David Amerland

How Google Learns to ‘See’ the Truth

Google uses human agents to verify breaking news

There are four Vs to semantic search which also apply to Big Data: Volume, Velocity, Variety and Veracity. The last one is also the most problematic. You can use algorithms to deal with data that’s going viral, is topical and has acquired huge importance because of a news angle but how do you verify the sources? How do you make sure that what your algorithms are indexing is coming from a trustworthy source? How, in other words, do you ensure that the systems you have in place to quickly propagate news also do not spread misinformation? 

These are the issues Google has to content with when it comes to breaking news and news sources which is why the company has set up a novel initiative to help tackle it. Eyewitness Video on YouTube announced a three part initiative that aims to introduce the human factor into the equation and provide a first-basis verification process that can then be used by Google’s algorithms to produce smarter, more accurate results.   

The spread of smartphones and the emergence of the devices as a primary means of connecting to the web for many developing country residents makes video the easiest means of breaking any kind of news story. Verifying the accuracy of that, often raw, video footage is critical. Google’s initiative introduces three very important layers to the equation: 

  • A curated feed on YouTube that allows humans to establish trustworthiness in raw video footage that might be newsworthy (that’s YouTube Newswire)
  • The training of first reporters in verification techniques and ethics, establishing a broad coalition of partners who can be trusted to vet and propagate raw content across the web and, more importantly, provide an initial layer of context by incorporating it in their reporting (that's being called "the First Draft Coalition".
  • The establishment of a focal point for collecting and vetting crowdsourced stories and video footage, through the Witness Lab site.

These are broad initiatives that take into account the evolution of the web: 

  • Video as a primary communication tool
  • Crowdsourced information uploaded directly from smartphones and mobile devices to sites like YouTube and Twitter
  • The increased incorporation and featuring of citizen journalism in professional news reports. 

Video, News Reports and Factual Journalism

The importance of the initiative cannot be understated. By getting involved its own YouTube platform as well as a broad coalition of journalists and news sources google establishes a primary net that uses the equivalent of human rates to filter trustworthy news. News is data. What happens to it in terms of how far it travels, how it morphs, where it is cited, how it is incorporated on other news reports and who is involved along the chain of all these processes can be tracked and assessed, on the fly, with an algorithm. 

Provided the initial kernel is verified everything else in semantic search becomes a signal that shows, importance, trustworthiness, context and relevance. What’s more the tracking of the activity around specific data on the web allows semantic search to identify “bad actors” who promote misinformation or are uncritical about their sources and move to reduce their impact by lowering their importance and raising question marks about their trustworthiness as sources. 

Google is doing a very smart thing here by becoming involved at the primary level of news reporting, the critical stage when news breaks from unexpected sources as these are of peak search interest. It is adding, therefore, a pretty strong string to its figurative brand bow. At the same time it is forging partnerships and helping shape the news gathering industry, at that level, to better help it sift through fabrications at the delicate initial stage where the available data is not sufficient for an algorithm to assess the veracity of something new. 

As an aside, this also shows the weakness of the current stage of development of machine learning. Unless there is a sufficient volume of data for algorithms to work with and begin to organize it into patterns that allow inferences such as “true” or “false” to be made, a human is still better at understanding the context of a piece of information that is entirely new.  

How Does This Affect Your Marketing?

The web is becoming a more truth-orientated place. Just like in the real world, trust takes a long time to develop and is easily lost. If you’re incorporating news sources, video, Tweets or other primary sources of data in your content or sharing activity, being able to verify their provenance is a necessary skill. 

Before you share content from sources you’re not familiar with, you will also have to do your own due diligence to make sure that they’re current and can be trusted, otherwise their mistakes will affect your reputation. 

Your social sharing activity is every bit as important to your marketing as your content marketing because it now impacts on your trustworthiness. You may, for instance, create some great content for your Real Estate business but if you also share a whole lot of conspiracy theory material that suggests the world is ruled by a group of shape-shifting lizards from another planet, at some point your reputation will start to suffer in the eyes of Google’s algorithms and, just like in the real world, your activity in one area will begin to impact upon your work, in another. 


© 2019 David Amerland. All rights reserved