Page last updated at 00:59 GMT, Monday, 29 March 2010 01:59 UK

How spam filters dictated Canadian magazine's fate

By Jude Sheerin
BBC News

Generic pic of beaver
Beaver: A semi-aquatic rodent, but also a vulgar slang term

After 90 years, one of Canada's oldest magazines, The Beaver, is changing its name.

Its publishers say it was only natural that a Canadian history journal should have been named in honour of the industrious dam-building creature which is the country's national emblem.

But in recent times the magazine's attempts to reach a new online audience kept falling foul of spam filters - particularly in schools - because beaver is also a slang term for female genitalia.

The publishers of the magazine - now to be known as Canada's History - also noticed that most of the 30,000 or so visitors to their website per month stayed for less than 10 seconds.

Scunthorpe problem

And they suspect that learning about the trade in beaver fur which built Canada's early economic fortunes was not what they were interested in.

In 2004, the Horniman Museum in London failed to receive e-mails as filters thought its name was a version of "horny man"
A history website for children, RomansInSussex, was blocked due to the substring "sex"
Filters have trashed e-mails with the word "socialist"; substring "Cialis" is erectile dysfunction drug
Harmless e-mails promoting the pantomime Dick Whittington have been shot down by filters
E-mails to councils objecting to planning applications rejected for having the word "erection"

Deborah Morrison, publisher of the Winnipeg-based journal, told the BBC News website: "Back in 1920, The Beaver was a perfectly appropriate name.

"And while its other meaning is nothing new, its ambiguity began to pose a whole new challenge with the advance of the internet. The name became an impediment to our growth."

Similar concerns in part prompted Beaver College in the US state of Pennsylvania to change its name in 2001.

The blocking of harmless websites or e-mails by trigger-happy filters is nothing new.

In 1996, residents in the British town of Scunthorpe were initially banned from registering with internet service provider AOL because the town's name contained an obscenity.

This became known as the Scunthorpe problem.

Cyber prudery

Elsewhere in England, residents of the South Yorkshire town of Penistone and Lightwater in Surrey had the same trouble.

The talent of quickly inferring humans' message intent is very hard to automate

Christian Kreibich
International Computer Science Institute

Such ham-fisted cyber prudery might have been forgivable in the internet's infancy.

But why are some filters still apparently unable to distinguish between clean and dirty content?

Are their chastity belts fastened too tight?

The problem, technology experts say, is that with a worldwide web awash with filth, broad blocking rules often result in a false positive, meaning innocent content is binned just to be on the safe side.

Furthermore, distinguishing the e-mail sheep from the goats can be a tricky business, especially given the intensity of the so-called "arms race" between the filters and the spammers.

The spammers develop ever-more sophisticated techniques for slam-dunking our inboxes with ads extolling the benefits of manhood enlargements, pornography and virility pills, among other things.

Clbuttic mistake

Then the spam filter engineers have to hit back by creating smarter deterrents, in a perpetual game of cat and mouse.

American sprinter Tyson Gay
US sprinter Tyson Gay was the subject of a disastrous filtering error

Christian Kreibich, from the International Computer Science Institute in Berkeley, California, says: "The talent of quickly inferring message intent that humans have is very hard to automate.

"You can try to parse the language, but you'll find a lot of it is unfortunately deliberately misleading or broken, for example, with entirely unrelated passages from the latest Dan Brown novel."

The research scientist adds: "The filtering systems are - all in all - working rather well, but of course, individual users don't realise this when a few slip through into their inboxes yet again."

In a farcical sequel to the Scunthorpe problem, some filters have been known to act like rogue spell checkers, replacing "rude" words with what they deem to be more acceptable variants.

"Butt" replaces "ass", "breast" is substituted for "tit", and so on, even within longer words containing the banned letter combinations.

This is known as the "clbuttic mistake", because classic is edited as clbuttic, passport becomes pbuttport, and so on.

Spam tsunami

In 2008, a news website run by the American Family Association censored an Associated Press article on the sprinter Tyson Gay.

Unsolicited e-mails on a Hong Kong computer screen
Up to 400bn spam e-mails a day would deluge us without filters, says Cisco

A filter decided that "gay" was an offensive word, which should be replaced with "homosexual".

The resulting article began with the memorable headline: "Homosexual eases into 100m final at Olympic trials".

Some believe correcting the glitches that see countless legitimate e-mails automatically jettisoned into the junk is not in the financial interests of spam filterers.

Bennett Haselton, webmaster for Peace Fire, an organisation supporting free speech for young people on the internet, said: "The main problem is not that filters are dumb.

"Lots of technology starts out dumb and improves over time.

"The problem here is that there are no marketplace incentives for the filters to improve, because end-users often don't even realise what e-mails they've missed."

Not so, according to Henry Stern, Cisco Systems' senior security researcher, who says spam filters are rated by the industry on their ability not to block legitimate e-mails or websites.

He says without the blockers, a tsunami of between 250bn and 400bn spam messages a day would deluge our inboxes, crashing computer networks around the world.

The spam filter engineer reckons Scunthorpe-related gremlins are largely a thing of the past.

However, at least one e-mail, mentioning the Scunthorpe problem and sent to a university for this article, was zapped by a filter.

And if you try forwarding this piece to a friend at their workplace, don't be too surprised if it doesn't reach its intended recipient.

Print Sponsor

How I helped nail the 'manhood' spammer
02 Dec 09 |  Magazine
How to stop becoming phish food
09 Oct 09 |  Technology
The internet's conscientious objectors
06 Aug 09 |  Magazine
Coping with the malware deluge
20 May 09 |  Technology

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

Has China's housing bubble burst?
How the world's oldest clove tree defied an empire
Why Royal Ballet principal Sergei Polunin quit


Americas Africa Europe Middle East South Asia Asia Pacific