5 mistakes Google made with Knol
Wednesday, July 23rd by Ben YatesKnol — Google’s new Wikipedia for experts — is a spectacular engineering feat. Not many companies could set up an effortless cell-phone integrating ID verification system. (See yesterday’s post.)
But Google’s engineers always mess up the social side, and they did it again this time.
Here are 5 things they need to fix.
1. The Articles on the Front Page.
A front page isn’t just a sidenoted showcase window — it should be like a business card, conveying a focused impression of what the site is. It should excite people and make the site seem like a fun place to hang out.
Knol wants to attract experts that feel uncared for on Wikipedia — so the front page showcases articles by doctors.
This is what Google wanted people to see:
And this is what people actually saw:
If you think the top part is bad, scroll through the whole picture. The really amazing thing is how many articles are both depressing and gross.
This is bad because…
Because reading the front page makes you want to kill yourself. (And ironically, there is no “suicide prevention” article. Unironically?)
What were they thinking?
Well, they were focused on being engineering geniuses. Knol is built from a thousand intricately interlocking components — layers and layers of javascript frameworks, an interface to the database, sophisticated authentication tying the site into people’s google accounts, meticulously constructed algorithms to fillet the content for indexing.
And plus, Googlers are used to viewing front pages as unimportant fluff — they built a search engine dedicated to retrieving the best deep-buried content and laying it out nicely on a results page, sidebars and formatting be damned.
But front pages are super important.
To be fair, Google might also be trying to spread knol via word-of-mouth among physicians. But there has to be a way to do that without a front page of death.
How Wikipedia Does it
As with Knol, Wikipedia’s front page is pulled in from other pages (from “templates”, technically).
But unlike the Engineers who chose the featured knol articles, the people who edit those templates take their duty seriously. They’ve been immersed in wikipedia for years, and that spark of whoa is second nature. They’re basically coolhunters, staffing the most popular cool-stuff blog in the world, working for the admiration of their fellow otaku.
Deputy dog would kill to hire them away. Knol should be desperate to attract them. But they aren’t much interested in reading about autism and genital warts.
How Knol should do it
By injecting a human touch. Stop treating the front page of Knol like the front page of Google — as a pristine space to be populated by checking form boxes — and start putting messages from the team there. If you’re looking for physicians, say you’re looking for physicians.
2. Information Architecture of the Front Page
Sorry to harp on this, but looking at a front page should be like going into a coffeeshop — you should be able to look around, figure out what the crowd is like at a glance, see unwashed hippies with dreadlocks in the back or some college students on their laptops in the front corner.
The last thing you want to see is a bunch of empty seats. Then you’re not staring through a window into the world; you’re just sitting by yourself looking at a computer monitor.
So giving the impression of activity is vital for a community site, especially when the site is small. You want to show exactly what’s happening right now, who’s online, etc.
When wikipedia was younger, its most important place was Recent Changes, where you could see all the edits that just happened. Only vandal-fighters use the page now (some wikitourists, too), but it’s still in the left-hand sidebar of every wikipedia article.
Having a Recent Changes page makes it more fun to contribute, because you can be sure someone will see what you wrote. Everything2 has it, blogger has it, reddit has it, digg has it, livejournal used to have it. Even dating sites let you browse by “online now” and “recently updated”.
It’s one of the most important doorways into the site.
So where’s the Knol equivalent?
I looked long and hard (and let me tell you, I’m good at surfing the net). It’s not there.
At one point I hit upon a link in — seriously — 8 point text at the very bottom of the page that said “browse” (it only shows up if you’re logged in). But it went to a long, unformatted list of more “featured” articles (beginning with the ones I’d already seen).
What were they thinking?
“We have to fight spammers.” Google spends a lot of time and money fighting search-engine gaming, and they’re good at it.
The way they see it, a recent list would be a huge incentive for shady self-promotion, link-farming, blah blah blah.
How Wikipedia does it
The same way the french revolution did it: by making its participants feel like they’re full owners of something amazing, and god damn it, if we have to break a few filthy spammer necks we will.
Wikipeida users have the tools to oust spammers themselves. Knol users don’t.
How Apple does it
Leopard Wiki does Wikipedia one better, and actually displays a list of individual recent changes right in the global sidebar.
How Knol should do it
Knol lets users rate articles, so I was sure I’d see a list of the top ones somewhere — and low-rated articles could be kicked out. This is probably in development.
3. There Aren’t Any Interior Links
That is, there are no links from one Knol article to another.
This is bad because…
Because interior links tie the site together — they make it a single entity, not a bunch of disparate articles.
