» Publishers, Monetize your RSS feeds with FeedShow: More infos (Show/Hide Ads)
Yes, this is the last day for the kozoru blog. We're removing it in preparation for our launch on June 5th. Now you can keep up to date with us at our news page.
By the way, all the blog content will still be accessible at http://www.kozoru.com/main.html. If you have any problems, or if something goes down, please email me at gardner at kozoru dot com and I'll see if I can get it resolved.
Thanks for reading and we look forward to your comments about byoms.
We'll have more soon, but in the meantime take a look at the press release to find out more details about byoms.
Filed under: company, development
One of the things that distinguishes kozoru from other search engines is our ability to return sentences, rather than snippets, as results. To do this kozoru breaks a web page into sentences.
We have decided to release this piece of the kozoru technology under the Creative Commons Attribution license.
The code, written in newLISP, breaks raw HTML into sentences using two key functions.
- HTML Cleaning - This removes html markup and converts the page to text.
- Sentence Boundary - This breaks a given piece of text into sentences.
Since HTML contains more data on sentence boundaries than plain text, the HTML cleaning function can insert known sentence breaks into its output. This feature can then be leveraged by the sentence boundary function.
Filed under: company, development
When Justin told me that my last post about 43 things is the most popular post on this blog I didn't believe him. Who knew? So you can imagine that writing a followup is a lot of pressure.
Well, here I go. Another 43 things I've learned at kozoru.
- Employee’s can actually get tired of having free lunch brought in everyday.
- Fiji Bottled water
- The website for naked juice is NOT www.naked.com but www.nakedjuice.com
- A dashboard is not just a part of a car’s interior.
- As soon as a computer has been decommissioned and sold, it will be needed again.
- Cache is not to be confused with CASH.
- Corporate Lawyers can be very expensive for a startup.
- Debugging a kernel does not involve taking the worms off an ear of corn.
- Thai Orchid has the best Thai Food in the Kansas city area.
- Non-native Kansans find our weather slightly extreme.
- What exactly noodling for catfish involves.
- Blackberries are not always edible.
- A RAZR is not for shaving.
- How to mount a drive.
- That a video card is necessary even if you do not plan to watch videos on your computer.
- Google Maps
- Bandwidth does not mean the drummer is overweight.
- The 80/20 rule
- You can get a new passport the same day in Chicago and San Francisco.
- Everything is connected via Six degrees of Separation.
- Odwalla juice.
- Whole Foods – a great organic grocery store!!
- Every Enterprise Car Rental location is independent of each other. You cannot easily return a car to a location different from the one in which you rented it.
- At Hertz, you can EASILY return a rental car to a location different from the one in which you rented it.
- How to use SMS on my cell phone.
- Emoticons :-)
- The online computer parts vendors that offer FREE shipping.
- An appliance does not have to be a physical or touchable item.
- How to sell stuff on ebay.
- Protein drinks taste similar to sweetened sandy pond scum.
- Just how valuable a marble is.
- Midwest and Southwest Airlines are basically the only airlines that fly non-stop out of Kansas City.
- A road map doesn’t have to show highways.
- Room temperature drinking water is better than ice water.
- Airborne
- Crackers are not just something you dunk in your potato soup.
- Computer Spam is not canned meat, but is equally nasty.
- Shrook
- It’s hard to keep a white keyboard clean.
- Having two weeks off at Christmas time is awesome.
- Whiteboard markers grow legs.
- Humpty Dumpty fell off the wall because he was kicked by the cow jumping over the moon.
- Working for a startup can be very demanding, but also very rewarding.
Gred Linden of Geeking With Greg points out a paper written by four Google employees that discusses the search experience and what gets in the way of satisfying searching.
Note these sections Linden has pointed out. They're about spamming and judging whether or not certain ranking changes are effective...
One particularly intriguing problem in web IR arises from the attempt by some commercial interests to unduly heighten the ranking of their web pages by engaging in various forms of spamming .... Classification schemes must work in an adversarial context as spammers will continually seek ways of thwarting automatic filters. Adversarial classification is an area in which precious little work has been done.
Proper evaluation of [relevance rank] improvements is a non-trivial task ... Recent efforts in this area have examined interleaving the results of two different ranking schemes and using statistical tests based on the results users clicked on to determine which ranking scheme is "better". There has also been work along the lines of using decision theoretic analysis (i.e., maximizing users' utility when searching, considering the relevance of the results found as well as the time taken to find those results) as a means for determining the "goodness" of a ranking scheme.
