERPs likely on top of the list so I anticipate considerable reader exposure.

I don’t have stats on this, but would be curious to know what percentage of Wikipedia’s traffic comes from search engines (and, in particular, Google referrals) vs searches on the site itself. I suspect a lot is from the former. Sure, Wikipedia is known and used by a lot of people, but I would guess most still just turn to a search engine with many queries instead of going to Wikipedia directly so if the Wikipedia links started to show up lower on the SERPs replaced by a similarly relevant (or even just seemingly similarly relevant) page then Wikipedia would start losing audience share.

This brings us to the second point: What about quality? This one is trickier, and certainly not to be confused with relevance or popularity per se, since we know full well from lots of other examples that most relevant or highest quality doesn’t necessarily beat out sites popular for other reasons.

The announcement states the following: “Anyone will be free to write. For many topics, there will likely be competing knols on the same subject.”

So then what? How does Google decide which entry about some topic should be ranked higher on a SERP? Is it based on comments? The quality or the quantity of comments? The rating the entry received? Whose rating? The following important point is not clear from the announcement (either for knol authors or commenters): do people have to be registered with some verified name to rate and comment? Who verifies and how?

Related: whom is the service most likely to attract as potential contributors? Can we expect the most knowledgable experts to contribute? Why might they? The example piece they showcase seems pretty extensive. Is that the best use of my time for some topic in which I am an expert? It might be once there is a critical mass of people using the service, but will there be if the people contributing aren’t necessarily experts? (Sure, I know you don’t need experts for popularity, but I still wanted to raise the issue.)

While the option of commenting is there, whatever is in the core article will get the most attention. I suspect few will care to read the comments, especially if they are just looking for some basic information about something. What this implies is that despite rating and comments, misinformatin could still remain relevant. (I understand that presumably Google will try to figure the ratings and comments into its search algorithm, but how?)

There is so much information out there already (ironically, part of the motivation for all this, isn’t it?). Much of it is very good while, of course, much of it is not. Won’t we see a lot of replication this way? Many people go to WebMD for medical issues, and likely get reliable information. Now such material will be replicated (not word-for-word per se, but in many cases I’m assuming close) on a google.com domain. Sure, Google gets more traffic, but I don’t quite see why the information will necessarily be better than what’s already out there in many forms (dependent on the topic, of course) for the reader.

Which brings us to my third point: because this information will be on a google.com domain, if you are logged into your Google Account (e.g., using GMail) then Google will have more information on what topics you’re exploring, what articles you’re reading thereby adding to the extensive profile they already have about you. Under current circumstances, they know what you search for, but unless the page you follow uses Google Analytics, they don’t know where you go from there. Now they will.

Finally, the whole thing reminds me of Google Base a little, one of Google’s initiatives that didn’t seem to take off too widely. Obviously, knols have all sorts of other features, but the basic idea of putting our content on a google.com domain is there. That one didn’t overwhelm search results so likely for that reason it didn’t seem to go very far. But if Google searches start featuring knols in prominent positions and given the new service’s community aspects that many people seem to appreciate, with the right level of exposure these knols may have the potential to gain considerable popularity.

{ 2 trackbacks }

michaelzimmer.org » Archives » Google Knol: Wikipedia with a Splash of Advertising and a Twist of Web 2.0
12.14.07 at 2:12 pm
"Google Sets Its Sights On Hosting Knowledge" in Disruptive Library Technology Jester
12.17.07 at 10:03 pm

{ 18 comments }

1

chris y 12.14.07 at 2:05 pm

For many topics, there will likely be competing knols on the same subject

And you shouldn’t trust anyone whose knol is in Dealey Plaza.

2

abb1 12.14.07 at 2:46 pm

How does Google decide which entry about some topic should be ranked higher on a SERP?

On the SERP they could link a list of entries (the whole ‘knol’ thingy) instead of each individual entry. Then the question is: how do they order the list? Well, they could have several options for the sort order. But which sort order should be the default? Well, probably the ‘popularity’.

