Tuesday, July 29, 2008

The Real Problems With Knol

I'm a big fan of Knol, and I seem to have been spending an undue amount of time defending it from critics who, I believe, have the wrong end of the stick.

But. Just because most people are judging Knol using an inappropriate comparison, doesn't mean it's perfect.

So what could Knol be doing better?

Simple Internal Article Linking
It's pretty common for the specialist topics featured in Knol to reference terms that are unfamiliar to readers, but out of scope for the current article. Everything2 and Wikipedia have shown that jumping from topic to topic is compelling. Knol needs a simple mechanism that lets you create internal links automagically using topic text.

Related Knols
Everything2 does this really well though the use of soft links. E2 builds a related articles list by tracking transitions between articles. Given Google's prowess when it comes to finding related advertisingcontent it should be cinch to dynamically find Knols related to the current one. Internal linking would make this even easier, as would article tags. Which brings me to:

Article Tagging
The idea of a formalized taxonomy is clearly a non-starter. Fair enough, but that still leaves a healthy requirement for non-hierarchical cataloguing. Tags are a well established metaphor for cataloging to identify similar or related content. The 'alternative titles' field will makes it easier to search for a topic that's referred to by different names, but doesn't do much to help group or explore similar topics.

Topic Shells
E2 uses Node Shells to act like a shelf for a given topic. It contains all the articles with the same name written by various authors displayed as a kind of summary, indicating the reputation details for each of the contained Knols.

KnolShells would answer one of the criticisms related to ranking and linking to a topic covered in Knol. Currently you need to link to a specific author's Knol, or a search for the topic. KnolShells would let you link to the topic and offer all the available articles. This is particularly handy combined with internal links to streamline internal navigation, letting authors link to a topic rather than a particular article as it eliminates the 'first-in-most-linked' effect.

Improved Ratings and Reputation
Reputation systems are always problematic, but it would be nice to offer more statistical information on the popularity of each article beyond the a ambiguous 5 star rating mechanism. The E2 model uses the thumbs up / thumbs down model with the added ability for experienced users to flag particular articles as 'Cool'. Popular articles can then be used to populate the 'featured Knols' on the front page.

I like to know how popular my articles are and get an idea of who's reading them. Seems like a no-brainer to include Google Analytics the way they did Adsense.

Outbound Links
I fully understand the spammy reasons outbound links are no-followed, but it would be nice if this rule was applied more discriminantly. It would be very nice to see outbound links become 'followed' based on rating, author reputation, or/or time since posting. No one wants Knol filled with V1agr4 links, but there's no reason people shouldn't be able to link back to their own sites, and no reason those sites shouldn't get the associated link juice.

RSS Feeds and Feedback
Knol doesn't yet have a very robust feedback system. I'd want to subscribe to someone's New Knol Feed, and I'd like to subscribe to comments and reviews on Knols as well. As a Knol author email notifications when someone comments on, or reviews, my articles would be good too.

A private message mechanism for people to send messages to Knol authors within the Knol ecosystem would be a handy way to let people comment or suggest changes without publicly posting a comment or going through the 'edit' process.

In Closing...

These ideas certainly aren't comprehensive, but I think seeing some of these changes would really improve Knol. What do you think?


  1. The lack of openness around knols really bugs me. There is nothing that I can not find out about the wikipedia project (except the real names of the contributors). Is there a simple way to find out how many knols there actually are.

  2. Interesting question (and I don't have an answer).

    It would be nice to hear from the Knol team directly, AFAIK there's no blog or Google Group that most Google projects use for communicating with users.

  3. i recently joined:


    it's a community (social network) site for knol users.