In any event, definitely worth a look.
Filed under: industry
When the Department of Justice request a sample of Google's index, we here at kozoru had this observation :
It's pretty ridiculous for the government to expect a company to open up and show them all their data. And by the way, the reason for doing this (seeing how much traffic certain sites are getting) doesn't seem like it would help the government's case at all.
Well, looks like Google had the same thoughts in their official response.
From the Google blog:
Google is, of course, concerned about the availability of materials harmful to minors on the Internet, but that shared concern does not render the Government's request acceptable or relevant. In truth, the data demanded tells the Government absolutely nothing about either filters or the effectiveness of laws. Nor will the data tell the Government whether a given search would return any particular URL. Nor will the URL returned, by its name alone, tell the Government whether that URL was a site that contained material harmful to minors.
Exactly.
Filed under: industry
Well, the butler has officially left the building, and with him one of the last characters from web 1.0.
From the Ask blog:
The majority of people are ambivalent about the character. But even for these people, Jeeves means something -- a place to go to ask questions -- and in their minds it will be difficult to ever think of him as anything more. (And for those who haven't tried the Ask Jeeves since 1999 or 2000, it meant asking questions and not getting answers.) He was so well-branded during the late '90's and early 00's, he became part of pop culture, appearing everywhere from the Macy's Thanksgiving Parade to Arrested Development. Some even think of Jeeves as representative of the dot-com hype, like the Sock Puppet.
[...]
So, we're going to take the leap and strike out for a fresh identity, one that fits more with who we've become than who we used to be. One that revolves more around the site, and what it does for our users, rather than around a character.
It'll be interesting to see what they come up with. In the meantime, you can visit their special Jeeves retirement site.
Ahh marketing...
Filed under: industry
You get to avoid phone calls like this...
3pm: I call our data center vendor for additional space. He says he has only 1 rack left at 3x the original price and only on a 1 year contract (we are currently month to month). He turns me on to another.
Even though we've had our various misadventures with vendors , not having any rack space available when you really need to grow is particularly painful.
Good luck Munjal!
Filed under: company
When you're building a search technology that answers question, these are two issues that inevitably pop up in the process.
Given that, Search-Science's Xan Porter has a good overview of the unique challenges facing those who tackle these problems in the posts "Topic Detection" and "The Pros And Cons Of The Semantic Web."
Filed under: industry
I post about this because it has the potential to affect the credibility of search engines everywhere, even ones that haven't launched yet. Of course the government is presently just looking at the biggest indexes on the block, but when will they start to look elsewhere? It could be sooner than you may think.
From the Mercury News:
The Bush administration on Wednesday asked a federal judge to order Google to turn over a broad range of material from its closely guarded databases.
The move is part of a government effort to revive an Internet child protection law struck down two years ago by the U.S. Supreme Court. The law was meant to punish online pornography sites that make their content accessible to minors. The government contends it needs the Google data to determine how often pornography shows up in online searches.
In court papers filed in U.S. District Court in San Jose, Justice Department lawyers revealed that Google has refused to comply with a subpoena issued last year for the records, which include a request for 1 million random Web addresses and records of all Google searches from any one-week period.
The Mountain View-based search and advertising giant opposes releasing the information on a variety of grounds, saying it would violate the privacy rights of its users and reveal company trade secrets, according to court documents.
Nicole Wong, an associate general counsel for Google, said the company will fight the government's effort ``vigorously.''
Good for Google. It's pretty ridiculous for the government to expect a company to open up and show them all their data. And by the way, the reason for doing this (seeing how much traffic certain sites are getting) doesn't seem like it would help the government's case at all.
More from Danny Sullivan:
Here's a thought. If you want to measure how much porn is showing up in searches, try searching for it yourself rather than issuing privacy alarm sounding subpoenas. It would certainly be more accurate.
Getting a list of all searches in one week definitely would let US federal government dig deep into the long tail of porn searches. But then again, the sheer amount of data would be overwhelming. Do you know every variation of a term someone might use, that you're going to dig out of the hundreds of millions of searches you'd get? Oh, and be sure you filter out all the automated queries coming in from rank checking tools, while you're add it. They won't skew the data at all, nope.