3

robertdfeinman 12.14.07 at 3:03 pm

They have just reinvented the encyclopedia. I have two 60 year old editions of encyclopedias on my shelf. One is the Britannica from 1940 and the other Grove’s dictionary of music.

Reading some of those articles now is amazing. The prejudices of the authors (or editors) are just so clear, yet at the time they were regarded as authoritative.

One of my favorite examples concerns the huge article on English composer Hamilton Harty (who?). He is now regarded as a minor musical figure, but he got the exposure because he was a friend of Grove.

Wikipedia has the potential to eliminate this type of bias. At least the “facts” will be the prejudices of the majority and not the whims of the site owner. The correction mechanism is also more robust if something of doubtful authority gets published.

4

Eszter 12.14.07 at 3:05 pm

Yes, I thought of that, too abb1: just link to all related knols, but as you say, it still leaves the problem of how to list them on that page then.

So the question remains, how to decide “popularity” or “relevance” as they like to think about what’s of main interest, I think. They could rely on their usual algorithms, but that would ignore comments and ratings.

5

Jacob T. Levy 12.14.07 at 3:07 pm

I don’t have stats on this, but would be curious to know what percentage of Wikipedia’s traffic comes from search engines (and, in particular, Google referrals) vs searches on the site itself. I suspect a lot is from the former.

I suppose most of my wikipedia searches go through google– but that’s in part because I already know the wikipedia entry will be in the first few links on google. So if I’m aiming at wikipedia (or amazon or imdb) I just type the search into the google box in my toolbar and know that I’m one click away.

If the first page of google results started getting taken up with ads, sponsored links, etc, I’d just go directly to the site I was looking for.

6

Tim McG 12.14.07 at 3:13 pm

I’m still rather mystified by how they’ll combine editing (particularly as a “strong community element”) with emphatic authorship. Will edits have to be approved by authors? what about authors that aren’t interested in upkeep of their knols–they’re no longer mowing the grass, as it were? If there’s a good article that needs to be updated, there’s not a quick and easy way to make a correction: one would have to create a whol knew knol–and not a derivative work from the original–in order to make a small but important correction.

7

Tim McG 12.14.07 at 3:24 pm

One further thought: if this spurs improvement on the part of Wikipedia, that would, I think, be the best outcome. Perhaps a way to take credit for one’s contributions (optional) and verify one’s identity (again, optional) and a link on each article to a ‘credit map’ showing who created what, and who edited it, and perhaps even a way for knowledgeable readers to vouch for the content of the page.

8

abb1 12.14.07 at 4:02 pm

I almost always use google to get to wikipedia, typing the word “wiki” in addition to the search words (e.g. bittorrent wiki). Wikipedia doesn’t have spelling suggestions.

9

Sebastian Holsclaw 12.14.07 at 6:35 pm

I use the same technique as abb1. Google is actually much faster at searching for the wiki entry than wikipedia proper, and if it gets to the relevant entry more often. Also if there is a popular wiki that isn’t wikipedia, it will find that without me having to know about its existence beforehand.

10

mpowell 12.14.07 at 6:51 pm

abb1 and SH give good methods for finding data on your own. What’s missing, though, is that this is not what most people will be doing.

The concern is: what happens when the search engine is also providing content?

Normally, if a search engine compromises it system by linking too much to its own stuff (or equivalently, to those who pay it money) people will use different search engines. But Google is pretty big at this point. Can there be a monopoly threat with regards to search engines?

11

vivian 12.15.07 at 1:36 am

Ezster, how many people who click on the first link also go back and click on others? Because, in addition to using Google the way Jacob does, when I’m seriously looking for something, I will click my way through the first few pages (of a hundred links each). Then again, I’m not the average user. It’s a lot more reasonable to check the first link – even the first sponsored link if I’m shopping anyway – than it is to rely on it. How did you check the users’ choices against the adequacy of the content – how bad was the first hit, and how bad would it have to be before they checked another?

