Was it possible for technology companies like Facebook, Twitter and Google to realize that foreign agents were buying ad space for the purpose of spreading misinformation?
Yes, they have standards and policies in place, and have proven they are capable of identifying these types of activities. However, having ad standards and policies in place needs to become more than a gesture; they need to be investing in the appropriate resources to monitor and enforce their own policies. As technology leaders, they need to be applying their massive data engineering and data analysis talent pool to detect and prevent the spread of misinformation, or bring on the kinds of partners who can specialize in doing that on their behalf.
How does data analysis fit into this narrative?
There are a number of ways to identify concerted efforts by foreign nationals to influence the political landscape within the US, ranging from manual (human) reviews to advanced data analysis methods like machine learning and artificial intelligence. The most effective approach is a combination of all available techniques. Proper analysis of the content, the social networks, and even the metadata and attributes of ad buys can reveal patterns that indicate misinformation campaigns and other types of efforts to thwart policy and compliance.
What should Facebook/Twitter/Google’s data teams have been looking for in order to recognize what was taking place?
There are a few specific signals that the data teams might have been looking at to recognize that misinformation campaigns orchestrated by foreign nationals were taking place. For instance, there is a wealth of data associated with an ad buy, including a few data points that either directly or indirectly associated with location that can help identify patterns related to foreign interests. These might include things like the name and location of the purchaser, the IP addresses from which the purchase was made, and even language targets for the ads.
The next concern might be identifying patterns in concerted efforts among groups of individuals. As industry leaders in social networks, the obvious answer is social network analysis, which can help identify associated groups and links between those groups. Buying patterns such as dates and times can also help identify possible associated groups of buyers.
Lastly, beyond the metadata associated with ad buys, the content of the campaign provides enormous value in identifying non-compliance with ad campaign policies. The ads don’t have to mention specific parties or individuals — there is an enormous amount of data that can be used to train models to classify the political natures and party alliances. These types of ads are clear red flags, but combined with location analysis, social network analysis, and content analysis, patterns can be immediately revealed to uncover attempts by foreign powers to stoke political tensions.
What lessons should other corporations take from the challenges these tech giants are currently facing?
Corporations should learn from these incidents, and understand that the dangers associated with fraudulent or nefarious activities could affect more than their reputation or their customer satisfaction. The stakes are so high today, and data security breaches and fraudulent activities will continue to plague the world’s corporations and economies. Neither policies nor compliance will have an effect unless smart data analysis initiatives are put into place to counteract these types of problems.
In the future, how can companies better foresee troubling patterns like this?
To better detect these emerging patterns, companies must at least be implementing proper data analysis methods internally, or partnering with companies like Signafire that specialize in Big Data with a focus on data fusion and content analytics. With more advanced data analysis technology in place, companies like Facebook, Google and Twitter can mitigate risk while ensuring integrity, and keep their main attention on their business and customers.