On November 2nd, 2010 at 5:26pm EDT, Politico's Tony Romm posted a rather scathing article. The article focuses on the claim that Google's Election Center, which is a service intended to match a voter's address with the location of their ballot box, had potentially misdirected "more than 700,000 households" in twelve battleground states.

Worth noting is that the source for this magical 700,000 number is not an objective source by any means, but rather a competitor, Aristotle, which runs a similar polling place location service.

While this story has been reported and re-reported all across the internet, as near as I can tell it began with this FastCompany.com article. While the Politico article does not make any mention of FastCompany, it is the only article I could find posted about the issue before Aristotle piped up and Politico got a hold of it. Later in the evening, FastCompany posted an update article after they were contacted by Aristotle.

If there is one thing that this story truly demands is that we take a step back from all the finger-pointing and consider the sources and statistics. For the purposes of this examination, I'm going to draw on the content of the following articles (I've culled the plethora of articles I've found that are merely single-source reporting on Politico's article)

1. Why Did Google Get My Polling Place Wrong - FastCompany.com
2. Google polling app misleads voters - Politico.com
3. Google Tool May Have Had Wrong Polling Place Locations for Hundreds of Thousands of Voters - FastCompany.com
4. Google Sent Users to Wrong Polls? Nope, Board of Elections Did - ReadWriteWeb.com

Note: I've arranged the links above into roughly chronological order by post date.

The first article on the list cites a primary source, namely FastCompany.com's own Deputy Editor Tyler Gray who discovered a discrepancy in the location indicated by Google's tool. The tenor of the article is in no way hostile towards Google and even goes so far as to point out that "Google itself is probably not at fault here. Ultimately it's just a clearinghouse for information provided by others" and also notes that "a sidebar on the Google tool urges users to 'check your voting location on your state voting place locator'". The overall message of the piece is summed up in an editor's note in the first paragraph "[Ed: Two sources. Always]", in other words, don't just rely on a single source and assume it is right.

The second article, coming from Politico, appears to go against that most basic of journalistic principles. The sole source indicated in the article for finding the issue with Google's tools is the aforementioned competitor, Aristotle. While Mr. Romm does disclose the conflict of interest, John Phillips, the CEO of Aristotle, appears to be his only source for discovering Google's apparent flaws.

Where things begin to look fishy is in the third article, in which FastCompany.com posts a follow-up to their original one. "Following our story, Aristotle, which creates its own polling place locator tool, contacted Fast Company to tell us that, in the previous two weeks, they had been testing the data Google was using to determine its accuracy." Okay. Clearly this update post was not driven by further investigation, but rather was prompted by a call from the same competitor.

For the first time, however, we are given a glimpse not only into the process, but we also get a chart of the actual data collected by Aristotle. Now I'm not sure at this point whether it is typos, poor reporting on the part of Fast Company's E.B. Boyd, or intentional misinformation on the part of Aristotle, but there are a few details that don't match up, ranging from a simple mathematics error (1 error out of a sample of 999 addresses is as 0.1% error, not 0.001%) to something much more sinister.

The problem cannot be detected in this third article alone. Fast Company reports that, over the course of the previous two weeks, Aristotle "pulled 1,000 random addresses belonging to registered voters" and used Google's polling place lookup tool. Aristotle then took that set of addresses and entered them "into tools on the websites belonging to official election agencies". Their analysis was then based on a comparison of those two sets of results.

It's not until you compare that description of Aristotle's analysis of Google's tools with the one given by Brandi Travis, who is the manager of grassroots services at Aristotle. As quoted in the fourth article, "Aristotle took 1,000 random records from each state and ran those records against whereivote.com".

Did you catch the distinction?

If I add this quote from the Politico article, does it help? "... said John Phillips, CEO of Aristotle, which maintains its own ballot box location service called whereivote.com"

When Aristotle contacted Fast Company, at that time having published an article that is perhaps the most sympathetic to Google's position, Aristotle claimed that their research was based on a comparison with data from the "official election agencies", yet Aristotle's Brandi Travis is quoted as saying that their research was based on a comparison with Aristotle's own database.

Let's think about what that means. Aristotle, over the past two weeks, has basically been compiling a list of discrepancies between Google's service and their own, assuming that whenever there is a discrepancy, their database is correct and Google's is wrong. Based on those findings, they have gone around beating the drum on Election Day, passing their figures out to all the news media and saying "Bad Google".

My question is, how can we be sure that in every single discrepancy, Aristotle's database was correct? They claim to use "real-time official information, as provided by the State and County elections' offices", which is pretty much exactly where Google is claiming to pull its data.

So what we ultimately have here is a case where two companies are providing the same product, which receives data from the same sources. How can we trust one company to be objective and impartial in judging the other, under those circumstances? That would be like HTC reviewing an iPhone, or Chrysler calling up Consumer Reports to offer its findings on the fuel efficiency of the Toyota Prius.

In this age of rampant hyperbole and reckless sensationalist journalism, I think there is one quote worth holding on to. Repeat this to yourself whenever you read anything (no matter how legitimate the source may appear to be):

"[Ed: Two sources. Always]"