Linking articles together is fun — it feels like you’re building something important. And it means all the good articles get read, as people follow links from other articles.
What were they thinking?
“You can make a regular link to any site on the web, including other Knol articles! Why would you need anything else?”
This is misguided for two reasons. First, you’re not able to link to yet-unwritten articles — more on that in a second. Second, baking in a specialized link tool creates a social convention that encourages inter-linking and makes social interaction more interesting.
How Wikipedia does it
On Wikipedia, you can link to any article by surrounding its title with square brackets, [[like this]].
Because you can make a link out of any phrase, it’s possible to link to articles that don’t even exist yet. (These links show up in red.)
Redlinking was one of the most important ingredients in Wikipedia’s early growth; it showed readers of one article possible writing opportunities in another. It allowed editors to make their to-do lists and their articles the same thing.
How Everything2 does it
Like Knol, Everything2 is a bucket of articles with a rating system and individual authorship. But on Everything2, any user can add links to any article (”softlinks”).
And, as on Wikipedia, authors can link to written or unwritten articles (”nodes”, in e2-speak. Sounds a lot like “knol”, yes?).
How Knol Should do it
I like the double-square brackets, personally. I know this breaks with the use-buttons-for-everything Knol formatting convention, but it’s worth it.
4. The Search Function Doesn’t Work
This is a really strange one. If you type in a word that doesn’t exist in an article title, the article doesn’t come up.
How Wikipedia does it
Exactly the same way! Just kidding. Wikipedia’s search utility has gotten a lot better, though it’s still no match for a site-specific Google search of the Wikipedia domain. (And an army of Wikipedians go around manually creating “redirects” from one name to another.)
What were they thinking?
Hell if I know. They should at least have provided a general google-search in addition to their strict title-search.
5. The Whole Thing’s Half-baked!
This is a summary.
There are no opportunities for cross-pollination between articles. There is no way to sort articles by most-viewed, or by rating. There is no way to see a list of users. There is no way to search properly.
Basically, Knol seems unfinished. Until Google fixes these problems — and it should be pretty easy, honestly — Knol is not going to take off.
What were they thinking?
They were thinking, look at all these search results that Wikipedia is at the top of. We could be running ads on these!
Silly Google. If it were easy to make a Wikipedia, everyone would have done it. (You can read this part as either Bugs Bunny or Marylyn Monroe.)
Seriously, I think they’re probably taking things touch-and-go. But at least in the short term, Knol articles won’t rise to the top of the results unless Google gives them special treatment in the algorithm.
See also: Nobody pwns Wikipedia.
For some mistakes Wikipedia is making, see Nuking the Fridge and deletionism by the numbers.








July 24th, 2008 at 12:24 am
Great post. I will read your posts frequently. Added you to the RSS reader.
July 24th, 2008 at 6:29 am
You’ve got some valid points, but I think your analysis is a bit biased:
* Why does “Guantanamo bay trial” on wikipedia gets a “whoa” but “Lung cancer” on Knol gets a “depresssing”
* If you really wanted to “resarch [sic] [your] paper”, would you rather have “an authoratitive set of articles” or an “an encyclopedia anyone can edit” - would you base your research on? (assuming both had the content)
* I think you’re overstating impact of the links on the front page: The first couple of times it’s nice, but most users go straight to the search box and skip the rest of the front page.
* Wikipedia “makes its participants feel like they’re full owners of something amazing”. Okay, I’ve never contributed to the wikipedia but I’ve heard so many stories about deletionism and edit wars in wikipedia the prospect of having my work being undone puts me off doing it: Valid or not, it’s a perceived problem.
Big biases aside, your complaints are:
* Knol doesn’t have as many articles because it’s early days yet.
* Knol’s front page is biased towards medicine: Valid point - even if Knol is 90% medical articles, the front page should be broarder.
* Knol’s front page doesn’t tell you what’s going on. A “news” front page isn’t necessarily the right thing on every site.
* There aren’t any interior links: valid point. That’s what keeps people clicking around in wikipedia.
* They haven’t introduced a good search yet. Early days?
Okay, you have some points, but slagging off an emerging site as “half-baked” and praising wikipedia is harsh and rather ignores the fact that wikipedia was once an emerging site and may could have been called “half-baked” on similar metrics in its early days, too.
July 24th, 2008 at 6:50 am
How ’bout you copy this article into a Knol for “Criticisms of Google Knol”?
July 24th, 2008 at 6:56 am
Love your annotation of Knol and Wikipedia pages, especially the Wikipedia main page. Great post. :)
July 24th, 2008 at 7:21 am
That’s why it says Knol *BETA*. Of course it’s unfinished…
July 24th, 2008 at 7:50 am
You have a few good points, but you do realize that it’s still in the early stages of beta, right?