Agreed, these moves by the government don't make a lot of sense. Maybe I'm missing something, but it really doesn't feel like this is helping them determine who is going to a given site or not.
In any event, stay strong Google. If not just for you, but for all of us. Well, at least most of us. Several other undisclosed search engines have already shared their info with the government. No names were mentioned, but it doesn't take a genius to understand who they are.
Ugh.
Filed under: industry
Every year, since 1996, I have given the entire team (except the Networking and Support folks, unfortunately) the week of December 25th off work. It is a kind of Holiday bonus meets attaboy and helps everyone take the time to recharge, refuel and decompress over the holiday break.
If people have kids or family, they can spend the week with them, if not -- 80 hours power-leveling an Undead Mage on the Draenor server of World of Warcraft is good for the mind, body and spirit.
And so, December 22nd through January 3rd was the "holiday break" for the company. No one had to stay behind and work (we don't yet have a public product or customers, so the network guys are off too).
This year also marked the strangest event I've witnessed in my decade-long experience starting, flipping, crashing and creating companies: Not one single e-mail to the kozoru internal mailing lists.
This puzzled and then bothered me, before I became confused and then finally settled into a kind of relaxed calm.
I realized, as I had never realized before, this probably the hardest year any of us had ever experienced as a company, team or employee. We'd been through a near acquisition, meetings with major players in the search space, five product revisions, a new direction (mobile) and several rewrites of the core technology platform. I remember a time over the summer where more than half the kozoru team worked for over fifty days in a row without a single day off, including weekends and holidays.
We were naive in 2004 when we started the company. We believed if we could just build something which would answer questions, it would be enough to drive users and partners. We believed the world needed an "Ask Jeeves that works!"
Turns out we were wrong, but not completely.
We realized, but it never sunk in, that the world does need kozoru. It's just that the web doesn't. Not really. At least, not yet.
And that's what this break gave us. Time to spend with our families, friends and thoughts. Time for us to pull together a new plan of action and a way of approaching the market. Time to get some partners lined up and turn a few possible partners away for strategic reasons. Time to deliver on the promise we have built and evangelized for the past year.
We're back and ready to move the company to the next level.
Things are about to get interesting and in a very public way.
I hope you'll stick around for the ride. It may not be perfect or what you expected, but I promise it will be interesting.
Filed under: company
Yes, it's been quiet around the kozoru blog lately. We've been away for a week having a nice break before the new year.
Happy 2006 and here's to a great year.
Filed under: company
We've been reading Don Dodge's blog and recently he posted an interesting look at what defines a great startup.
It is interesting to note that VCs invest in startups for the same reasons that the major companies acquire startups; 1) Great team of extraordinary people, 2) innovative technology, 3) growing user base, 4) a hot new market with huge growth potential. These are pretty much in rank order of importance.
Truly great engineering talent is hard to find. They are worth their weight in gold. Using a basketball analogy, Michael Jordan was only 1/5th of the Chicago Bulls starting lineup, but he was 100% of the reason they won so many games.
One superstar engineer/visionary like Ray Ozzie is worth 100 really great engineers. And, one really great engineer is worth another 100 good engineers. This is the normal order of things, yet few CEOs understand this. The truth is you need all levels of talent to build out a team, but without the superstar it will be tough to win.
I also thought this summation was particularly noteworthy...
Real estate investment is all about location, location, location. Technology investment is all about people, people, people.
Filed under: company
Yahoo just launched their Yahoo Answers site today, and it walks and talks a lot like all of the other social network answering services we've seen lately - those being Google Answers and Wondir. Well, except Wondir and Yahoo Answers are free. Google Answers makes you set a price to get your information.
So obviously, this idea isn't exactly exciting and new. If anything it's decidedly me-too. What's the point then?
Zawodny has this to say in his launch statement on the Yahoo Blog:
We want to give anyone a place to tap the collective wisdom of the Internet for advice, recommendations, theories, jokes, … whatever. Anyone can answer. It’s free. And once your question is answered, you get to pick the best answer and the whole thread is archived and searchable. So in addition to getting answers, you’re helping to add to the “the Internet’s” collective knowledge.
Interesting, but again, nothing new. And tapping the collective wisdom of the Internet can only be done in a very limited way when you rely on humans. These inherent limitations are fairly obvious.
