Newsspam: The Yahoo and Opera stories

Early in December a Yahoo story spread quickly across the web, MAN DATES GAL ON INTERNET FOR SIX MONTHS — AND IT TURNS OUT SHE’S HIS MOTHER!. Yesterday, almost three weeks later, the story was published unchecked by Dagbladet, Norway’s third largest newspaper where it became the most widely spread and read story that day. The only problem was that this story wasn’t true and it wasn’t intended to be true either.

Writing infective stories is largely to follow a formula. One such story with a seemingly reliable source is all that is needed for it to pass. The source of the Yahoo page was Weekly World News, but the apparent origin was Yahoo! Entertainment. Until the WWN Yahoo stories are better marked it will lower the trust in Yahoo News! (or for that matter Dagbladet) as a reliable newssource.

A few days later a blog came up with the story that Google could buy Opera. That story was just about believable as a rumour. The blogger had connections, and while you could wonder why the anonymous source would spill the beans, but a titillating rumour all the same. The story made the rounds and within a couple days had mutated into that Google actually had bought Opera (but to my great disappointment nobody told us what our price was). That the original was in French added mystique and frisson to the story.

It was hard to believe that people would actually fall for the next story, that Microsoft had bought Opera (in past tense). This story was made by some obscure web site with no connection postulating a business plan that would make absolutely no sense for either Microsoft or Opera, no collaborating evidence, and with story holes that would require massive suspension of disbelief. But it was written according to the formula, the timing was good, and the story involved everyone: Opera users seeing Opera Software being sold out to its worst enemy, Firefox users in fear of a pincer maneuver, IE users for the hint of Microsoft dropping IE as their browser, web users in general as another corner of the web taken over by the Microsoft machine, and mobile web users in particular for Microsoft getting another foothold on their beaches.

Rumours are part of the territory for public companies. The share price generally increase for the company rumoured to be taken over (since purchasing a company comes at a premium) and usually decrease for the company doing the buy-out (since they have to pay a premium, and for all the synergy claims most merged companies do worse than they would have on their own). John Dowdell (Adobe Systems) observed that the recurrent “microsoft to buy macromedia” dream always happened in December. It is yet to be seen what delusions will come next December. Technorati perhaps? [Disclosure: John Dowdell is the recent owner of Opera Software ASA]

But for cui bono there are more motives for creating rumours than stock price manipulation. Making rumours may have its own financial incentives. Stock manipulation is usually intangible and anonymous (it is illegal so you don’t want it tracked back to you), but media has always known that juicy rumours increase circulation and ad income balanced by loss of trustworthiness and lawsuits if the rumours are wrong. On the web a rumour would increase not just the number of hits to the publication, but also the page rank as so many high-rated pages will now link to it.

It could be possible to abuse the system to create newsspam. Reputable news sites can’t create fictitious news without loss of credibility, but no-brand sites can. The referral pattern would be similar to cross-referral search engine scams, but would allow incoming links from external sources which is getting harder when counter-measures like rel=nofollow is getting common. Creating a successful rumour isn’t that easy, most will be duds, but with the critical faculty spam filter missing in the blogosphere this strategy might do well on average.


Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.