Internet Marketing
« Is It Time For Search Marketing Standards? | Main | The Personalized Search Revolution »
February 27, 2008
Louis Monier: Past, Present & Future of Search
We're here in the main room for our Wednesday keynote with Louis Monier. I am nursing a cup of coffee and a sore neck. The pillows here at the Hyatt are not my friends. I'd also like to register a complaint about the lack of juice at breakfast. Danny might be able to live on Diet Coke alone but I'm not made of such stern stuff.
Let's get down to business, shall we? Louis Monier, for those who don't recognize the name, is the founder of the AltaVista search engine. I actually met someone who still uses Altavista. She said she was too set in her ways to change. Some people are old school.
Louis is now at Cuill (pronounced Cool) after being at eBay and Google.
Danny asked him to cover the past, present and future of search. What is the point of search? To find information on the Web. These are all his views, he's not speaking on behalf of Cuill. The creation of the internet is on par with the creation of writing. I agree.
The Web is only as good as its index. This captured the need that people have and always will have about the Web. It's not some little teeny safe walled garden, but the whole sprawling thing.
Only full text search fo the Web would do. Search was such a need that no one argued against or rejected it. Words that used to be strictly jargon have become common "query", "keyword", etc.
1995 was when Altavista launched with a whopping 16 million documents and responded in a fraction of a second. It also had a feature that no one else did. The "link:" command. It basically gave rise to the search marketing community, as people figured out how to you it.
Altavista was nearly called Gotcha. Ouch.
For 2-3 years, they were the lead.
Banner ads appeared right about the same time.
The search result page is much more expensive to produce than a regular page. So search engines turned to portals to keep people on the page longer. The trouble is that this is incompatible to the mission of a search engine and so the quality of search became an afterthought.
Around 1998, something bad happened the first wave of spam…er, "aggressive marketing" appeared and it nearly killed search. Very precise queries would return a good result but most wouldn't. Search engines trusted the content of the page and didn't really have the ability to do any kind of quality control.
Right about this time, link analysis as the measure of a page came up. Google managed to revolutionize search. Their secret weapon was a core focus on link analysis for ranking, search, discreet text ads.
Where are we now? One dominant competitor, a couple of trailing players and a lot of very distant engines just struggling to get going.
"Trust me!" Who knows whether or not search is really good right now. You have to trust that the query really is returning the right answer because you can only see the first few results. Everything else is hidden. Now most of the time there is a right answer ("I'm Feeling Lucky") but sometimes there isn't.
Answering those types of long queries, the ones without a precise answer, hasn't progressed much in the last few years.
We get no help from the search engines. They don't offer 'how to phrase a query' or even just answering the question 'now what?"
E-commerce sites are much better at that. You get lots of help in the search process. Not just "shoes" but what kind and size and style. They don't just assume that the first result will make you happy. The query isn't finished so they offer refinement.
In some ways, it's still 1995. You go and you do a query. We haven't evolved.
Old school: Giving weight to the words in the query on the page.
Next: Weighing anchor text and linking information
After: Doing both.
The user interface: It's still 10 blue links.
So what's changed in the last 10 years? He thinks it's mostly questions of scale. We used to talk in millions now we talk in billions but he thinks that the Web is growing even faster than that. Size matters. You need to have access not just to a few select sites but to ALL sites.
If you have millions of results and none of them are the page you want, you're not going to be happy and the search engine isn't doing good enough.
To me, search is not just looking for my keys. It's about doing research, finding, learning. It's about figuring out what I want and gaining insight. That's the future of the Web.
Human powered search: Directories. Good content but teeny teeny amounts. Not a scalable solution.
Personalization: Sounds good. We all like to feel special. Diamond -- rock or baseball? The problem is this. I'd happily volunteer information to a search engine if it would improve my results. However, the trouble is that it's not going to because people are growing and evolving.
Social Search: My friends already have the answers. But how many people would it take? Not just my friends, but many, many more people. Again, not scalable.
Vertical Search: It's very clear that a special search engine with very specialized interested can do a better job in their niche. The catch? No one wants to remember 10000 different sites to search for the right topic. For some things, you make the effort (fine wines) but for most things, you go general (grocery store). That's going to limit the reach of vertical sites.
Natural Language Processing: That's more of a feature than a contender. How much good language is there on the Web? Not enough. And really who wants to write an essay in the search box.
Semantic Search: Webmasters are going to label everything to define what it mean! The problem is, what's the motivation? And if it's not going to be done by humans, then it'll take artificial intelligence. He's a little nervous about promises to take us from the Flintstones to the Jetsons in one shot.
Search engines are not just about navigation. Example, search as a Spell Checker: Testing your spelling in a search engine "blogging" takes two Gs, "blogology" takes one. There's also just learning from the search engine without clicking through. Search on an acronym and see all the various types of pages that are returned.
Why don't search engines take the same kind of approach?
10 years from now, we're not going to be typing two words in a search box. We'll be looking less at search engines as navigation but more as research assistants. It would not go out, do analysis and return a report.
In conclusion: These are the things that he's hoping to see from …
BABIES! There are babies on the screen. Aw, he's got the teeniest cutest twins ever.
We all accept a very narrow definition of search and we're being trained to stay in that. It's okay to want more. Right now search engines are the only game in town. Any choice that a search engine makes to cut you off from information is a bad thing.
In the end, size matters. We talk about billions of results but it's not enough if it isn't the right response.
Q&A
Why haven't refinement/clustering engines taken off? Why isn't structured data being used more often?
It's really hard… Two possibilities: it's too hard or it's just not precise enough yet. It's still too shallow.
Why not structured data? It's the nature of the Web. It's very hard to get structure into the data.
What do you think about blended search?
I didn't mentiont hat because I didn't wan this t take 3 hours. It's fine. To me that's the Web. The problem is that we're not doing any better processing of that. It's the same as Web search, there's no novelty there. They all need work.
Why are you dismissing NLP?
When they were working on Babelfish, step one is NLP. Any little slip will throw you off. A lot of the text on the Web isn't grammatical or even spell checked.
What about voice recognition, eye scanning, etc?
Not really related to search. It's great but it's not search.
Posted at February 27, 2008 9:52 AM
View related entries in: smxwest2008
Stumble It!
Add to Del.icio.us
