Fact Checking Google’s Fact Checker


Yesterday, we mentioned a widely reported potential change to Google’s algorithm. Today, after further investigation, we believe most of the reporting to be over-hyped. This is what we found.

Over the past few days, a number of websites have reported that Google will begin using the “facts” on a website to replace the links that are currently an important ranking factor. Though there are many articles, all of them use this New Scientist report as the source.

The New Scientist article does not say that such a change is coming, though the headline is a bit misleading: “Google wants to rank websites on facts not links.” What the article actually says is that some Google researchers have published a paper on using the facts on a website as a ranking factor. The facts would be compared to those found in Google’s Knowledge Vault.

One confusing, and oft-quoted sentence is “Instead of counting incoming links, the system – which is not yet live – counts the number of incorrect facts within a page.” Many readers took this to mean that the system would replace the page rank algorithm and provide public-facing search results. Most likely it just means that the software, created by the researchers, has not yet been deployed. Google has many alternate algorithmic tests running all the time, and the results are constantly compared to the main search engine. When the system is live, if will just become one of these alternate algorithms, at least for now.

But what it if did become a significant part of Google’s ranking process? Would it really wipe out all the bad pseudo-science on the web? Would conspiracy theorists lose all the attention they crave? Would Google become the final arbiter of truth? Will the iron fist of Google crush dissent and innovation?

First, Google currently uses over 200 ranking factors. The idea that replacing page-rank with this new knowledge-trust test would change everything is ridiculous. At most, it would be one more additional factor.

Second, the source of these facts is not a politically motivated committee somewhere, it is the so-called Knowledge Vault, which uses a complex algorithm to rank the trustworthiness of “facts.” The algorithm looks at several things, consensus and consistency among them. It isn’t easily manipulated. This Knowledge Vault itself was a bit overhyped. The original story came from the same New Scientist reporter, and Google felt the need to correct some of the stories based on this source. The Knowledge Vault is also an experimental project at the moment.

Third, as was reported at the very bottom of one of these over-excited articles:

A spokesman for Google said there were no specific plans to use the algorithm in their public search engine yet.

He said: ‘This was research – we don’t have any specific plans to implement it in our products. We publish hundreds of research papers every year.’

Ironically, all of the alarmist writers have actually missed the one actual threat that their websites face. Google is interested in providing direct answers to searchers when they ask a simple question. If I want to know a simple fact like “Who is the President of the United States?” Google would rather answer my question than send me to some website that happens to include those words. The project called the Knowledge Graph (which is actually in operation) already provides some of the answers. If the Knowledge Vault database becomes operational, it could provide many more. It won’t be a question of how well or poorly a website is ranked so much as whether the user has any reason to leave the Google search results page for questions like this.