I think there's an incredible opportunity for mis/disinformation researchers to take the next step and embrace network science in their research

Info pollution spreads on networks. Info ops target networked audiences. Divisive content fragments entire networks

1/

All of these social media threats are fundamentally networked processes

It's easy to grab keyword tweets from the Twitter API and visualize them in Gephi. And the ease of that has allowed journalists and researchers to do some amazingly impactful investigative work

2/

But to comprehensively understand how misinfo connects across events and over time, how info ops infiltrate audiences, and how to scale these insights to the level of entire platforms, we *need* to move beyond simple network viz and draw more fully on network science

3/

Networks are incredibly complex objects. For any modest misinfo story spreading for more than a few hours, the resulting network will simply be too large for you to visualize and reliably filter without destroying fundamental information

4/

(Let alone larger networks of stories, communities, and users over long periods of times which may have *at minimum* hundreds of millions of edges)

5/

Network science methods help us in at least two critical ways

First, they let us identify important structural patterns when we can't visual a network at all. We need network science to help us pull out patterns when there are simply too many nodes and edges for a single viz

6/

Second, even in the case where we *can* visualize a network, we still need network methods to identify patterns. Network visualizations can be unintuitive and misleading. Our minds are not good with comprehending multi-step relational data. And they can miss critical insights

7/

Consider a sophisticated information operation that has subtly embedded itself in a community. One that can seed stories for other high profile actors to pick up, but which itself is never in the center spotlight. We know info ops and media manipulators aim to do this

8/

If we only rely on network viz and basic measures like (in-)degree, we may never pick up on this operation. By definition, those viz and measures will not highlight the adversarial actors because they intentionally do *not* get many retweets, and instead seed to others

9/

Network science gives us a host of sophisticated methods for drawing out these subtleties (and for the example above, my current research shows how you could still identify those actors)

Network science as a field is ready to support the work of info pollution researchers

10/

There is so much potential for network science to help us map long term, platform level info pollution, and to help us do deeper investigative work with mixed methods. Network science can help is build more just and equitable info ecosystems

11/11

Thread inspired by @CarlottaDotto and @cward1e who recognize the trickiness of networks for information pollution research https://twitter.com/cward1e/status/1281553458894704640?s=19