Welcome to the new Independent website. We hope you enjoy it and we value your feedback. Please contact us here.


Rhodri Marsden: Why a search engine seldom gives you a straight answer


Faced with the prospect of going to watch a plumber called Melvyn perform as Elvis Presley in a tribute act called, somewhat predictably, Melvis, I scoured the internet to try and find out whether he was any good. But combinations of Melvyn, Melvin, Elvis, Melvis and plumber yielded up nothing from Google, so I set out for the evening with trepidation, while pondering a time in the future when I might be able to ask of a computer: “Is Melvis worth seeing in St Albans tonight?” and getting a straight answer in return.

A Google-sponsored study at the University of Maryland has been looking into the simplification of the internet searching process with the assistance of a group of seven, nine and 11-year-olds, amid concern that the often tortuous process of eking out information online isn’t particularly intuitive for children who are trying to do their homework. Of course, search engines strive to make their results as relevant as possible, suggesting related searches or correcting possible spelling errors, but it’s often hard to make ourselves understood to a Google or a Bing when we’re feeling vague or have limited information to give them. Equally, the search engine has its work cut out in returning results from a web of over 50 billion pages, many of which are trying to make themselves appear relevant but are merely designed to bring in advertising revenue, and where chunks of text – prices, product names, song titles, news items, video descriptions and so on – all appear the same to the computers doing the indexing.

The vision of web inventor Sir Tim Berners-Lee is the “semantic web”, where computers are understanding partners in our information trawl, taking our hands and guiding us in the right direction, rather than slinging a mass of unhelpful junk back at us or expecting us to use confusing operands like AND, ~, NOT or * in order to finely hone our searches. A step towards this is Wolfram Alpha, a British “answer engine” launched earlier this year which analyses our queries and computes the answers rather than guessing them, and offers a glimpse of how people of all ages and abilities will one day be able to make the most of the web. For example, entering “20th November 2010” returns a mass of useful information about that date, while Google merely delivers a page from a forum discussing wedding plans. Wolfram Alpha can’t yet tell me that Melvis can be relied upon to deliver a storming version of “Suspicious Minds”, but hey, it’s early days.

Email any technology gripes to cyberclinic@independent.co.uk