Just back from the week off, and much is up in the search world. I’ll have a roundup later in the day, but the top news for now is Ask, which has declared its intention to become the Fox of the search engine world – in other words, to come into the game late, with low odds, and hope to strike ratings gold. We knew this play was coming – it was clear when IAC and Diller bought Ask. Now the company has declared. Steve Berkowitz, CEO of Ask.com. throws down the gauntlet in the release:
“People deserve a search engine that gives them the tools to get what they need faster, not just a bunch of links on a page. Ask.com takes search to the next level.”
You’re not sure they’re serious? Diller is keynoting this week’s SES conference, for goodness sake (I won’t be there, focusing on FM then traveling to London later this week). More on all of this in the Ask Blog here.
Gary Price, who is joining Ask from SEW, has some interesting thoughts here.
A summary of the news: As expected, Jeeves is history. It’s now Ask.com. Teoma.com is no more, it’s been rebranded as “ExpertRank”. Ask has a new homepage, a new customizable “toolbox” function, an upgraded Maps feature, and some other features which you can read about in their release here.
What I find interesting is Ask’s decision to compete based on differentiation of approach to sponsored listings.
From the release:
Ask.com now has the fewest ads of any major search destination on the first screen of results. With Smart Answers, Ask.com is also the only search engine to place editorial results above advertisements.
More on this as it develops.
“People deserve a search engine that gives them the tools to get what they need faster, not just a bunch of links on a page. Ask.com takes search to the next level.”
Wasn’t there some similar sabre-rattling coming from Bill Gates a few months ago? Didn’t he also say something like ‘Google’s goal is to organize the world’s information, according to how they think it should be organized; our goal is to give you the tools so that you can easily organize it the way you want it’?
Rhetoric or not, it is an excellent point. And with now a fair number of search engines offering what is known in the academic world as ‘query expansion’ and/or topic-based clustering/reviewing (e.g. Ask, Gigablast, Vivisimo, heck even AltaVista had something similar years ago), one has to wonder why we don’t see any of these tools from Google.
I mean, c’mon. I thought their mission statement was to be good at one thing: Search. And there are so many well-known, proven search-enhancing tools out there, such as query expansion, relevance feedback, and clustering. Why does Google not offer these tools? Too much success == complacency?
I hope this play succeeds, because it might force the majors to start innovating again, instead of spending their time on chat and horoscopes. Sigh.
The bottom line is relevancy –
Will they tweak or do away with Subject Specific Popularity –
All the bells and whistles are nice – but can the ALGOs be improved to deliver valid results quickly?
Ask has changed their SERPs as of a few days ago – and many many many good sites are missing
Google has clustering technology, and have for a long time now. For some reason, they don’t see a need of making it a main public feature right now. Perhaps because most users don’t take advantage of it – maybe that will change.
SEW: Yes of course the bottom line is relevance. But I think you’re missing the fundamental point when you even think of these information retrieval tools as “bells and whistles”.
Take “relevance feedback” for example. It is much about ALGO improvement as anything else. You do a search for, say, “asian earthquake”. There is no way Google is ever going to know if you are talking about the pakistani earthquake or the indonesian earthquake. PageRank might favor one above the other, but relevance is not defined by PageRank, it is defined by YOUR information need. So if it puts indonesia at the top, and you wanted pakistan, it will fail. If it puts pakistan at the top, and you wanted indonesia, it will fail.
So lets say the results are all about the tsunami and indonesia, when what you really meant was pakistan. So, if relevance feedback were available, you would give a quick thumbs down to the top 1 or 2 indonesia results. The organic ALGO then uses this feedback, computes features of these documents you’ve “thumbsed down”, and re-ranks the list in such a way that all the pakistan earthquake pages get pulled to the top. All automatically/algorithmically, based on your quick couple of feedback clicks, in order to better and more quickly get you to the documents that are relevant to you.
Relevance feedback is a known technology. It’s been around for 30+ years, in the information retrieval (translation: non-web search) world. It is not a bell or a whistle. And yet Google won’t do it? Or they’ll do chat or horoscopes or finance first? Bah.