To be fair, though, Zawodny's talking about the kinds of questions you can't get answered through keyword search. But still...I think sometimes we know that the information is out there, it's just that current search engines just can't find it. Given that, should we rely on humans to fill in that ever widening gap or create better search technology? The answer here seems to be both.
So then I take a walk around the blogosphere to see what other people are saying. I find Gary Price from Search Engine Watch covering this story and talking with Ofer Shaked, Director of Engineering for Yahoo Search.
Shaked told me that Yahoo Answers is built to on the company's social search technology with a focus on answering day to day questions of a subjunctive nature (Where can I buy..., What's a good show to see..., etc.) and organizing them into a more structured info base with the use of several features including categories. It's motto is: Ask--Answer--Discover. Shaked also said in testing "day to day" types of questions have been most popular. Yahoo strives for this to be a self policing community.
Still, I feel like I'm missing something. This is a new product for Yahoo's mega-portal and it will add value to their index. Yes, got all that. But where's that last piece of the puzzle? Surely there has to be more?
Michael Arrington over at Tech Crunch calls for tagging:
While this certainly helps structure the data for easier search, it isn’t very useful to the publisher. It would be so much easier if, like Flickr, the person asking the question could tag it with a few descriptive terms. They have an incentive to get it right, and Yahoo would quickly have rich enough data to create a virtual category on the fly as users search or browse through the listings.
Making this a free, community-driven service takes it way beyond Google answers. Take the next step: ditch this impossible to maintain category system and move to tagging and dynamic, on-the-fly taxonomy.
Sure, tagging could be interesting, but that scales as well as humans answering questions do. Again, what am I missing?
Ahhh, but then I see it. On my Yahoo Answers home page I find this at the bottom: Earn money for your answers. All I have to do is join the Yahoo Publisher Network (YPN) to earn money for my best knowledge. How that happens is not really clearly outlined, though. I just have to sign up and wait.
So is Yahoo Answers just a way to get people into the YPN? Interesting...
Then I go back to Gary Price, who seems to have this take-away from the whole thing:
One thing I can't help but wonder (am I getting a bit old) if this is another 96-120 hour story. In other words, lots of talk about it for a few days and then we don't here about it again, at least for awhile. With all of the new services that are being released these days it seems like we see more and more of these 96-120 stories. Why? The public and press (minus the tech geeks, of course) don't have time and, sad to say (is it?), aren't nterested in keeping up with EVERYTHING NEW. I wonder if a future tactic companies might take is instead of developing products and then giving them to the early adopters, develop services and tools and promote them directly at certain user groups. From teachers, to lawyers, to nurses.
Any of the search giants will tell you that search technology is in its infancy. Well, if that's the case, will question-answering sites, which rely solely on the human touch, be forever young? And what will it take for these sites to become more than me-too ideas?
I'd like your thoughts on this.
Filed under: industry
As many of you already know, JSF often writes about life, work and everything in between over at his blog.
Now he's revealed quite a bit more about why we've been quiet recently.
It is a strange and completely surreal experience when one tries to convince someone else of their value. It requires a delicate balance between confidence and humility and it almost never works out that the things you say and the things you mean are exactly the same for the person on the opposite end of the negotiation.
I am in the fortunate situation of being surrounded by a team of people who I believe to be the smartest, most talented and most interesting group of people I have ever had the pleasure of working with. We've been through a few near misses with our technology (one when we presented the technology to Apple Computer, showing them how we could improve their Knowledge Base, once at Google and others I cannot mention because of disclosure issues).
Later on in the post, there's this...
Likely I am remembering a quote from someone else, but I do believe technology should be thoughtful, imaginative and adventurous. I also believe technology should solve a problem people actually care about.
Make no mistake, however, as you will be solving a lot of problems no one but you cares about along the way, unless some other poor, unfortunate person has solved them already (c.f. Sentence Boundary Detection in previous entries).
On the one hand, I seriously want to release what we've done in a more open forum and let people play with it. On the other hand, making a move right now, when people are seriously standing on us to negotiate a relationship with them, I don't want to go through what a Stanford MBA would call, "ejaculation, premature."
Maybe we'll do something for Christmas and call it a holiday present to the world...
As "they" say, read the whole thing.
And while you're at it, read some thoughts from Battelle and Threadwatch.org.
Filed under: company