Hey, where's my truth tester, voter dude?

You didn't? Well, clearly you don't read England's Guardian newspaper. Here's what a bit of what columnist Greg Palast (an American, incidentally), had to say

As in Florida in 2000, and as in Ohio in 2004, the exit polls show the voters voted for the progressive candidate. The race is "officially" too close to call. But they will call it - after they steal it.

Reuters reports that, as of 8pm eastern time, as voting concluded in Mexico, exit polls showed Andrés Manuel López Obrador of the "leftwing" party of the Democratic Revolution (PRD) leading in exit polls over Felipe Calderón of the ruling conservative National Action party (PAN).

We've said again and again: exit polls tell us how voters say they voted, but the voters can't tell pollsters whether their vote will be counted. In Mexico, counting the vote is an art, not a science - and Calderón's ruling crew is very artful indeed. The PAN-controlled official electoral commission, not surprisingly, has announced that the presidential tally is too close to call.

Calderón's election is openly supported by the Bush administration.

On the ground in Mexico city, our news team reports accusations from inside the Obrador campaign that operatives of the PAN had access to voter files that are supposed to be the sole property of the nation's electoral commission. We are not surprised.

This past Friday, we reported that the US Federal Bureau of Investigation had obtained Mexico's voter files under a secret "counter-terrorism" contract with the database company ChoicePoint of Alpharetta, Georgia.

In fact, if you copy the first sentence of his rant above into the Google search field (but without quotation marks) and hit return you'll get 80,200 hits -and the first couple of pages of listings all seem to echo both his paranoia and his complaint.

So if we set aside personal political opinion and our natural response at being asked to deal with an obvious nutter, how do we know first that this guy is wrong and secondly that the people who either quote him approvingly or offer similar views have gone little off the deep end?

What's going on with these people is a special case of group think - some unhappiness and reasonable doubt in a few cases being amplified by the instant feedback and credibility created by Internet blogging and psuedo-news websites.

There's an important question here: how can we protect ourselves and others from the social and economic consequences of getting swept up by what amounts to an emotional disinformation mob? And "mob" is the right analogy, they're not limited to rampaging rioters burning cars and buildings -and if you don't believe that try publishing something that points out that SCO has case, that there's no evidence for human culpability in global climate change, or that data processing predates computing by forty years and has nothing to do with business applications.

All of which brings us to the real bottom line: how do we know who or whether to trust when a Google search produces a large number of hits that slant predominantly one way on some issue? It's not like Reuters is going to publish a report under the headline "George Bush Didn't Steal Mexican Election."

So how can we tell truth from falsehood in an Internet document? More directly how can a search engine like Google help us do this without appointing itself or anyone else as the unbiased arbiter of Truth and Value?

I don't know - but I do have a suggested direction for research into the issue.

Google currently uses citations (links are a weak proxy for this) as part of its page ranking scheme and we don't know much about what the author of some page found in a Google search is all about - but Google does because it has access to whatever else he has offered on the Internet. Thus Google could alter its page ranking and display mechanisms to report both forward and backward citations: who cites the author, and who does the author cite - across all of the author's work Google has access to and across all of the documents on the site the particular hit refers to. Classify those citations in terms of peer review, public journal, educated newspaper, or tabloid and you have both the beginnings of a reader's guide to the likely sanity of the writer's position and an easily user customizable mechanism for search result rankings.