The world’s largest search engine is setting itself up to become the global arbiter of truth online.
Researchers at Google are developing a new system to rank search results based on the number of facts a website contains, rather than the number of incoming links. Under Google’s current system, the higher the number of incoming links to a website (popularity), the higher up it appears in search results, regardless of its content outside of key words included in the search.
Under the new system described by Google researchers in a recently published paper, those results will be based on a truth score given to each website called “Knowledge-Based Trust,” which ranks a website lower in search results based on the number of incorrect facts counted on the site.
Such facts are determined by comparing a website’s information against Google’s “Knowledge Vault,” which Google describes as “a database of 2.8 [billion] facts extracted from the web.” If websites include information contradictory to the Knowledge Vault, they rank lower on search results based on their number of incorrect facts. Sites with fewer contradictions rank higher.
The Knowledge Vault is an automated database that identifies “facts the web unanimously agrees on,” according to New Scientist, which reported the paper first.
“Informally, we define the trustworthiness or accuracy of a web source as the probability that it contains the correct value for a fact (such as Barack Obama’s nationality), assuming that it mentions any value for that fact,” Google researchers wrote.
Google’s software pulls “knowledge triplets” from websites to determine their factual accuracy. In the test case of President Obama’s nationality, Google extracted the words “Barack Obama,” “nationality” and “USA” to determine accuracy.
By pulling 2.8 billion knowledge triplets from the web, Google was able to “reliably predict the trustworthiness of 119 million webpages and 5.6 million websites,” according to Google researchers.
In a test comparison between Google’s traditional ranking system and Knowledge-Based Trust, 15 websites considered online gossip hubs including TMZ, Gawker, and E! Online ranked in the bottom 50 percent of searches, while under the current system, they ranked in the top 15 percent.
The new system is not live and no timeline has been given for when it may replace the current system. Any moves forward on Google’s part to implement Knowledge-Based Trust are sure to raise concerns among users of the world’s most popular search engine, which currently receives more online advertising revenue than any other online company.