bbc.co.uk Navigation

Rory Cellan-Jones

Searching for another Google

  • Rory Cellan-Jones
  • 6 May 09, 10:35 GMT

"I think it's going to be pretty exciting. A new paradigm for using computers and the web." That's what the creator of Wolfram Alpha told the world recently about a new way of extracting information from the web.

 Image of Wolfram Alpha websiteStephen Wolfram believes his computational knowledge engine will provide a better way of executing a certain type of search query, where you are looking for data, than Google. It will also give you answers directly rather than pointing you to websites where you may find them.

We won't really know how well it works until it goes live later this month - but Technology Review managed to get hold of a login and put it to the test. For the most part, it came up with some quite satisfactory results, particularly for searches involving hard numbers - the term "GM Ford" turned up an array of graphs and tables giving a good statististical picture of the two American car companies.

Inspired by this, and lacking a Wolfram Alpha login, I decided to put some of the other pretenders to Google's crown to the test. Over the last year or so, Cuil, True Knowledge, and Kosmix have all appeared on the scene, offering alternative ways to search - whether by using semantic web techniques or by presenting results in a different way.

Cuil, set up by some ex-Google employees, raised a pile of money and a good deal of publicity last year, promising more detailed results, but has since faded from view.

True Knowledge is a semantic search company started by some very clever folks in Cambridge, England, and is still at the beta stage. And Kosmix is a Silicon Valley start-up, just across town from Google in Mountain View - it promises to give you a rounded view of a subject by assembling data from a variety of searches and sources.

So I tried three of the same searches with which Technology Review tested Wolfram Alpha - plus a take two to give each engine a better chance - and then threw in one of my own.

Here are the results:

SEARCH TERM: Sydney New York

Google: provided links first to cheap flights.

Cuil: linked to helicopter tours of Sydney harbour.

True Knowledge: replied " Sorry, I don't understand that question."

Kosmix: linked to Yahoo answers - with a reply to "how many hours from Sydney to New York."

Take two: distance Sydney to New York

Google: linked to distance calculator sites, which would then provide an answer.

Cuil: "No results were found..."

True Knowledge: "Sorry, I don't understand that question."

Kosmix: Led to WikiAnswers: with the answer - "the distance between Sydney, Australia and New York, New York is 9935 miles."

SEARCH TERM: 10 pounds kilograms

Google: Linked first to a site where you can make metric to imperial calculations.

Cuil: Linked first to a news story about a drug seizure.

True Knowledge: "It sounds like 10 pounds kilograms may be an object that I don't know about yet."

Kosmix: Linked to the same conversion site provided by Google - and a lot of conversations about diets.

Take two: What is 10 pounds in kilograms

Google:10 pounds = 4.5359237 kilograms

Cuil: First result was a Wikipedia entry about "free recoil", apparently about small arms.

True Knowledge: This time after scratching its head the semantic engine comes up with the answer too: 4.5359237 kilograms.

Kosmix: Once again, linked to conversion sites and conversations about diets.

SEARCH TERM: GM Ford

Google: First on the list is a link about a writer called GM Ford.

Cuil: comes up with a list of essays comparing GM to Ford.

True Knowledge: "Sorry, I don't understand that question."

Kosmix: First links to the same Google results - but later to other sites with comparisons of the companies.

Take two: Compare GM with Ford

Google: First link is to a news story on "GM and Ford threaten merger".

Cuil: Link to news story "President Bush speaks out on GM and Ford.

True Knowledge: "Sorry, I don't understand that question."

Kosmix: A bunch of similar links to Google but mostly irrelevant.

Finally, I posed my own factual question - just the kind that Wolfram Alpha is apparently good at answering - Who was the president of France in 1957? And here were the answers:

Google: Went straight to a Wikipedia page about heads of state, with the answer that it was Rene Coty.

Cuil: Wikipedia: Takes you to another Wikipedia page to find that "Felix Gaillard was elected president of France on November 12, 1957."

True Knowledge: Became very confused and asked me to select from these two questions - what or who is the president of the organisation the French national football team in the year 1957? Or "what or who is the president(head of state) of France..." I chose the latter - and got this: "Sorry, I don't know the answer to that question. It sounds like this is something True Knowledge doesn't know about yet."

Kosmix: links to the same Google search results, but also to other resources about French presidents.

As far as I can see, Rene Coty is the right answer - Felix Gaillard was actually the president of the French assembly. But then again, Wikipedia could be wrong.

To sum up, none of the three Google rivals really merited a detour away from most people's search engine of choice, though Kosmix did perform better than the other two in helping you towards an answer, and True Knowledge has the excuse that it's still in beta. Then again, Google itself didn't really supply speedy access to the kind of data-rich results that Wolfram Alpha promises.

There is a need for an alternative way of turning up all kinds of data - I'm always struggling to find government statistics, corporate results, long lost batting averages - so perhaps that will be on offer soon.

But as Kosmix, True Knowledge and Cuil have shown, it's easy to boast about exciting new technologies that will provide "a new paradigm for using computers and the web". It's rather more difficult to deliver.

Comments

  • Comment number 1.

    Oh please! Not another paradigm. They're so '90s.

  • Comment number 2.

    Seriously... it's all good and great if it works... but no one, and I mean no one has made it work. Where do they get their investors? No really.... I got a chili recipe and a burger restaurant concept more likely to succeed than taking on Google.... I'm just saying....

    Liv

  • Comment number 3.

    It's going to be interesting to see if Google is challenged. It will be good for the industry and all of us seeking out information to have an alternative. If nobody challenges Google successfully, we might see the search engine become the next 'evil-empire' in the footsteps of Microsoft.