12

Shane 12.15.07 at 1:47 am

Wikipedia is likely to be fine from this; in fact, Google’s action seems a direct response to Wikipedia’s refusal to monetize.

It’s Citizendium that’s just been irrevocably murdered. Not that I think it ever had much of a chance at success, but it seems that Google’s formulation is a far superior way of getting to where Citizendium was aiming for.

If having quality content that’s easy to get to is a good thing, then this is a good thing.

13

Karen Carr 12.15.07 at 6:33 am

Some points I haven’t seen raised anywhere (yet):

– Wikipedia pages are extensively hyperlinked to other Wikipedia pages. Will people writing knols link to other knols? Will they go back and add more links as their first ones get broken? Or will it be like reading a newspaper article, mostly dead text?

– Will there be any way to rate a page child-friendly or not? Much of the research done on the Internet is done by children for school.

– Will Christian anti-evolution sites all float to the top of the rankings as thousands of Christian believers vote for them? Democracy has never been known for its ability to promote accurate scientific inquiry.

– Will Google encourage knols on any topic, including the ones Wikipedia has decided to discourage, like knols on your irritating next-door neighbor, or knols on your kid’s new rock band?

– Won’t most people just take Wikipedia pages and repost them on knols, just like many web encyclopedias do now?

14

MNPundit 12.15.07 at 7:42 pm

People really click the highlighted link? My eyes naturally skip over that as an add. I had to do a quick search just now to check to make sure I wasn’t clicking the sponsored link–and I wasn’t. I never do–because it’s always trying to sell me stuff.

To me the first google result is the one in white right under the highlighted link.

Wow.

15

MNPundit 12.15.07 at 7:43 pm

Bizarre, sometimes the first link is the pay link even if it’s not highlighted. I still never click on it though….

16

FuzzyFace 12.16.07 at 11:31 am

Third, what’s in it for Google beyond the potential to showcase more ads?

That’s sort of the point, actually. People still think that Google is a search-engine company. They’re not. They are the internet equivalent of a television network – their entire business model is about providing content into which ads can be placed. If they can displace Wikipedia, they gain enormous numbers of eyeballs and ad revenue.

17

David Gerard 12.16.07 at 12:52 pm

At Wikipedia, we were mostly enormously pleased that the mockup PNG includes a CC-by licence tag. IF, I say IF, they require knols to be released under a proper free content licence, then that’ll be a big win for everyone. (And we’ll actively tell people what a good idea it is.) Wikipedia’s not actually about running a horribly popular and expensive website, but about writing stuff that can be copied and reused and adapted and spread; if other people are doing that, that fulfils our aims without us having to actually do the work ;-) Same for Citizendium, who I really doubt are going to pack up and go home any time soon. More freely-reusable content is a big win for the world.

18

Eszter 12.16.07 at 1:54 pm

MPowell – Regarding monopoly threat in the domain of search engines, I’ve been writing about related issues for almost a decade now.:) While a pure monopoly per se is unlikely, there is definitely the potential for one or a very few having a lot of say in this domain and thereby having a lot of say over what access is realistically within the reach of most users.

Vivian – That paper I link to addresses the issue of relevance, I think. I didn’t do that study so I don’t remember the details. But yes, of course one would have to see whether #10 is as relevant as #1 and if it is then it’s less of an issue to have people relying on it. As for clicking through to several following pages, I think it is a very rare event.

MNPundit – Definitely a lot of people click on those links and in some of my studies I’ve started finding out whether people realize that those are sponsored links, and it turns out that many do not.

Fuzzyface – You may be right that I assumed too much when I didn’t elaborate on the centrality of ads in Google’s business model. I figured the part about being able to display more ads through this service was fairly transparent so I didn’t say more here. My main point in this case was to note that beyond that central and fairly public aspect, there is an additional reason why this may be of benefit to Google.

Comments on this entry are closed.