Archives

Categories

’10 blue links’ can’t be the last word in search

Emil Protalinski of ars technica recently interviewed Stefan Weitz of Microsoft’s web search team and the resulting post throws up some familiar arguments about the future of search, but ones that are worth repeating.

Microsoft has been investing heavily in search for some time now and continues to do so in the belief that the current Google experience can be bettered.  I remember hearing Ballmer say in 2006 that it is crazy that we have all learnt to search with queries with an average of 2.x words, and that giving the search engine more information by using more words leads to worse results.  He then went on to say that Microsoft was good at coming from behind (e.g. spreadsheets and the Xbox) and would do so again in the search market.

Since then it is Google rather than Microsoft that has got stronger, despite the latters $100m acquisition of natural language search business Powerset in July last year.

Judging by the interview with Weitz this lack of progress hasn’t dented their belief that they can catch Google, and in addition to the Ballmer arguments above about the paucity of the current search experience they are now also saying the following:

In the last 6 to 12 months, Microsoft has learned that only about a quarter of users are satisfied with the results of their first query, about half end up having to refine their query or start with a completely different one, and another quarter abandon their search altogether.

and

the future of search wouldn’t be centered around the “10 blue links” that has become the user’s expectation for web search

These are all good points and when you think that the current search process is really very unatural then it starts to become obvious that there is massive scope for improvement and innovation going forward.  This is not to say that Google isn’t great – it is, and the current search experience allows us to do things that we didn’t even dream about 20 years ago, and it is not as if they are standing still either.  Nor is it to say that Microsoft will necessarily win here either – they might, but it will be tough for them, although the way they are going about trialing their search products internally and the hints about rebranding both bode well.

What I am here to say is that if you buy into the idea that search has a long way to go then it becomes credible that a startup might succeed in this space.  Particularly when the existing players are locked into the existing ‘one search box which returns some links’ paradigm.  The next phase of search might be different enough that it is hard for Google to iterate their existing offering to get there.  It may even re-frame the problem in a way that is orthoganal to Google’s current thinking, allowing a new competitor to slip in under their radar.  Realtime search might work this way.

Reblog this post [with Zemanta]
  • martinowen

    Most people don't explicitly have Boolean algebra – although we appear to be born with it.

    In the dark ages I used to rely on people I discovered who had pages of links to stuff that I would find useful and interesting. Now I can trawl the Delicious pages of people I think know something about what I want to know. Doing something with social search and something magical with JSON would be cool. Trusted folk and People-like-us are great filters. I am hoping to talk with CognitiveMatch – who may have some answers.

  • http://andybeard.eu AndyBeard

    I like to look at this from a different perspective

    The web currently is a competition with each publisher of content looking to earn more page views,a click or a sale, thus navigation between useful resources has deteriorated.

    A Google/MSN/Yahoo first query should be enough to get you to a landing page which as well as having relevant content to the query, also contains links that lead you to other content based on multiple probability factors… linguistic, price, value, conversion, social media votes.

    The closest example of this is About.com but that is limited to a few hundred subdomains – you could also liken it to the content portfolios of Yahoo, or the way Amazon suggests other purchases.
    For information though non-commercial and lacking commerical satisfaction metrics, Wikipedia is similar, and has thus become effectively “tag space”, the default place people link to on any topic.

    There isn't a product based equivalent that works, because there needs to be sufficient incentive to cooperate.

    Big rock projects however are hard to get off the ground.

  • http://www.theequitykicker.com brisbourne

    Andy – the challenge with this approach is that someone editorially has to decide what is interesting/relevant. Despite the inevitable problems it creates with people trying game the system my gut is that the leading search play will always be truly universal (or close to it).

    Thanks for the comment.

  • http://andybeard.eu AndyBeard

    I was watching an old video from Techstars 2 years ago, a presentation by 2 of the Mybloglog founders Todd & Eric http://vimeo.com/227450 – when Scott Rafer joined them they began to finally realise their most valuable asset – “It's all in the data”

    A single small website doesn't gather enough data for it to really be meaningful.

    If you control the backend for 100,000 online marketers, content creators, bloggers, affiliates and even backend sales data via API or explicit code, it is a totally different ball game, and you can come up with far better suggested links, as Amazon does partially, but they don't have all the data.

    If someone searches for a Nikon D90 camera, Google themselves isn't going to suggest a review for an equivalent Canon that is better value. Adwords advertisers might, because either as a human or cross populating keywords it makes sense to bid for competing products, as long as they avoid problems with brand names.

    It isn't gaming the system when Ebay collect data from shopping.com epinions.com half.ebay.com and ebay.com, and then heavily interlink them.
    It isn't gaming the system when Amazon does the same with IMDB, or I gather from a 2007 press release http://www.dpreview.com, though that relationship isn't clear on the site, they seem to be far less integrated.

    The leading search play is important – I tangled a little with Google for 18 months over paid reviews, paid links, employee links, shareholder links and possibly most important, affiliate links, just to narrow the grey area a little.

    Ultimately it is possible to give the end user a much better experience after they have left the search engine, or followed a referral link from social media, whilst still guiding them willingly down a path to an informed purchase.

    At the same time the earning potential for smaller content creators can be magnified by an order of magnitude.

  • http://www.online-pharmacy-adviser.com/atarax-hydroxyne.html buy atarax

    Atarax (hydroxyzine) for Anxiety: I take this along with Lexapro… I was having such hard times falling asleep due to nine million things going on through the day … then the anxiety kicked in…sometimes I would be up for hours while everyone is sound to sleep…When I started taking this I fall asleep within thirty mins and usually sleep sound. It is not addicting… and good price for the good things the meds do

  • http://www.online-pharmacy-adviser.com/atarax-hydroxyne.html atarax25mg

    Atarax (hydroxyzine) for Anxiety: I take this along with Lexapro… I was having such hard times falling asleep due to nine million things going on through the day … then the anxiety kicked in…sometimes I would be up for hours while everyone is sound to sleep…When I started taking this I fall asleep within thirty mins and usually sleep sound. It is not addicting… and good price for the good things the meds do

  • http://www.1stpagekillerearningsreview.com 1stpagekillerearning reviews

    Nice blog.