  • Comment number 4.

    it may help to change how you ask the question.

    replace - "who is the president of france?" (searching for questions)

    with

    "the president of france is" (searching for answers)

    helped me out with my problems finding the red bricks in lego bat man.

  • Comment number 5.

    I am a strategist in an advertising agency and use search engines all the time for desk research. A great way to find a specific piece of information on a search engine is to make an "exact search" using the words you expect to form the answer. For example: "The capital of Australia is", or "BMW's market share is". You can also use wildcards (*) to fill in any gaps. For example: "the property market declined by * in 2008". This is also a good technique to find people with a particular lifestyle, attitude or experience. For example: "I wish my bank would" or "I read the Guardian because". Hope this helps.

  • Comment number 6.

    In Google I typed: 'how far is sydney from new york', the first link gave me a page saying 9935 miles (15989 km).
    You'd think Google maps would be able to produce an 'as the crow flies' distance between two points; but no, they gave 16,228 miles - as if I could drive from the USA via Hawaii and Japan to Australia!

    Google can convert [for example] kilos to pounds (or vice versa) if you prefix the question with the word 'convert': e.g. convert 10 kilograms to pounds - it usefully shows the answer above the first link. (10 kilograms = 22.0462262 pounds.)
    You can also use this prefix to convert from/to imperial or metric systems for lengths (inches, ft, yds, miles etc), weights, volumes, etc. You can also convert from one currency to another; more than one country uses pounds and dollars so use 'GB pounds' and 'US dollars'.

    There's a list of 101 Google search tips n tricks here:
    http://www.pcanswers.co.uk/node/4122

    But its not about knowing how to phrase a search question. Were still a long way from a semantic web, where information can be pulled together from various sources and presented in easy to understand form (though various niche websites are attempting this). As peoples expectations rise it may never exist in a truly satisfactory form. Still, competition means companies will keep on trying to reach it.

    PS. Rene Coty was French President in 1957 (1954 1959).

  • Comment number 7.

    For the time being, there's nothing "wrong" with Google. The most likely thing to bring it down is if its search results start being less than useful. For example, if their advertising-based business model goes a bit crazy so you suddenly get nothing but ads, or if they start removing search results to certain sites (for example, should they link to thepiratebay or not?). They also have copyright issues of their own to contend with, as some search results may point to copyrighted files.

  • Comment number 8.

    It's an interesting experiment but not quite fair. Different search engines do different things. Google give you a list of links so the first result is important. But the other engines use a different approach,so Cuil's results have a sidebar that does a category search. What's the point if every new engine just tries to copy Google? Expecting Wolfram to ape Google would defeat the whole purpose.

  • Comment number 9.

    Whilst this proves interesting to read, I need to call into question why we are looking for new search engines? We already have search engines such as dogpile.com which acts to collate all searches/information/pages from google, yahoo, ask and windows live search. This provides the best then of four worlds, and more importantly of repsected and regularly used search engines.Dogpile.com so everyone know sis a metasearch engine.I have witnessed a description of dogpile.com as the tesco of search engines. It can be said, however to be ironic how I came across this through a google search, no?

    As for this new engine 'wolfram Alpha' appears to be a 'google' version of wikipedia.

  • Comment number 10.

    SmartInsight,

    Take a question like "Is the population of Sydney greater than the population of Birmingham UK?" No pure text-matching search engine could handle that type of question unless it can find articles containing the specific result, so even sites like dogpile can't get the answers in a single search.

    That's why everyone (including the major search companies) is on the look-out for technology that can compute knowledge instead of blindly looking it up.

  • Comment number 11.

    The design looks awful.

  • Comment number 12.

    True Knowledge DOES answer the first four questions. It just expects them in the phrase of a real question instead of keywords. I just typed "What is the distance between Sydney and New York City" and "What is 10 pounds in kilgrams" and got the correct answers straight away.

    Jed

  • Comment number 13.

    This may be the beginning of a new kind of service.The term "Engine computing " may be more appropriate to describe this service rather than "computational knowledge engine ". we could look at cloud computing for an example.
    Using the term computational knowledge engine could be confusing and may lead to misinterpretation of the service.

  • Comment number 14.

    "helped me out with my problems finding the red bricks in lego bat man. "

    Oddly enough, I typed in Red Bricks Lego Batman and the first link was gamefaqs.com, which gave a complete walkthrough, including red bricks.

    I've noticed this before with Google - sometimes, two people in the UK typing in the same search get different results.

    I wonder if it's tailored per region as well as domain area?

  • Comment number 15.

    True Knowledge isn't designed to do web searching so queries like "GM Ford" and "sydney new york" which don't have a direct answer (other than a list of pages) are outside its scope.

    It's designed to complement search by providing direct answers to questions with keyword searches (such as Google results) being producable instead when it doesn't understand or answer the question.

  • Comment number 16.

    The phrase "i don't understand the question" surely should have given our author the clue, that TrueKnowledge was requiring a question

    as for the red brick question - google's quieter sponsored links/results doesn't help with consistancy (and incidentally, makes it very difficult to prove that what you are paying for is what you get when attempting to ensure your link ends up at the top of the list) as it randomises the responses. whoever pays most will end up top more often, but not necessarily every time.

    I found youtube was eventually the best source of info!

  • Comment number 17.

    Personally I feel that there is definitely a need for an improvement on Google. On Twitter the other day, 2 of my friends tweeted about how they were unsatisfied they couldn't find what they wanted in the space of 2 hours of each other. To me that signifies a need for change.

 

The BBC is not responsible for the content of external internet sites

BBC.co.uk