If there are multiple articles on the same topic, how would the user link to them if they HAVEN’T BEEN MADE YET? Even with the red-links, which one would you point to?
If Google really wants to pursue this project, I believe it’ll take off with a bang after they polish it.
July 24th, 2008 at 9:46 am
Dead-on accurate critism/critique.
From the description of Knol elsewhere I was really, really hoping for a Digg-o-Pedia minus anonymous Bozos.
One thing you didn’t mention: Where oh where are the tags? If they don’t provide a taxonomy or catalog anywhere (hence the “Plain old bag o’ Knols” on the front page) then they’ve got to support tags (by authors and readers).
So, (combining with your ideas) I hope it evolves into a Flickr-Digg-o-Pedia with these features:
1. Flickr-like tags and favorites
2. Digg-like rankings and trends w/ subcategories
3. Wikipedia-like interlinking and future-linking
4. (oh yeah) And Google-like search by title, content and tags
July 24th, 2008 at 10:00 am
i’m pretty surprised they launched like this. every other product launch I can remember was MUCH better… well, I’m happy to know that sometimes google shit’s too
July 24th, 2008 at 11:20 am
Mike: On everything2, all articles on the topic are listed on one page — so it’s possible to link to an empty “nodeshell” that will later be filled with multiple topics. (And the whoa was really about osama bin ladin’s chauffeur, but I took it out for clarity.)
July 24th, 2008 at 1:15 pm
I wrote my own article on Knol. I agree that it was a total disappointment.
The “it’s beta” excuse doesn’t fly. It’d be practically a total rewrite to get something that doesn’t suck.
July 24th, 2008 at 2:50 pm
Whether or not I have ever even seen the front page for Wikipedia, I surely do not remember it. I get to specific Wikipedia pages directly from Google searches. The same will be true for Google’s Knol. In fact, I avoid Wikipedia unless the topic of concern is unlikely to contain their usual biased, half-baked rubbish. I’ve taught my children never to use Wikipedia for anything school related and to always assume Wikipedia articles are wrong unless they find independent confirmation for each fact. Perhaps if folks like yourself focused more on content quality and how people use sites like Wikipedia and Knol and less on superficialities that dress up a pig to look like a princess, there wouldn’t be such a desperate need for something like Knol purports to be. If Knol pans out, with or without a front page that makes you happy, then I’ll be delighted. Front pages are not “super important”, but quality content is!
July 24th, 2008 at 9:01 pm
Ollie — with all due respect, there are hundreds of wikipedia editors who spend hours every day going to the library to verify facts on wikipedia, to make sure everything’s accurate. This is why wikipedia tends to stack up fairly well in the accuracy department against Britannica et al — see this article in the journal Nature.
You should be teaching your kids to evaluate the writing style of each wikipedia article to determine whether it’s been through the community mill.
Also: front pages might not be vital for an individual who’s interested in getting facts, but they are vital in attracting a large base of contributors.
July 25th, 2008 at 7:10 am
There is no way that the wikipedia editors can go through every article, or even a majority of them, on a regular basis. There are just too many. While I wouldn’t go to the extreme that Ollie mentioned, you can’t take anything seriously in Wikipedia unless you can find it in the link citations at the bottom. No link citations = opinion-based, erroneous article. It’s a generalization, but probably true.
July 25th, 2008 at 7:26 am
Allow me to clarify my point. I do think Wikipedia is a great resource, just one that needs to be utilized with caution. Many large “important” articles are trawled through more often by editors (I would assume) and many of these articles have over 100 citations/links to outside sources.
Also, I do think that your commentary about the front page is dead-on. We can argue about the factual information in Wikipedia all day, but the front page of Knol is a disaster… when the title is just a horrible disease, it certainly doesn’t make you want to read it…
July 25th, 2008 at 7:43 am
I think academics regularly underestimate Wikipedia’s reliability — but that’s a topic for another post.
July 25th, 2008 at 1:09 pm
Hey, Knol updated their front page! (The top articles are about music, pancakes, solar eclipse, etc.)
July 25th, 2008 at 5:13 pm
Some interesting points here.
July 25th, 2008 at 5:32 pm
Tangential point: I like how you did the brush stroke swooshes in the screenshots. :-)
July 26th, 2008 at 12:04 am
Haha, thanks. Illustrator is my baby (but you should know that brush strokes are, at the moment, deeply uncool).
August 14th, 2008 at 7:43 am
I’m sold :)) +1