BBC Homepage World Service Education
BBC Homepagelow graphics version | feedback | help
BBC News Online
 You are in: Sci/Tech
Front Page 
UK Politics 
Talking Point 
In Depth 
Monday, 15 May, 2000, 15:59 GMT 16:59 UK
Half the internet is going nowhere
Web is a dead end
Where can't you go today?
By BBC News Online internet reporter Mark Ward

The internet is less like an information superhighway and more like a one-way street leading to lots of dead ends.

A study of the way that the internet is wired up has revealed that the net very often sends traffic in one direction and leads surfers to websites that link to nowhere.

About 30% of the
1.4 billion unique documents on the web are rubbish

Joe Frost Inktomi

In one of the most comprehensive studies of internet connections, researchers from IBM, Compaq and Altavista traced the 1.5 billion links between 200 million separate web pages.

Their findings contradict previous work that suggested it was possible to reach any page from any other by, at most, 19 steps.

Instead, the researchers found that the web was much more intricate. While some websites are heavily connected, some lead nowhere and others are almost impossibly hard to find.

The study found that only at the heart of the internet are web pages linked to each other. The researchers estimate that only 30% of all web pages make up this heavily interconnected core which most of the search engines index. They dubbed the core the 'giant strongly connected component'.

Big Heart

Beyond this core are websites that are increasingly isolated. The researchers, led by Ravi Kumar from the IBM Almaden laboratory, found that 24% of websites have links that lead towards the core.

A further 24% are destination pages that can be reached by following links out of the core pages. The final 22% are utterly isolated from the core or self-contained islands between friends or special interest groups. This means that 46% of web pages go nowhere.

The results are backed up by the work that web search company Inktomi has done on indexing the links between pages. Inktomi has indexed around 1.4 billion web pages to get a better idea of which pages come up as a result of a web search.

"About 30% of the 1.4 billion unique documents on the web are rubbish or computer generated and stuff that people are not interested in," said Joe Frost, marketing director at Inktomi.

The study contradicts work carried out by Alberto-Laszlo Barabasi from the University of Notre Dame that suggested it took a maximum of 19 clicks to get from one side of the web to the other. This latest report found that it can take hundreds of clicks to reach some web pages and others cannot be reached at all.

The results of the study are being presented at the 9th International World Wide Web Conference that is being held in Amsterdam this week.

Search BBC News Online

Advanced search options
Launch console
See also:

07 Jul 99 | Sci/Tech
Web engines could do better
Internet links:

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

Links to more Sci/Tech stories are at the foot of the page.

E-mail this story to a friend

Links to more Sci/Tech stories